Social sharing of emotion following exposure to a negatively valenced situation

28
Social sharing of emotion following exposure to a negatively valenced situation Olivier Luminet Belgian National Fund for Scientific Research and University of Louvain at Louvain-la-Neuve, Belgium Patrick Bouts University of Louvain at Leuven, Belgium Fre ´de ´rique Delie University of Louvain at Louvain-la-Neuve, Belgium Antony S.R. Manstead University of Amsterdam, The Netherlands Bernard Rime ´ University of Louvain at Louvain-la-Neuve, Belgium COGNITION AND EMOTION, 2000, 14 (5), 661–688 Please send correspondenc e and requests for reprints to Olivier Luminet, University of Louvain, Department of Psychology, Research Unit for Clinical and Social Psychology, 10, Place Cardinal Mercier, B-1348 Louvain-la-Neuve , Belgium; e-mail: [email protected]. The design of the research reported here grew out of a research workshop on emotion taught by Bernard Rime ´ and Tony Manstead at the Summer School of the European Association of Experimental Social Psychology, San Sebastian, Spain, in June 1991. Sincere thanks are due to all the other students who participated in the workshop and helped to develop the design for Experiment 1 reported here: Mirella Amatulli, Meltem Avci, Stefano Boca, Javier Cerrato, Elmarie Egan, Tali Freund, Mariagrazia Monaci, Neo Morojele, Renate Scha ¨fer, and Agnes Vayreda. We also wish to thank Craig A. Smith and three anonymous reviewers who provided invaluable comments and suggests on an earlier draft. The research reported in this paper was supported by a grant from the Fonds de De ´veloppemen t Scientifique of the University of Louvain and by grant 8.4510.94, 2.4546.97, and 1.5.1214.00 from the Belgian National Fund for Scientific Research. It was also facilitated by support from the Maison des Sciences de l’Homme, Paris to the Consortium for European Research on Emotion (CERE). Ó 2000 Psychology Press Ltd http://www.tandf.co.uk/journals/pp/02699931.html

Transcript of Social sharing of emotion following exposure to a negatively valenced situation

Social sharing of emotion following exposure to anegatively valenced situation

Olivier LuminetBelgian National Fund for Scientific Research and

University of Louvain at Louvain-la-Neuve Belgium

Patrick BoutsUniversity of Louvain at Leuven Belgium

Frederique DelieUniversity of Louvain at Louvain-la-Neuve Belgium

Antony SR MansteadUniversity of Amsterdam The Netherlands

Bernard RimeUniversity of Louvain at Louvain-la-Neuve Belgium

COGNITION AND EMOTION 2000 14 (5) 661ndash688

Please send correspondence and requests for reprints to Olivier Luminet University of LouvainDepartment of Psychology Research Unit for Clinical and Social Psychology 10 Place CardinalMercier B-1348 Louvain-la-Neuve Belgium e-mail luminetclisuclacbe

The design of the research reported here grew out of a research workshop on emotion taught byBernard Rime and Tony Manstead at the Summer School of the European Association ofExperimental Social Psychology San Sebastian Spain in June 1991 Sincere thanks are due to allthe other students who participated in the workshop and helped to develop the design for Experiment1 reported here Mirella Amatulli Meltem Avci Stefano Boca Javier Cerrato Elmarie Egan TaliFreund Mariagrazia Monaci Neo Morojele Renate Schafer and Agnes Vayreda We also wish tothank Craig A Smith and three anonymous reviewers who provided invaluable comments andsuggests on an earlier draft The research reported in this paper was supported by a grant from theFonds de Developpement Scientifique of the University of Louvain and by grant 84510942454697 and 15121400 from the Belgian National Fund for Scientific Research It was alsofacilitated by support from the Maison des Sciences de lrsquoHomme Paris to the Consortium forEuropean Research on Emotion (CERE)

Oacute 2000 Psychology Press Ltdhttpwwwtandfcoukjournalspp02699931html

Three experimental studies are reported in which we tested the prediction thatnegative emotion elicits the social sharing of the emotional experience In twoexperiments participants arrived at the laboratory with a friend and then viewedone of three film excerpts (nonemotional moderate emotion or intense emotion)alone Afterwards the participants who saw the film had an opportunity to interactwith the friend and their conversation was recorded In both experimentsparticipants who had seen the intense emotion excerpt engaged in significantlymore social sharing than did participants in the other two conditions A thirdexperiment extended the investigation to social sharing in everyday life andreplicated the previous results Nonemotional explanations of the effects observedwere also ruled out Results are discussed in the context of the relation betweenemotional intensity and amount of social sharing

Being exposed to an emotional condition has a marked impact on a personrsquosmotivation to seek social contact In a series of now classic experimentsSchachter (1959) demonstrated that participants who were anxious at theprospect of being administered electric shocks expressed an overwhelmingpreference for waiting in the company of other persons rather than alone Bycontrast when the level of threat was low the majority of participants preferredto wait alone These findings suggested that anticipation of stress increases theneed to be with others and stimulated considerable research interest in thelsquolsquostress and affiliation effectrsquorsquo Nearly forty sets of studies of this phenomenonwere reviewed by Cottrell and Epley (1977) These authors concluded that therewas general support for the hypothesis that fear elicits affiliation

In the period following Cottrell and Epleyrsquos (1977) review scientific interestin the stress-affiliation relation waned Schachter himself moved on to develophis influential two-factor theory of emotion (Schachter 1964 Schachter ampSinger 1962) When the key study upon which this theory rested failed to bereplicated (Marshall amp Zimbardo 1979 Maslach 1979) Schachterrsquos viewsmdashand in particular the notion that emotion sometimes entails an ambiguousinternal statemdashcame to command less influence among researchers HoweverShaver and Klinnert (1982) advocated that it would be a mistake to discard thework on stress and affiliation They argued that this research yielded data thatare consistent with findings from research on attachment initiated by Bowlby(1969 1973) This research provided ample demonstration that both primatesand human infants seek contact with others at times of uncertainty and distress(eg Ainsworth Blehar Waters amp Wall 1978 Sroufe amp Waters 1977) Shaverand Klinnert (1982) concluded that for both humans and non-human primatesthe relationship between distress and affiliation is maintained throughout life

More recently fresh efforts have been made to investigate the tendency toaffiliate under stress (eg Gump amp Kulik 1997 Kirkpatrick amp Shaver 1988Kulik amp Mahler 1987 1989 Kulik Mahler amp Earnest 1994 Kulik Moore ampMahler 1993) Particular attention was paid to overcoming the limitations ofSchachterrsquos (1959) original research procedure which relied solely on

662 LUMINET ET AL

anticipated electric shock Affiliation was investigated both under real-lifethreats with room-mates among hospital patients expecting to undergo majorsurgery (Kulik amp Mahler 1987 1989 Kulik et al 1993) and in laboratorystudies (Gump amp Gulik 1997 Kulik et al 1994) To summarise the stress-affiliation effect is alive Under stressful conditions there is a tendency ofhuman beings to seek the company of others

Being exposed to an emotion-arousing situation not only impacts on anindividualrsquos motivation to social contact Emotion also entails importantconsequences for the personrsquos cognitive activity First clinical and field studiesshowed that after witnessing a traumatic event many people experienceintrusive thoughts or images (eg Bownes OrsquoGorman amp Sayers 1991McCammon Durham Allison amp Williamson 1988 Tait amp Silver 1989Weisaeth 1989 Wilkinson 1983) Second laboratory research demonstratedthat such cognitive consequences are not limited to those who seek psychiatrictreatment after experiencing a highly negative life event Such symptoms aremanifested after even relatively minor or moderate emotion-inducing events (fora review see Horowitz 1975 1992) Third real-life data based on volunteersrsquoreports of recent autobiographic episodes in which they experienced emotionalso showed that not only trauma but emotional situations in general elicitcognitive consequences (Rime Mesquita Philippot amp Boca 1991a RimeNoel amp Philippot 1991b for a review see Rime Philippot Mesquita amp Boca1992) More than 95 of the respondents reported having had recurrent thoughtsabout the emotional episode in the hours or days that followed it A considerableproportion of them mentioned that they have had such thoughts often or veryoften

To summarise one line of evidence reveals that exposure to an emotionalsituation generally entails social consequences A second line of evidence showsthat exposure to an emotional situation has cognitive consequences in the formof intrusive and repetitive thoughts and images related to the emotion-elicitingsituation Considering the two lines of findings jointly gives rise to a newprediction regarding the interpersonal effects of an emotional episode If peopleexposed to an emotional event have an urge to affiliate and if they experiencerecurrent emotion-related thoughts and images the resulting interpersonalsituation can be expected to involve what has been called the lsquolsquosocial sharing ofemotionrsquorsquo (Rime 1989 Rime et al 1991a) The social sharing of emotioninvolves (a) the evocation of the emotion in a socially shared language and (b)at least at the symbolic level an addressee It occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their own feelings and emotional reactionsThere are a number of theoretical and empirical arguments suggesting thatcommunicating about a negative emotional experience may fulfil adaptativefunctions Thus Pennebaker (eg 1989) has developed an extensive line ofresearch on the ways in which talking about traumatic events impacts on

SOCIAL SHARING OF EMOTION 663

physical and psychological adjustment His influential inhibition theory arguesthat talking about highly stressful events significantly decreases psychologicaldistress health complaints and physician visits Considering together intrusivethoughts and talks about a traumatic event Lepore and his colleagues (egLepore 1997 Lepore Silver Wortman amp Wayment 1996) argued that whenboth processes emerge in a supportive environment they facilitate the cognitiveprocessing of the emotional event thereby reducing emotional distress

The reasoning developed in the present introduction leads us to predict thatpeople spontaneously incline in developing this potentially adaptative behaviourwhich consists in talking about their distressing experiences The centralargument is that the stress-affiliation relationship is paralleled by a broader-spectrum phenomenon that occurs in a variety of emotional circumstances andwhich involves talking about the emotion Studies in which volunteers reportedrecent episodes of everyday life emotion show that any emotional episode evenif mild or moderate is generally associated with an urge to talk and to share theemotional experience with others (Rime et al 1991a b for a review see Rimeet al 1992) Participants who were instructed to retrieve from their memory aninstance of an emotion they recently experienced also answered questions suchas Did you talk about the episode with other people With whom How longafter the emotion How often Eight independent studies based on thisprocedure were reviewed by Rime et al (1992) These involved a total of 913participants whose ages ranged from 12 to 60 years and entailed 1384 emotionalepisodes The observed proportion of cases in which the respondent reportedhaving talked about the emotion with others varied from 90 to 96 Talkingabout a negative emotion generally began soon after the eliciting situation andthis sharing was typically repeated A majority of respondents answered thatthey first talked about the event lsquolsquoon the same dayrsquorsquo and that they talked aboutthe episode lsquolsquoseveral times with several personsrsquorsquo Addresses were almostinvariably intimates (ie parents siblings friends or spousepartner) Finallymore disruptive episodes were talked about more extensively than lessdisruptive ones

The results just reviewed are consistent with the present prediction thatpeople who were exposed to an emotional episode spontaneously incline insocially sharing the emotion they experienced However this existing evidenceis correlational in nature A straightforward experimental test of the assumptionthat negative emotion elicits the social sharing of emotion would consist ofexposing participants to negative emotional stimuli varying in intensity and thencomparing the extent of spontaneous talking about the stimuli the reactions tothe stimuli or both Such a test has to our knowledge never been conducted Inthis paper we report three experiments of this type Manipulation of theintensity of emotional experience was achieved by exposing participants todifferent film excerpts For ethical reasons the intensity range was restricted tothe low-to-moderate range that can reasonably be used in a laboratory context

664 LUMINET ET AL

In the first two experiments shortly after being exposed to the movieparticipants were left in a waiting room situation with a close friend who had notseen the same excerpt Close friends were used because previous research hasestablished that they are the most usual type of partner for the social sharing ofemotion among young adults (Rime et al 1991a) The conversations betweenparticipants and friends were unobtrusively tape recorded and later scored withrespect to the extent of social sharing In the third experiment we examinedwhether the social sharing observed under laboratory conditions could bereplicated when assessed in a real-life context In this third experiment we alsoattempted to rule out some plausible nonemotional explanations of the effectsobserved in the first two experiments

EXPERIMENT 1

Pre-testing of emotional stimuli

The hypothesis to be tested concerns the effects on the social sharing of emotionof variations in emotional intensity We therefore needed to vary the intensity ofemotional experience while keeping constant the quality or type of emotioninduced We achieved this by using appropriately selected film excerpts asemotional stimuli The pre-testing procedure was undertaken with nopreconceptions regarding the type of emotion elicited The goal was to identifyat least two excerpts that were comparable in the type of emotion elicited butdiffered in intensity Excerpts from the set of films described by Philippot(1993) and from movies and video-recordings available from the universitymedia centre were shown to volunteer students who rated them for felt emotionsAlthough the pre-tests involved many excerpts only two were found to elicitsimilar emotions of different intensity Both lasted about three minutes anddepicted violence inflicted on animals One came from a television movie oncockfighting and portrayed violence inflicted by animals on animals The otherwas taken from a commercially available film called Faces of Death (Philippot1993) and depicts cruelty inflicted by humans on animals Pre-tests showed thatthe latter was more emotionally arousing than the former

A formal pre-test was then conducted in order to ascertain the appropriatenessof the two selected stimuli Twenty undergraduate psychology students (10 menand 10 women) participated on a voluntary basis They were individually showneach of the two excerpts in an order counterbalanced across participants Aftereach excerpt they rated on 7-point scales ranging from 0 (not at all) to 6 (verymuch) the intensity of their reactions with respect to four basic emotionsmdashsadness anger fear and disgust These emotions were selected as appropriateconsidering the type of films displayed Because these four emotional intensityscales exhibited a high internal consistency a = 82 scores were averaged to forma single index of emotional intensity This index was found to be higher for theFaces of Death excerpt (M = 308 SD = 083) than for the cockfighting one (M =

SOCIAL SHARING OF EMOTION 665

204 SD = 098) t(18) = 397 p lt 001 Figure 1 shows that the profiles ofemotions elicited by each of the two movie excerpts were very similar Consistentwith this a MANOVA with movie condition as a between-participants factor andthe four basic emotions as a within-participant factor resulted in a nonsignificantinteraction F(3 54) lt 1 It was thus confirmed that although the Faces of Deathexcerpt elicited more intense emotional reactions than did the cockfightingexcerpt the two excerpts were nevertheless comparable with respect to the qualityof induced emotion A nonemotional control movie was also needed to test thecentral hypothesis A 3-minute excerpt of a film called Wildlife in the Himalayasthat depicted animal life without any violence was chosen for this purpose

Method

Participants

Thirty undergraduate psychology students at the University of LouvainLouvain-la-Neuve Belgium volunteered to participate in a study on lsquolsquocoopera-tionrsquorsquo Each of them agreed to arrive at the laboratory at a specified timetogether with a friend Due to technical problems (ie inaudible tape-recording)7 dyads had to be eliminated The excluded dyads were evenly spread acrossconditions (3 in the nonemotional condition 2 in the moderate emotioncondition and 2 in the intense emotion condition) Thus they were 46participants 17 men and 27 women (gender failed to be recorded for one pair ofparticipants) in the final sample Dyads were equally distributed amongconditions (n = 7 for the nonemotional condition n = 8 for the moderate emotioncondition and n = 8 for the intense emotion condition)

Figure 1 Profiles of emotions as a function of the intensity of the stimulus (pre-test)

666 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Three experimental studies are reported in which we tested the prediction thatnegative emotion elicits the social sharing of the emotional experience In twoexperiments participants arrived at the laboratory with a friend and then viewedone of three film excerpts (nonemotional moderate emotion or intense emotion)alone Afterwards the participants who saw the film had an opportunity to interactwith the friend and their conversation was recorded In both experimentsparticipants who had seen the intense emotion excerpt engaged in significantlymore social sharing than did participants in the other two conditions A thirdexperiment extended the investigation to social sharing in everyday life andreplicated the previous results Nonemotional explanations of the effects observedwere also ruled out Results are discussed in the context of the relation betweenemotional intensity and amount of social sharing

Being exposed to an emotional condition has a marked impact on a personrsquosmotivation to seek social contact In a series of now classic experimentsSchachter (1959) demonstrated that participants who were anxious at theprospect of being administered electric shocks expressed an overwhelmingpreference for waiting in the company of other persons rather than alone Bycontrast when the level of threat was low the majority of participants preferredto wait alone These findings suggested that anticipation of stress increases theneed to be with others and stimulated considerable research interest in thelsquolsquostress and affiliation effectrsquorsquo Nearly forty sets of studies of this phenomenonwere reviewed by Cottrell and Epley (1977) These authors concluded that therewas general support for the hypothesis that fear elicits affiliation

In the period following Cottrell and Epleyrsquos (1977) review scientific interestin the stress-affiliation relation waned Schachter himself moved on to develophis influential two-factor theory of emotion (Schachter 1964 Schachter ampSinger 1962) When the key study upon which this theory rested failed to bereplicated (Marshall amp Zimbardo 1979 Maslach 1979) Schachterrsquos viewsmdashand in particular the notion that emotion sometimes entails an ambiguousinternal statemdashcame to command less influence among researchers HoweverShaver and Klinnert (1982) advocated that it would be a mistake to discard thework on stress and affiliation They argued that this research yielded data thatare consistent with findings from research on attachment initiated by Bowlby(1969 1973) This research provided ample demonstration that both primatesand human infants seek contact with others at times of uncertainty and distress(eg Ainsworth Blehar Waters amp Wall 1978 Sroufe amp Waters 1977) Shaverand Klinnert (1982) concluded that for both humans and non-human primatesthe relationship between distress and affiliation is maintained throughout life

More recently fresh efforts have been made to investigate the tendency toaffiliate under stress (eg Gump amp Kulik 1997 Kirkpatrick amp Shaver 1988Kulik amp Mahler 1987 1989 Kulik Mahler amp Earnest 1994 Kulik Moore ampMahler 1993) Particular attention was paid to overcoming the limitations ofSchachterrsquos (1959) original research procedure which relied solely on

662 LUMINET ET AL

anticipated electric shock Affiliation was investigated both under real-lifethreats with room-mates among hospital patients expecting to undergo majorsurgery (Kulik amp Mahler 1987 1989 Kulik et al 1993) and in laboratorystudies (Gump amp Gulik 1997 Kulik et al 1994) To summarise the stress-affiliation effect is alive Under stressful conditions there is a tendency ofhuman beings to seek the company of others

Being exposed to an emotion-arousing situation not only impacts on anindividualrsquos motivation to social contact Emotion also entails importantconsequences for the personrsquos cognitive activity First clinical and field studiesshowed that after witnessing a traumatic event many people experienceintrusive thoughts or images (eg Bownes OrsquoGorman amp Sayers 1991McCammon Durham Allison amp Williamson 1988 Tait amp Silver 1989Weisaeth 1989 Wilkinson 1983) Second laboratory research demonstratedthat such cognitive consequences are not limited to those who seek psychiatrictreatment after experiencing a highly negative life event Such symptoms aremanifested after even relatively minor or moderate emotion-inducing events (fora review see Horowitz 1975 1992) Third real-life data based on volunteersrsquoreports of recent autobiographic episodes in which they experienced emotionalso showed that not only trauma but emotional situations in general elicitcognitive consequences (Rime Mesquita Philippot amp Boca 1991a RimeNoel amp Philippot 1991b for a review see Rime Philippot Mesquita amp Boca1992) More than 95 of the respondents reported having had recurrent thoughtsabout the emotional episode in the hours or days that followed it A considerableproportion of them mentioned that they have had such thoughts often or veryoften

To summarise one line of evidence reveals that exposure to an emotionalsituation generally entails social consequences A second line of evidence showsthat exposure to an emotional situation has cognitive consequences in the formof intrusive and repetitive thoughts and images related to the emotion-elicitingsituation Considering the two lines of findings jointly gives rise to a newprediction regarding the interpersonal effects of an emotional episode If peopleexposed to an emotional event have an urge to affiliate and if they experiencerecurrent emotion-related thoughts and images the resulting interpersonalsituation can be expected to involve what has been called the lsquolsquosocial sharing ofemotionrsquorsquo (Rime 1989 Rime et al 1991a) The social sharing of emotioninvolves (a) the evocation of the emotion in a socially shared language and (b)at least at the symbolic level an addressee It occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their own feelings and emotional reactionsThere are a number of theoretical and empirical arguments suggesting thatcommunicating about a negative emotional experience may fulfil adaptativefunctions Thus Pennebaker (eg 1989) has developed an extensive line ofresearch on the ways in which talking about traumatic events impacts on

SOCIAL SHARING OF EMOTION 663

physical and psychological adjustment His influential inhibition theory arguesthat talking about highly stressful events significantly decreases psychologicaldistress health complaints and physician visits Considering together intrusivethoughts and talks about a traumatic event Lepore and his colleagues (egLepore 1997 Lepore Silver Wortman amp Wayment 1996) argued that whenboth processes emerge in a supportive environment they facilitate the cognitiveprocessing of the emotional event thereby reducing emotional distress

The reasoning developed in the present introduction leads us to predict thatpeople spontaneously incline in developing this potentially adaptative behaviourwhich consists in talking about their distressing experiences The centralargument is that the stress-affiliation relationship is paralleled by a broader-spectrum phenomenon that occurs in a variety of emotional circumstances andwhich involves talking about the emotion Studies in which volunteers reportedrecent episodes of everyday life emotion show that any emotional episode evenif mild or moderate is generally associated with an urge to talk and to share theemotional experience with others (Rime et al 1991a b for a review see Rimeet al 1992) Participants who were instructed to retrieve from their memory aninstance of an emotion they recently experienced also answered questions suchas Did you talk about the episode with other people With whom How longafter the emotion How often Eight independent studies based on thisprocedure were reviewed by Rime et al (1992) These involved a total of 913participants whose ages ranged from 12 to 60 years and entailed 1384 emotionalepisodes The observed proportion of cases in which the respondent reportedhaving talked about the emotion with others varied from 90 to 96 Talkingabout a negative emotion generally began soon after the eliciting situation andthis sharing was typically repeated A majority of respondents answered thatthey first talked about the event lsquolsquoon the same dayrsquorsquo and that they talked aboutthe episode lsquolsquoseveral times with several personsrsquorsquo Addresses were almostinvariably intimates (ie parents siblings friends or spousepartner) Finallymore disruptive episodes were talked about more extensively than lessdisruptive ones

The results just reviewed are consistent with the present prediction thatpeople who were exposed to an emotional episode spontaneously incline insocially sharing the emotion they experienced However this existing evidenceis correlational in nature A straightforward experimental test of the assumptionthat negative emotion elicits the social sharing of emotion would consist ofexposing participants to negative emotional stimuli varying in intensity and thencomparing the extent of spontaneous talking about the stimuli the reactions tothe stimuli or both Such a test has to our knowledge never been conducted Inthis paper we report three experiments of this type Manipulation of theintensity of emotional experience was achieved by exposing participants todifferent film excerpts For ethical reasons the intensity range was restricted tothe low-to-moderate range that can reasonably be used in a laboratory context

664 LUMINET ET AL

In the first two experiments shortly after being exposed to the movieparticipants were left in a waiting room situation with a close friend who had notseen the same excerpt Close friends were used because previous research hasestablished that they are the most usual type of partner for the social sharing ofemotion among young adults (Rime et al 1991a) The conversations betweenparticipants and friends were unobtrusively tape recorded and later scored withrespect to the extent of social sharing In the third experiment we examinedwhether the social sharing observed under laboratory conditions could bereplicated when assessed in a real-life context In this third experiment we alsoattempted to rule out some plausible nonemotional explanations of the effectsobserved in the first two experiments

EXPERIMENT 1

Pre-testing of emotional stimuli

The hypothesis to be tested concerns the effects on the social sharing of emotionof variations in emotional intensity We therefore needed to vary the intensity ofemotional experience while keeping constant the quality or type of emotioninduced We achieved this by using appropriately selected film excerpts asemotional stimuli The pre-testing procedure was undertaken with nopreconceptions regarding the type of emotion elicited The goal was to identifyat least two excerpts that were comparable in the type of emotion elicited butdiffered in intensity Excerpts from the set of films described by Philippot(1993) and from movies and video-recordings available from the universitymedia centre were shown to volunteer students who rated them for felt emotionsAlthough the pre-tests involved many excerpts only two were found to elicitsimilar emotions of different intensity Both lasted about three minutes anddepicted violence inflicted on animals One came from a television movie oncockfighting and portrayed violence inflicted by animals on animals The otherwas taken from a commercially available film called Faces of Death (Philippot1993) and depicts cruelty inflicted by humans on animals Pre-tests showed thatthe latter was more emotionally arousing than the former

A formal pre-test was then conducted in order to ascertain the appropriatenessof the two selected stimuli Twenty undergraduate psychology students (10 menand 10 women) participated on a voluntary basis They were individually showneach of the two excerpts in an order counterbalanced across participants Aftereach excerpt they rated on 7-point scales ranging from 0 (not at all) to 6 (verymuch) the intensity of their reactions with respect to four basic emotionsmdashsadness anger fear and disgust These emotions were selected as appropriateconsidering the type of films displayed Because these four emotional intensityscales exhibited a high internal consistency a = 82 scores were averaged to forma single index of emotional intensity This index was found to be higher for theFaces of Death excerpt (M = 308 SD = 083) than for the cockfighting one (M =

SOCIAL SHARING OF EMOTION 665

204 SD = 098) t(18) = 397 p lt 001 Figure 1 shows that the profiles ofemotions elicited by each of the two movie excerpts were very similar Consistentwith this a MANOVA with movie condition as a between-participants factor andthe four basic emotions as a within-participant factor resulted in a nonsignificantinteraction F(3 54) lt 1 It was thus confirmed that although the Faces of Deathexcerpt elicited more intense emotional reactions than did the cockfightingexcerpt the two excerpts were nevertheless comparable with respect to the qualityof induced emotion A nonemotional control movie was also needed to test thecentral hypothesis A 3-minute excerpt of a film called Wildlife in the Himalayasthat depicted animal life without any violence was chosen for this purpose

Method

Participants

Thirty undergraduate psychology students at the University of LouvainLouvain-la-Neuve Belgium volunteered to participate in a study on lsquolsquocoopera-tionrsquorsquo Each of them agreed to arrive at the laboratory at a specified timetogether with a friend Due to technical problems (ie inaudible tape-recording)7 dyads had to be eliminated The excluded dyads were evenly spread acrossconditions (3 in the nonemotional condition 2 in the moderate emotioncondition and 2 in the intense emotion condition) Thus they were 46participants 17 men and 27 women (gender failed to be recorded for one pair ofparticipants) in the final sample Dyads were equally distributed amongconditions (n = 7 for the nonemotional condition n = 8 for the moderate emotioncondition and n = 8 for the intense emotion condition)

Figure 1 Profiles of emotions as a function of the intensity of the stimulus (pre-test)

666 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

anticipated electric shock Affiliation was investigated both under real-lifethreats with room-mates among hospital patients expecting to undergo majorsurgery (Kulik amp Mahler 1987 1989 Kulik et al 1993) and in laboratorystudies (Gump amp Gulik 1997 Kulik et al 1994) To summarise the stress-affiliation effect is alive Under stressful conditions there is a tendency ofhuman beings to seek the company of others

Being exposed to an emotion-arousing situation not only impacts on anindividualrsquos motivation to social contact Emotion also entails importantconsequences for the personrsquos cognitive activity First clinical and field studiesshowed that after witnessing a traumatic event many people experienceintrusive thoughts or images (eg Bownes OrsquoGorman amp Sayers 1991McCammon Durham Allison amp Williamson 1988 Tait amp Silver 1989Weisaeth 1989 Wilkinson 1983) Second laboratory research demonstratedthat such cognitive consequences are not limited to those who seek psychiatrictreatment after experiencing a highly negative life event Such symptoms aremanifested after even relatively minor or moderate emotion-inducing events (fora review see Horowitz 1975 1992) Third real-life data based on volunteersrsquoreports of recent autobiographic episodes in which they experienced emotionalso showed that not only trauma but emotional situations in general elicitcognitive consequences (Rime Mesquita Philippot amp Boca 1991a RimeNoel amp Philippot 1991b for a review see Rime Philippot Mesquita amp Boca1992) More than 95 of the respondents reported having had recurrent thoughtsabout the emotional episode in the hours or days that followed it A considerableproportion of them mentioned that they have had such thoughts often or veryoften

To summarise one line of evidence reveals that exposure to an emotionalsituation generally entails social consequences A second line of evidence showsthat exposure to an emotional situation has cognitive consequences in the formof intrusive and repetitive thoughts and images related to the emotion-elicitingsituation Considering the two lines of findings jointly gives rise to a newprediction regarding the interpersonal effects of an emotional episode If peopleexposed to an emotional event have an urge to affiliate and if they experiencerecurrent emotion-related thoughts and images the resulting interpersonalsituation can be expected to involve what has been called the lsquolsquosocial sharing ofemotionrsquorsquo (Rime 1989 Rime et al 1991a) The social sharing of emotioninvolves (a) the evocation of the emotion in a socially shared language and (b)at least at the symbolic level an addressee It occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their own feelings and emotional reactionsThere are a number of theoretical and empirical arguments suggesting thatcommunicating about a negative emotional experience may fulfil adaptativefunctions Thus Pennebaker (eg 1989) has developed an extensive line ofresearch on the ways in which talking about traumatic events impacts on

SOCIAL SHARING OF EMOTION 663

physical and psychological adjustment His influential inhibition theory arguesthat talking about highly stressful events significantly decreases psychologicaldistress health complaints and physician visits Considering together intrusivethoughts and talks about a traumatic event Lepore and his colleagues (egLepore 1997 Lepore Silver Wortman amp Wayment 1996) argued that whenboth processes emerge in a supportive environment they facilitate the cognitiveprocessing of the emotional event thereby reducing emotional distress

The reasoning developed in the present introduction leads us to predict thatpeople spontaneously incline in developing this potentially adaptative behaviourwhich consists in talking about their distressing experiences The centralargument is that the stress-affiliation relationship is paralleled by a broader-spectrum phenomenon that occurs in a variety of emotional circumstances andwhich involves talking about the emotion Studies in which volunteers reportedrecent episodes of everyday life emotion show that any emotional episode evenif mild or moderate is generally associated with an urge to talk and to share theemotional experience with others (Rime et al 1991a b for a review see Rimeet al 1992) Participants who were instructed to retrieve from their memory aninstance of an emotion they recently experienced also answered questions suchas Did you talk about the episode with other people With whom How longafter the emotion How often Eight independent studies based on thisprocedure were reviewed by Rime et al (1992) These involved a total of 913participants whose ages ranged from 12 to 60 years and entailed 1384 emotionalepisodes The observed proportion of cases in which the respondent reportedhaving talked about the emotion with others varied from 90 to 96 Talkingabout a negative emotion generally began soon after the eliciting situation andthis sharing was typically repeated A majority of respondents answered thatthey first talked about the event lsquolsquoon the same dayrsquorsquo and that they talked aboutthe episode lsquolsquoseveral times with several personsrsquorsquo Addresses were almostinvariably intimates (ie parents siblings friends or spousepartner) Finallymore disruptive episodes were talked about more extensively than lessdisruptive ones

The results just reviewed are consistent with the present prediction thatpeople who were exposed to an emotional episode spontaneously incline insocially sharing the emotion they experienced However this existing evidenceis correlational in nature A straightforward experimental test of the assumptionthat negative emotion elicits the social sharing of emotion would consist ofexposing participants to negative emotional stimuli varying in intensity and thencomparing the extent of spontaneous talking about the stimuli the reactions tothe stimuli or both Such a test has to our knowledge never been conducted Inthis paper we report three experiments of this type Manipulation of theintensity of emotional experience was achieved by exposing participants todifferent film excerpts For ethical reasons the intensity range was restricted tothe low-to-moderate range that can reasonably be used in a laboratory context

664 LUMINET ET AL

In the first two experiments shortly after being exposed to the movieparticipants were left in a waiting room situation with a close friend who had notseen the same excerpt Close friends were used because previous research hasestablished that they are the most usual type of partner for the social sharing ofemotion among young adults (Rime et al 1991a) The conversations betweenparticipants and friends were unobtrusively tape recorded and later scored withrespect to the extent of social sharing In the third experiment we examinedwhether the social sharing observed under laboratory conditions could bereplicated when assessed in a real-life context In this third experiment we alsoattempted to rule out some plausible nonemotional explanations of the effectsobserved in the first two experiments

EXPERIMENT 1

Pre-testing of emotional stimuli

The hypothesis to be tested concerns the effects on the social sharing of emotionof variations in emotional intensity We therefore needed to vary the intensity ofemotional experience while keeping constant the quality or type of emotioninduced We achieved this by using appropriately selected film excerpts asemotional stimuli The pre-testing procedure was undertaken with nopreconceptions regarding the type of emotion elicited The goal was to identifyat least two excerpts that were comparable in the type of emotion elicited butdiffered in intensity Excerpts from the set of films described by Philippot(1993) and from movies and video-recordings available from the universitymedia centre were shown to volunteer students who rated them for felt emotionsAlthough the pre-tests involved many excerpts only two were found to elicitsimilar emotions of different intensity Both lasted about three minutes anddepicted violence inflicted on animals One came from a television movie oncockfighting and portrayed violence inflicted by animals on animals The otherwas taken from a commercially available film called Faces of Death (Philippot1993) and depicts cruelty inflicted by humans on animals Pre-tests showed thatthe latter was more emotionally arousing than the former

A formal pre-test was then conducted in order to ascertain the appropriatenessof the two selected stimuli Twenty undergraduate psychology students (10 menand 10 women) participated on a voluntary basis They were individually showneach of the two excerpts in an order counterbalanced across participants Aftereach excerpt they rated on 7-point scales ranging from 0 (not at all) to 6 (verymuch) the intensity of their reactions with respect to four basic emotionsmdashsadness anger fear and disgust These emotions were selected as appropriateconsidering the type of films displayed Because these four emotional intensityscales exhibited a high internal consistency a = 82 scores were averaged to forma single index of emotional intensity This index was found to be higher for theFaces of Death excerpt (M = 308 SD = 083) than for the cockfighting one (M =

SOCIAL SHARING OF EMOTION 665

204 SD = 098) t(18) = 397 p lt 001 Figure 1 shows that the profiles ofemotions elicited by each of the two movie excerpts were very similar Consistentwith this a MANOVA with movie condition as a between-participants factor andthe four basic emotions as a within-participant factor resulted in a nonsignificantinteraction F(3 54) lt 1 It was thus confirmed that although the Faces of Deathexcerpt elicited more intense emotional reactions than did the cockfightingexcerpt the two excerpts were nevertheless comparable with respect to the qualityof induced emotion A nonemotional control movie was also needed to test thecentral hypothesis A 3-minute excerpt of a film called Wildlife in the Himalayasthat depicted animal life without any violence was chosen for this purpose

Method

Participants

Thirty undergraduate psychology students at the University of LouvainLouvain-la-Neuve Belgium volunteered to participate in a study on lsquolsquocoopera-tionrsquorsquo Each of them agreed to arrive at the laboratory at a specified timetogether with a friend Due to technical problems (ie inaudible tape-recording)7 dyads had to be eliminated The excluded dyads were evenly spread acrossconditions (3 in the nonemotional condition 2 in the moderate emotioncondition and 2 in the intense emotion condition) Thus they were 46participants 17 men and 27 women (gender failed to be recorded for one pair ofparticipants) in the final sample Dyads were equally distributed amongconditions (n = 7 for the nonemotional condition n = 8 for the moderate emotioncondition and n = 8 for the intense emotion condition)

Figure 1 Profiles of emotions as a function of the intensity of the stimulus (pre-test)

666 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

physical and psychological adjustment His influential inhibition theory arguesthat talking about highly stressful events significantly decreases psychologicaldistress health complaints and physician visits Considering together intrusivethoughts and talks about a traumatic event Lepore and his colleagues (egLepore 1997 Lepore Silver Wortman amp Wayment 1996) argued that whenboth processes emerge in a supportive environment they facilitate the cognitiveprocessing of the emotional event thereby reducing emotional distress

The reasoning developed in the present introduction leads us to predict thatpeople spontaneously incline in developing this potentially adaptative behaviourwhich consists in talking about their distressing experiences The centralargument is that the stress-affiliation relationship is paralleled by a broader-spectrum phenomenon that occurs in a variety of emotional circumstances andwhich involves talking about the emotion Studies in which volunteers reportedrecent episodes of everyday life emotion show that any emotional episode evenif mild or moderate is generally associated with an urge to talk and to share theemotional experience with others (Rime et al 1991a b for a review see Rimeet al 1992) Participants who were instructed to retrieve from their memory aninstance of an emotion they recently experienced also answered questions suchas Did you talk about the episode with other people With whom How longafter the emotion How often Eight independent studies based on thisprocedure were reviewed by Rime et al (1992) These involved a total of 913participants whose ages ranged from 12 to 60 years and entailed 1384 emotionalepisodes The observed proportion of cases in which the respondent reportedhaving talked about the emotion with others varied from 90 to 96 Talkingabout a negative emotion generally began soon after the eliciting situation andthis sharing was typically repeated A majority of respondents answered thatthey first talked about the event lsquolsquoon the same dayrsquorsquo and that they talked aboutthe episode lsquolsquoseveral times with several personsrsquorsquo Addresses were almostinvariably intimates (ie parents siblings friends or spousepartner) Finallymore disruptive episodes were talked about more extensively than lessdisruptive ones

The results just reviewed are consistent with the present prediction thatpeople who were exposed to an emotional episode spontaneously incline insocially sharing the emotion they experienced However this existing evidenceis correlational in nature A straightforward experimental test of the assumptionthat negative emotion elicits the social sharing of emotion would consist ofexposing participants to negative emotional stimuli varying in intensity and thencomparing the extent of spontaneous talking about the stimuli the reactions tothe stimuli or both Such a test has to our knowledge never been conducted Inthis paper we report three experiments of this type Manipulation of theintensity of emotional experience was achieved by exposing participants todifferent film excerpts For ethical reasons the intensity range was restricted tothe low-to-moderate range that can reasonably be used in a laboratory context

664 LUMINET ET AL

In the first two experiments shortly after being exposed to the movieparticipants were left in a waiting room situation with a close friend who had notseen the same excerpt Close friends were used because previous research hasestablished that they are the most usual type of partner for the social sharing ofemotion among young adults (Rime et al 1991a) The conversations betweenparticipants and friends were unobtrusively tape recorded and later scored withrespect to the extent of social sharing In the third experiment we examinedwhether the social sharing observed under laboratory conditions could bereplicated when assessed in a real-life context In this third experiment we alsoattempted to rule out some plausible nonemotional explanations of the effectsobserved in the first two experiments

EXPERIMENT 1

Pre-testing of emotional stimuli

The hypothesis to be tested concerns the effects on the social sharing of emotionof variations in emotional intensity We therefore needed to vary the intensity ofemotional experience while keeping constant the quality or type of emotioninduced We achieved this by using appropriately selected film excerpts asemotional stimuli The pre-testing procedure was undertaken with nopreconceptions regarding the type of emotion elicited The goal was to identifyat least two excerpts that were comparable in the type of emotion elicited butdiffered in intensity Excerpts from the set of films described by Philippot(1993) and from movies and video-recordings available from the universitymedia centre were shown to volunteer students who rated them for felt emotionsAlthough the pre-tests involved many excerpts only two were found to elicitsimilar emotions of different intensity Both lasted about three minutes anddepicted violence inflicted on animals One came from a television movie oncockfighting and portrayed violence inflicted by animals on animals The otherwas taken from a commercially available film called Faces of Death (Philippot1993) and depicts cruelty inflicted by humans on animals Pre-tests showed thatthe latter was more emotionally arousing than the former

A formal pre-test was then conducted in order to ascertain the appropriatenessof the two selected stimuli Twenty undergraduate psychology students (10 menand 10 women) participated on a voluntary basis They were individually showneach of the two excerpts in an order counterbalanced across participants Aftereach excerpt they rated on 7-point scales ranging from 0 (not at all) to 6 (verymuch) the intensity of their reactions with respect to four basic emotionsmdashsadness anger fear and disgust These emotions were selected as appropriateconsidering the type of films displayed Because these four emotional intensityscales exhibited a high internal consistency a = 82 scores were averaged to forma single index of emotional intensity This index was found to be higher for theFaces of Death excerpt (M = 308 SD = 083) than for the cockfighting one (M =

SOCIAL SHARING OF EMOTION 665

204 SD = 098) t(18) = 397 p lt 001 Figure 1 shows that the profiles ofemotions elicited by each of the two movie excerpts were very similar Consistentwith this a MANOVA with movie condition as a between-participants factor andthe four basic emotions as a within-participant factor resulted in a nonsignificantinteraction F(3 54) lt 1 It was thus confirmed that although the Faces of Deathexcerpt elicited more intense emotional reactions than did the cockfightingexcerpt the two excerpts were nevertheless comparable with respect to the qualityof induced emotion A nonemotional control movie was also needed to test thecentral hypothesis A 3-minute excerpt of a film called Wildlife in the Himalayasthat depicted animal life without any violence was chosen for this purpose

Method

Participants

Thirty undergraduate psychology students at the University of LouvainLouvain-la-Neuve Belgium volunteered to participate in a study on lsquolsquocoopera-tionrsquorsquo Each of them agreed to arrive at the laboratory at a specified timetogether with a friend Due to technical problems (ie inaudible tape-recording)7 dyads had to be eliminated The excluded dyads were evenly spread acrossconditions (3 in the nonemotional condition 2 in the moderate emotioncondition and 2 in the intense emotion condition) Thus they were 46participants 17 men and 27 women (gender failed to be recorded for one pair ofparticipants) in the final sample Dyads were equally distributed amongconditions (n = 7 for the nonemotional condition n = 8 for the moderate emotioncondition and n = 8 for the intense emotion condition)

Figure 1 Profiles of emotions as a function of the intensity of the stimulus (pre-test)

666 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

In the first two experiments shortly after being exposed to the movieparticipants were left in a waiting room situation with a close friend who had notseen the same excerpt Close friends were used because previous research hasestablished that they are the most usual type of partner for the social sharing ofemotion among young adults (Rime et al 1991a) The conversations betweenparticipants and friends were unobtrusively tape recorded and later scored withrespect to the extent of social sharing In the third experiment we examinedwhether the social sharing observed under laboratory conditions could bereplicated when assessed in a real-life context In this third experiment we alsoattempted to rule out some plausible nonemotional explanations of the effectsobserved in the first two experiments

EXPERIMENT 1

Pre-testing of emotional stimuli

The hypothesis to be tested concerns the effects on the social sharing of emotionof variations in emotional intensity We therefore needed to vary the intensity ofemotional experience while keeping constant the quality or type of emotioninduced We achieved this by using appropriately selected film excerpts asemotional stimuli The pre-testing procedure was undertaken with nopreconceptions regarding the type of emotion elicited The goal was to identifyat least two excerpts that were comparable in the type of emotion elicited butdiffered in intensity Excerpts from the set of films described by Philippot(1993) and from movies and video-recordings available from the universitymedia centre were shown to volunteer students who rated them for felt emotionsAlthough the pre-tests involved many excerpts only two were found to elicitsimilar emotions of different intensity Both lasted about three minutes anddepicted violence inflicted on animals One came from a television movie oncockfighting and portrayed violence inflicted by animals on animals The otherwas taken from a commercially available film called Faces of Death (Philippot1993) and depicts cruelty inflicted by humans on animals Pre-tests showed thatthe latter was more emotionally arousing than the former

A formal pre-test was then conducted in order to ascertain the appropriatenessof the two selected stimuli Twenty undergraduate psychology students (10 menand 10 women) participated on a voluntary basis They were individually showneach of the two excerpts in an order counterbalanced across participants Aftereach excerpt they rated on 7-point scales ranging from 0 (not at all) to 6 (verymuch) the intensity of their reactions with respect to four basic emotionsmdashsadness anger fear and disgust These emotions were selected as appropriateconsidering the type of films displayed Because these four emotional intensityscales exhibited a high internal consistency a = 82 scores were averaged to forma single index of emotional intensity This index was found to be higher for theFaces of Death excerpt (M = 308 SD = 083) than for the cockfighting one (M =

SOCIAL SHARING OF EMOTION 665

204 SD = 098) t(18) = 397 p lt 001 Figure 1 shows that the profiles ofemotions elicited by each of the two movie excerpts were very similar Consistentwith this a MANOVA with movie condition as a between-participants factor andthe four basic emotions as a within-participant factor resulted in a nonsignificantinteraction F(3 54) lt 1 It was thus confirmed that although the Faces of Deathexcerpt elicited more intense emotional reactions than did the cockfightingexcerpt the two excerpts were nevertheless comparable with respect to the qualityof induced emotion A nonemotional control movie was also needed to test thecentral hypothesis A 3-minute excerpt of a film called Wildlife in the Himalayasthat depicted animal life without any violence was chosen for this purpose

Method

Participants

Thirty undergraduate psychology students at the University of LouvainLouvain-la-Neuve Belgium volunteered to participate in a study on lsquolsquocoopera-tionrsquorsquo Each of them agreed to arrive at the laboratory at a specified timetogether with a friend Due to technical problems (ie inaudible tape-recording)7 dyads had to be eliminated The excluded dyads were evenly spread acrossconditions (3 in the nonemotional condition 2 in the moderate emotioncondition and 2 in the intense emotion condition) Thus they were 46participants 17 men and 27 women (gender failed to be recorded for one pair ofparticipants) in the final sample Dyads were equally distributed amongconditions (n = 7 for the nonemotional condition n = 8 for the moderate emotioncondition and n = 8 for the intense emotion condition)

Figure 1 Profiles of emotions as a function of the intensity of the stimulus (pre-test)

666 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

204 SD = 098) t(18) = 397 p lt 001 Figure 1 shows that the profiles ofemotions elicited by each of the two movie excerpts were very similar Consistentwith this a MANOVA with movie condition as a between-participants factor andthe four basic emotions as a within-participant factor resulted in a nonsignificantinteraction F(3 54) lt 1 It was thus confirmed that although the Faces of Deathexcerpt elicited more intense emotional reactions than did the cockfightingexcerpt the two excerpts were nevertheless comparable with respect to the qualityof induced emotion A nonemotional control movie was also needed to test thecentral hypothesis A 3-minute excerpt of a film called Wildlife in the Himalayasthat depicted animal life without any violence was chosen for this purpose

Method

Participants

Thirty undergraduate psychology students at the University of LouvainLouvain-la-Neuve Belgium volunteered to participate in a study on lsquolsquocoopera-tionrsquorsquo Each of them agreed to arrive at the laboratory at a specified timetogether with a friend Due to technical problems (ie inaudible tape-recording)7 dyads had to be eliminated The excluded dyads were evenly spread acrossconditions (3 in the nonemotional condition 2 in the moderate emotioncondition and 2 in the intense emotion condition) Thus they were 46participants 17 men and 27 women (gender failed to be recorded for one pair ofparticipants) in the final sample Dyads were equally distributed amongconditions (n = 7 for the nonemotional condition n = 8 for the moderate emotioncondition and n = 8 for the intense emotion condition)

Figure 1 Profiles of emotions as a function of the intensity of the stimulus (pre-test)

666 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Procedure

The experiment was presented as being concerned with the effects ofcooperation on performance in different types of tasks On arrival at thelaboratory each pair of participants was met by two experimenters (E1 and E2)and it was explained that they would first perform a task individually and thenperform a task together One member of the pair was then randomly assigned tothe lsquolsquotargetrsquorsquo role leaving the lsquolsquopartnerrsquorsquo role to the other participant The targetwas then randomly assigned to view one of the three movie excerpts Heshe wasescorted by E1 to another room and seated facing a video monitor on which oneof the three movie excerpts was to be displayed The target was asked to focuson the film excerpt and when finished to wait till E2 arrived Meanwhile thepartner completed a test (Matrix D48) presented as a test of mentalconcentration in the original room Completing this task took about the sametime as watching the movie excerpt Immediately afterwards the partner wasbrought by E2 to the room where the target had seen the video On the pretext ofhaving to prepare the original room for the lsquolsquocooperation taskrsquorsquo E2 asked bothparticipants to wait there for a few minutes and then left All conversation thattook place during this waiting period was tape recorded for subsequent analysisAfter 5 minutes E2 returned to the room and asked the target to complete ratingsof intensity of emotions elicited by the movie These ratings were collected inorder to check the experimental manipulation They were deliberately takenafter the dependent measures to avoid the possibility that a procedure-inducedemotion labelling would impact on social sharing Participants were not madeaware before the experiment that their discourse would be being tape recordedbecause doing so would have undermined the naturalistic character of thesetting After completing these ratings the participants were thanked and fullydebriefed concerning the aims of the study and the need for the deceptionconcerning the lsquolsquocooperation taskrsquorsquo Participantsrsquo permission to use therecording was requested and they were assured that the recordings would betreated with utmost confidentiality All participants gave permission for theirrecordings to be used for research purposes

Movie excerpts

The cockfighting excerpt was used to induce moderate emotion It depicts thedifferent steps involved in a cockfight from the initial confrontation to injuryto and finally death of one cock The intense emotion condition was created byshowing the Faces of Death excerpt this depicts a scene in an lsquolsquoexoticrsquorsquorestaurant A group of people sitting at a table first kill and then eat the brains ofa small monkey The nonemotional control condition was created by showing anexcerpt from Wildlife in the Himalayas This depicts animals and plants in theHimalayan mountains with an accent on how species can adapt and survive inthis inhospitable environment

SOCIAL SHARING OF EMOTION 667

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Measures

Intensity of emotion As a manipulation check participants rated theintensity of emotions they had experienced while watching the movie on thesame four scales as used in the pre-test

Social sharing Extent of social sharing was assessed from the tape-recordings of the conversation held by the two participants during the 5-minutelsquolsquowaitingrsquorsquo situation The scoring procedure was based on the formal definitionof the social sharing of emotion which states that it occurs when individualscommunicate openly with one or more others about the circumstances of theemotion-eliciting event and about their emotional reactions to the event (Rime1989 Rime et al 1991a) Two independent judges who were blind to theparticipantsrsquo condition listened to the tape-recordings and pressed a key everytime the conversational content satisfied the above-mentioned criterion and keptthe key depressed until the criterion was no longer satisfied The judgesrsquo keyswere connected to a computer programmed to keep a cumulative record of howlong these keys were depressed In this way the judges recorded the time spentby the interactants in talking about the content of the film excerpt andor theemotions it elicited

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 90 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 20) = 5667 p lt 001 and post-hoc Tukey tests confirmed thatthe manipulation was successful in eliciting three different levels of intensity ofelicited emotion with M = 007 SD = 019 for the nonemotional condition M =227 SD = 088 for the moderate emotion condition and M = 354 SD = 093for the intense emotion condition

Social sharing1

The two judgesrsquo scores for the extent of social sharing exhibited highagreement r(21) = 95 p lt 001 A duration score was awarded to each pairof participants by averaging the time recorded by the two judges for that pairand by dividing this average by the total time spent by the pair in

1 Preliminary analyses of the present dataset and those collected in Experiments 2 and 3 using sexof participant as a factor revealed no significant effects involving gender This variable was thereforedropped from further analyses These results are consistent with previous studies (see Rime et al1992)

668 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

conversation2 Duration of social sharing scores were entered into a one-wayANOVA with movie condition as a between-participants factor This yieldeda significant main effect F(2 20) = 1000 p lt 001 Post-hoc Tukey testsrevealed that duration of social sharing was significantly higher in the intenseemotion condition (M = 0195 SD = 0114) than in the nonemotional (M =0044 SD = 0083) and moderate emotion (M = 0025 SD = 0022)conditions However the latter two did not differ significantly from eachother3 The intense emotion condition was the only one in which socialsharing was substantial enough to justify calculating the individual contribu-tions of targets and partners to the social sharing process Consistent with thehypothesis that a person exposed to an emotion would socially share ittargets contributed to the social sharing component of the conversation to amuch greater extent than did partners (818 and 182 of the timerespectively)

Discussion

With regard to the relation between emotional intensity and extent of socialsharing the present results are consistent with the notion of a nonlinear relationThe three films induced three different levels of emotional intensity Howeverthe moderate emotion condition and the nonemotional condition were equally

2 By considering as the denominator the total time spent in conversation rather than the total timeavailable we were able to assess specifically the time during which the control on the verbalisationprocess was optimal During moments of silence no control was possible as whether people were stillinteracting nonverbally (ie smiles glances) or whether they did not communicate at all eitherverbally or nonverbally Therefore although the present measure does not consider the nonverbalcomponent of social sharing it ensures a high control on the verbal aspect of communication

3 By using a relative index of social sharing rather than an absolute one it is not possible todistinguish the difference between two aspects (a) that the pairs in each condition talked forapproximately equivalent amounts of time with primarily those in the intense condition sharingemotions and (b) that only those in the intense condition were talking much at all In Experiment 1beside the variable of duration of social sharing we also used a measure based on the percentage ofwords used in social sharing This measure was computed by dividing the total number of wordsabout the film excerpt andor emotional reactions to the excerpt that were uttered during theconversation by the total number of words uttered by the two subjects during the time available forconversation The index of percentage of words used in social sharing evidenced a significant effectof movie type F(2 20) = 1504 p lt 001 Post-hoc tests revealed that percentage of words used insocial sharing was significantly higher in the intense emotional condition (M = 0429 SD = 0236)than in the moderate (M = 0065 SD = 0046) and the nonemotiona l ones (M = 0059 SD = 0107)but that these conditions did not differ from each other Despite this difference an ANOVA with thetotal number of words uttered by the two subjects during the total time available for conversation(ie 5 minutes) did not evidence a main effect of movie time This indicates that independently ofthe topic participants were talking about in the different conditions the total amount of conversationwas not affected by movie type As the index on duration of social sharing and the one on percentageof words spent in social sharing had been found to be highly intercorrelated (r = 95) this measurewas not reported in Experiment 1 and was not used in Experiment 2

SOCIAL SHARING OF EMOTION 669

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

low with respect to the amount of social sharing they induced whereas theintense emotion condition elicited more social sharing To establish whether thispattern of findings was reliable we conducted a replication

Consistent with the idea that being exposed to emotion would induce sharingthe results showed that the movie-related exchanges between targets andpartners were dominated by contributions from targets One question that arisesis whether targets really shared the emotional feelings they experienced whenexposed to the movie or whether they confined their contributions to a meredescription of the content of the movie These two types of social sharing werenot discriminated in the measure taken in this study If participants wereprimarily engaged in describing movie content this might be regarded as fallingshort of social sharing of emotion Addressing this question requires a more fine-grained analysis of the content of the recorded verbal exchanges

EXPERIMENT 2

We had two purposes in conducting this experiment First the same researchdesign and procedure were used in order to assess the replicability of the patternof findings obtained in the first experiment Second the measure of socialsharing involved a more detailed content analysis of the recordings of the verbalexchanges between targets and partners

Method

Participants

A total of 60 undergraduate students at the University of Louvain Louvain-la-Neuve Belgium were recruited via announcements made during introductorypsychology classes The study was described as investigating lsquolsquocooperationbetween friendsrsquorsquo Each participant was asked to arrive at the laboratory at apredetermined time together with a friend They participated in this study on anunpaid voluntary basis Due to technical problems (ie a substantial part of theconversation was inaudible) the data from seven dyads had to be eliminatedThe excluded dyads were evenly spread across conditions (two in thenonemotional condition two in the moderate emotion condition and three inthe intense emotion condition) Thus the final sample consisted of 106participants 16 of whom were men There were 15 dyads in the nonemotionalcondition 20 in the moderate emotion condition and 19 in the intense emotioncondition

Procedure

The procedure of Experiment 1 was replicated with only one modificationWe suspected that the intellectual task performed by partner participants inExperiment 1 while targets viewed the movie excerpt may have induced stress or

670 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

negative feelings In order to rule out any unwanted consequences of feelingsinduced in partners the intellectual task was replaced by a neutral situation inwhich the partner simply listened to relaxing music

Measures

Intensity of emotion As a manipulation check at the end of the waitingperiod and before the final debriefing the target was instructed to rate theintensity of the emotions elicited by the movie excerpt using the same fouremotional scales described previously

Rating of interest To check for the possibility that the movies would not becomparable with respect to the degree of interest they elicited participants werealso asked to rate the extent to which they had been interested by the film Thisrating was also made on a 7-point scale ranging from 0 (not at all) to 6 (verymuch)

Duration of social sharing Duration of social sharing was scoredseparately for each successive minute of the recorded conversation in orderto gain a better insight into the temporal dynamics of social sharing Theseassessments were performed independently by two judges who were blind withrespect to condition using the same procedure as in Experiment 1

Content of social sharing We adopted Balesrsquo (1951) lsquolsquointeraction processanalysisrsquorsquo (IPA) coding system for this purpose IPA was developed on thebasis of extensive observation of group discussions It comprises 12categories The content of each communication addressed by one individualto another can be classified into these categories Three categories describepositive statements (a) shows solidarity (b) shows release of tension bylaughing or by expressing satisfaction (c) shows passive agreement Threecategories describe negative statements (a) shows antagonism (b) showstension (c) shows passive disagreement Three further categories concerngiving information to others (a) gives information clarifies repeats orconfirms (b) expresses evaluation or feelings (c) makes suggestion Finallythree categories concern requesting information from others (a) asks forinformation repetition or clarification (b) asks for opinion or for expressionof feelings (c) asks for suggestion Two judges were trained in the use of thissystem They were then given transcripts of each target-partner conversationand were instructed to rate the partner and the target separately Utteranceswere not coded exclusively giving judges the possibility to select more thanone category for each conversation sequence Therefore the percentages indifferent categories can exceed 100

SOCIAL SHARING OF EMOTION 671

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Results

Manipulation check

The index of intensity of emotion based on participantsrsquo ratings of the fouremotion scales a = 86 was entered into a one-way ANOVA with moviecondition as a between-participants factor This analysis revealed a significantmain effect F(2 50) = 3216 p lt 001 Post-hoc Tukey tests revealed that thethree films were significantly different with M = 021 SD = 041 for thenonemotional condition M = 256 SD = 136 for the moderate emotioncondition and M = 378 SD = 132 for the intense emotion condition Theseresults are very similar to those of Experiment 1

Interest

On the rating scale ranging from 0 to 6 the nonemotional excerpt attracted anaverage rating of 371 (SD = 107) the comparable scores for the moderate andintense emotion excerpts were 380 (SD = 070) and 426 (SD = 087)respectively A one-way ANOVA with movie condition as a between-participants factor found no significant effect F(2 50) = 205 ns Z2 = 008The three films can therefore be considered to be comparable with respect to theamount of interest they elicited4

Duration of social sharing

Interjudge reliability was assessed by examining the correlation between thetotal social sharing times recorded by the two judges across pairs of participantsA high level of agreement was found r(58) = 95 p lt 001 The minute-by-minute social sharing time awarded to each pair of participants by each judgewas then averaged and entered into a two-way ANOVA with movie type as abetween-participants factor and time of measurement (first to fifth minute of thewaiting period) as a within-participants factor This analysis revealed significantmain effects for both movie type F(2 50) = 1450 p lt 001 and time ofmeasurement F(4 196) = 2287 p lt 001 These main effects were qualified bya significant interaction F(8 196) = 506 p lt 001 The effect size for the maineffect of movie type was large Z2 = 37 As can be seen in Table 1 duration ofsocial sharing was higher in the intense emotion condition than in the other twoconditions The main effect of time was due to a general tendency for duration of

4 The pattern of means for interest indicate that the nonemotiona l and the moderate emotionconditions are similar to one another and quite different from the intense emotion condition To testfor this possibility we conducted a planned comparison analysis pitting the intense condition againstthe other two This revealed an effect that just reached significance F(2 50) = 409 p = 05suggesting that the amount of interest elicited by the movie was slightly higher in the intense than inthe moderate and nonemotiona l conditions

672 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

social sharing to decline as time elapsed With regard to the significantinteraction post-hoc contrasts revealed that the intense emotion conditionelicited significantly more sharing during the first and the second minutesThereafter the difference between movie conditions disappeared

Covariate analyses

Another way to examine the respective contribution of interest and emotionalintensity to predict the duration of social sharing is to conduct analyses ofcovariance to test (a) the effect when interest is partialled out (b) the effectwhen emotional intensity is partialed out After controlling for interest the maineffect of the movie was still highly significant F(2 48) = 1292 p lt 001 theeffect size remaining similar Z2 = 34 After controlling for emotional intensitythe main effect of movie type also remained significant F(2 48) = 580 p lt 01However the effect size was considerably reduced Z2 = 19

Content of social sharing

As in Experiment 1 in both the moderate and the nonemotional conditionssocial sharing was too scarce in the conversations to lend itself to contentanalysis Such an analysis was therefore only conducted on the conversations inthe intense emotion condition involving 19 pairs of participants It was foundthat 893 of the social sharing time entailed contributions from targetswhereas partners spoke for the remaining 117 Inter-rater consistency in IPAscoring was measured by Cohenrsquos (1960) kappa and agreement was found to bevery high with k = 95 for targetsrsquo utterances and k = 89 for partnersrsquoutterances As can be seen in Figure 2 two of the 12 categories accounted for anoverwhelming proportion of targetsrsquo utterances They lsquolsquogave informationrsquorsquo aboutthe film in 947 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film in 842 of their utterances Compared to these proportions the scoresfor the 10 remaining categories are negligible Scoring of partnersrsquo utterances

TABLE 1Duration of social sharing (as proportion of total time spent conversing) as a function of

strength of emotional stimulus and time (Experiment 2)

Strength ofstimulus Minute 1 Minute 2 Minute 3 Minute 4 Minute 5

Nonemotional 077a 051ab 003a 000a 000a

Moderate 155a 011a 035a 021a 008a

Intense 434b 126b 075a 041a 075a

Note Subscripts refer to the results of post-hoc Tukey tests (within columns) Means not sharinga common subscript differ significantly (p lt 05)

SOCIAL SHARING OF EMOTION 673

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

was also dominated by two categories Partners lsquolsquorequested informationrsquorsquo fromthe target in 737 of their utterances and they lsquolsquoexpressed feelingsrsquorsquo related tothe film described by the target in 789 of the utterances

Discussion

The manipulation check based on target participantsrsquo self-report of felt emotionsprovided a straightforward confirmation of what had been observed in the pre-testand in Experiment 1 Three different levels of intensity of induced emotion wereunambiguously evoked by the movie excerpts Similarly the pattern of findingsrecorded for social sharing in Experiment 1 was precisely replicated in the presentstudy Social sharing occurred only to a negligible extent in the moderate and thenonemotional emotion conditions whereas it was substantial in the intenseemotion condition The findings again suggest that intensity of felt emotion andextent of sharing are not linked monotonically The present findings also indicatethat the differences in social sharing cannot be attributed to interest elicited by themovies Analyses of covariance indicated that the effect size for duration of socialsharing was considerably reduced when partialling out emotional intensitywhereas the effect size remained similar when controlling for interest

Before discussing the findings one other issue needs to be considered It couldbe argued that the social sharing observed in the intense emotion condition wasrelatively short-lived given that it disappeared after the second minute of the

Figure 2 Content of social sharing of emotion for the target and the partner by type of categoryfrom the interaction process analysis of Bales (Experiment 2)

674 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

waiting situation This contrasts with the findings of questionnaires and fieldstudies showing that the social sharing of emotion is a repetitive process thatbegins soon after the emotional event and develops across the days following thisexperience (Rime et al 1991a b for reviews see Rime et al 1992 RimeFinkenauer Luminet Zech amp Philippot 1998) It is of course possible that theartificial character of the laboratory situation attenuated the extent of sharing Yetto be consistent with the observations from questionnaires and field studies onewould anticipate the social sharing elicited by a laboratory-induced emotion to beevident beyond the two minutes immediately following the emotional event Itmay well be that participants share their movie-related emotion outside thelaboratory situation in the days following their participation in the experimentThis is a possibility that we examined in Experiment 3 reported later

A further purpose of the present experiment was to examine the extent to whichthe waiting room conversations occurring after the intense emotion conditionreally consisted of social sharing of emotion on the part of the target participantsA clear answer to this question was obtained from content analysis of the movie-related parts of these conversations First about 90 of the verbal exchangeswere utterances by targets Second targetsrsquo utterances predominantly consisted oflsquolsquogiving informationrsquorsquo and lsquolsquoexpressing feelingsrsquorsquo about the movie therebyconforming to the formal definition of the social sharing of emotion Third thepartnersrsquo utterances were found to be brief and to consist mainly of requests forinformation and expressions of feelings elicited by the targetsrsquo sharing In thisregard the findings are quite consistent with previous research showing that thesharing of an emotion elicits emotional responses in the listener (Archer amp Berg1978 Christophe amp Rime 1997 Shortt amp Pennebaker 1992) It is also worthnoting that in addition to expressing feelings related to the movie almost all thetargets provided information about the movie to their partner Similarly a highproportion of the partners requested such information These results suggest thatsocial sharing fulfils informational goals as well as emotional ones This seemsperfectly reasonable if we bear in mind the fact that if there were no sharedcontext the partner would not be able to understand what had happened to thetarget and would therefore not be able to provide appropriate social responses

EXPERIMENT 3

The main purpose of this experiment was to test the prediction that exposure toan emotional event in a laboratory would elicit social sharing of this emotion ineveryday life in the days following participation in the experiment Participantswere again individually assigned to one of the three conditions used in thepreceding studies Instead of collecting our dependent measure in the waitingroom situation participants were simply asked to complete a brief questionnairetwo days after the experimental session In the questionnaire they were askedabout social sharing relating to the experimental movie since they had seen it

SOCIAL SHARING OF EMOTION 675

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Additionally we took the opportunity to examine a number of possiblealternative explanations for the effects observed in the first two experimentsFirst we again checked the level of interest elicited by the excerpts Secondbecause strong emotional situations often involve vivid images that return toconsciousness easily (Horowitz amp Reidbord 1992) it may be that the vividnessof these mental images not emotion per se underlies any observed socialsharing We therefore examined the extent to which the various excerpts elicitedvivid mental images Third it may be the case that the excerpts vary with respectto the ease with which their contents can be verbalised Thus if the Faces of theDeath excerpt lends itself better to verbal description than the two other moviesthis might account for the observed differences in extent of social sharing Wealso examined this possibility in the present experiment Fourth it is possiblethat the movies differ with respect to the kinds of appraisals they evoke whichcould then lead to differences in social sharing It is generally accepted that theemotion process commences with a subprocess of appraisal in which theorganism scans internal and external information to assess whether any ongoingevent is relevant for its survival well-being goals andor values (Smith 1993)The environment is appraised with respect to several dimensions (eg Frijda1986 Leventhal amp Scherer 1987 Ortony Clore amp Collins 1988 Smith 1993)four of which were selected in the present experiment on the grounds of theirpresumed relevance for the type of excerpts used These dimensions werecertainty clarity novelty and importance

Method

Participants

These were 60 students at the University of Louvain Louvain-la-NeuveBelgium They were recruited by two students either from the library of thepsychology department or from among their acquaintances They were invited toparticipate (on an unpaid voluntary basis) in a study designed to assess peoplersquospersonal reactions to movies Five participants did not return the questionnairethey were supposed to complete two days after the experiment They weretherefore discarded for further analyses All the analyses were computed usingdata from the remaining 55 participants of whom 27 were men Participantswere evenly distributed across the three conditions (n = 19 for the nonemotionalcondition n = 18 for the moderate emotion condition and n = 18 for the intenseemotion condition)

Procedure

Participants arrived at the laboratory on their own The experimenterexplained that they would be shown a short excerpt from a movie which couldelicit negative emotions and that their task would simply be to complete

676 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

questionnaires after viewing the excerpt reporting how they had reacted to theexcerpt The participant was then randomly assigned to viewing one of the threemovie excerpts already used in the previous experiments A video-recorder wasstarted and the participant was left alone to view the excerpt At the end of themovie participants completed a questionnaire assessing their emotionalreactions Immediately after this it was explained that the study also requiredthem to complete a further brief questionnaire after a delay of 48 hoursParticipants were each given an envelope which they were instructed to open ata specified time two days later The experimenter insisted that the validity oftheir contribution to the study depended on their scrupulous following of theinstruction The envelope contained questions investigating participantsrsquo socialsharing of their movie experience with current life partners in the course of thetwo days following the experiment Participants were asked to return thequestionnaire to the experimenter at a pre-arranged time usually the same dayon which they had completed it During this visit they were thanked and fullydebriefed concerning the aims of the study

Measures

Intensity of emotion and additional variables The questionnaire containingmeasures of these variables was completed by participants immediately afterviewing the movie It first assessed the intensity of emotions elicited by viewingthe movie Participants rated their emotional reactions on the same four emotionscales used previously a = 88 These ratings were then followed by additionalitems intended to assess additional variables The degree of interest elicited bythe movie was measured using four items First participants rated the extent towhich they personally had been interested in the film on a 7-point scale rangingfrom 0 (lsquolsquoI was not interested at allrsquorsquo) to 6 (lsquolsquoI was very much interestedrsquorsquo)Further items concerned the extent to which the film was lsquolsquocaptivatingrsquorsquolsquolsquointerestingrsquorsquo and lsquolsquoattention grabbingrsquorsquo Ratings were made on 7-point scalesranging from 0 (completely disagree) to 6 (completely agree) Because thesefour items were highly intercorrelated a = 74 a single index of interest wascreated by averaging the collected responses The next item assessed the easewith which the excerpt could be visualised in terms of mental images Ratingswere ranked on a 7-point scale ranging from 0 (very difficult to visualise) to 6(very easy to visualise) Another item assessed the ease with which feelingselicited by the excerpt could be explained verbally Ratings were made on a 7-point scale ranging from 0 (very difficult to explain in words) to 6 (very easy toexplain in words) Finally respondents assessed the movie on four 7-pointappraisal dimensions with response scales ranging from 0 to 6 The firstdimension concerned the participantrsquos certainty about the movie and the scaleranged from lsquolsquoexpectedrsquorsquo to lsquolsquounexpectedrsquorsquo The second dimension assessed theclarity of the situation and the scale ranged from lsquolsquoclearrsquorsquo to lsquolsquoconfusedrsquorsquo The

SOCIAL SHARING OF EMOTION 677

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

third dimension concerned novelty and the scale anchors were lsquolsquousualrsquorsquo andlsquolsquounusualrsquorsquo The fourth dimension was about the personal importance of themovie and the scale ranged from lsquolsquounimportantrsquorsquo to lsquolsquoimportantrsquorsquo

Social sharing The questionnaire to be completed by participants two daysafter the laboratory session was deliberately kept brief in order to maximise thechances that participants would complete it Extent of social sharing wasmeasured by means of three questions chosen to reflect three key aspects ofsocial sharing First lsquolsquofelt the need to speak about the movie experiencersquorsquo wasintended to assess participantsrsquo motivation to engage in sharing ratings weremade on a 7-point scale ranging from 0 (never) to 6 (very often) Secondlsquolsquoactual frequency of social sharingrsquorsquo was intended to tap the extent of thesharing that occurred during the two days responses were made on a 7-pointscale ranging from 0 (never) to 6 (more than 8 times) Finally lsquolsquonumber ofpartners involved in sharingrsquorsquo was intended to assess the extent to which theinformation was spread across different persons responses were made on a 7-point scale ranging from 0 (none) to 6 (more than 8 people) The three scaleswere found to have a high degree of internal consistency a = 92 and they weretherefore averaged to form a compound measure of extent of social sharing

Results

Manipulation check

The index of intensity of emotion was entered into a one-way ANOVA withcondition as a between-participants factor This revealed a significant maineffect F(2 52) = 18398 p lt 001 and post-hoc Tukey tests showed once morethat the three conditions differed significantly with M = 016 SD = 030 for thenonemotional condition M = 265 SD = 064 for the moderate emotioncondition and M = 325 SD = 058 for the intense emotion condition Theseresults replicate those of Experiments 1 and 2

Additional variables

Each variable was entered into a one-way ANOVA with condition serving asthe between-participants factor No significant effects were found for interestF(2 52) = 093 ns Z2 = 03 ease of visualisation F(2 52) = 197 ns Z2 =07 ease of verbalisation F(2 52) lt 1Z2 = 00 certainty F(2 52) = 150 nsZ2 = 05 or clarity F(2 52) = 248 ns Z2 = 09 However a significant effectwas found for novelty F(2 52) = 3420 p lt 001 Z2 = 57 and for importanceF(2 52) = 599 p lt 01 Z2 = 16 Post-hoc Tukey tests showed that for noveltythe three conditions differed significantly from each other (M = 145 SD = 150for the nonemotional condition M = 318 SD = 159 for the moderate emotioncondition and M = 522 SD = 106 for the intense emotion condition) For

678 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

importance the nonemotional condition (M = 215 SD = 153) differed from theother two however the moderate emotion condition (M = 382 SD = 143) didnot differ significantly from the intense emotion condition (M = 367 SD =194)

Social sharing

The compound index of extent of social sharing was entered into a one-wayANOVA with condition as the between-participants factor The main effect wassignificant F(2 52) = 2665 p lt 001 The strength of the relationship was highwith Z2 = 51 Post-hoc Tukey tests revealed that although the extent of socialsharing in the nonemotional (M = 091 SD = 084) and the moderate emotion (M= 170 SD = 113) conditions did not differ from each other both differedsignificantly from the intense emotion condition (M = 319 SD = 091) The datashow once again that social sharing was low and comparable in thenonemotional and the moderate emotion conditions and that a higher level ofsharing follows exposure to the intense emotion condition The pattern offindings observed in Experiments 1 and 2 was therefore replicated here despitethe different nature of the measure of social sharing

Covariate and correlational analyses

The extent to which emotional intensity interest ease of visualisation easeof verbalisation certainty clarity novelty and importance were able to predictvariation in the duration of social sharing was examined in a series of analysesof covariance After controlling for interest the main effect of the movie wasstill highly significant F(2 51) = 2369 p lt 001 the effect size being onlyslightly lower Z2 = 48 A similar pattern was observed when controlling forease of visualisation F(2 51) = 2364 p lt 001 Z2 = 48 ease of verbalisationF(2 51) = 2610 p lt 001 Z2 = 51 certainty F(2 51) = 2774 p lt 001 Z2 =52 clarity F(2 51) = 2609 p lt 001 Z2 = 51 and importance F(2 51) =2017 p lt 001 Z2 = 44 When controlling for emotional intensity the maineffect of movie type remained significant F(2 51) = 736 p lt 01 but the effectsize was much smaller Z2 = 22 A similar pattern was observed whencontrolling for novelty F(2 51) = 691 p lt 01 Z2 = 21 The similarity ofresults observed when controlling for emotional intensity and for noveltysuggest some degree of overlap between these variables Indeed the product-moment correlation between emotional intensity and novelty is substantial r =71 p lt 001 indicating that their shared variance is just above 50 Furtherevidence that the two variables are linked is found in the very similar pattern ofcorrelations between emotional intensity and extent of social sharing r = 61 plt 001 and between novelty and extent of social sharing r = 63 p lt 001Strong overlap between emotional intensity and novelty is also evidenced bypartial correlations The partial correlation between emotional intensity and

SOCIAL SHARING OF EMOTION 679

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

extent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andnovelty) r = 32 p lt 05 was of the same order as that between novelty andextent of social sharing controlling for all the remaining variables (interest easeof verbalisation ease of visualisation certainty clarity importance andemotional intensity) r = 24 p lt 10 Together these results suggest that inthe present context the measures of emotional intensity and novelty tap closelyrelated constructs

GENERAL DISCUSSION

Data of the three experiments consistently confirmed that the movie excerptsused induced different intensities of emotion They also confirmed thatmoderate and intense conditions were homogeneous for type of emotionprovoked In each of the three experiments the intense emotion conditionconsistently led to a significantly higher extent of social sharing than did themoderate and the nonemotional conditions which consistently did not differfrom one another However the latter two conditions failed to be discrimi-nated for this variable The observed effect was not limited to the laboratorycontext When participants in Experiment 3 were asked to report the extent ofsocial sharing in the two days following exposure to the emotional stimulusthose in the intense emotion condition again engaged in more social sharingthan those in the two other conditions Together these results strongly suggestthat the emotional intensity of the stimulus determined the extent of socialsharing

When people socially share an emotional experience they may do so eitherby explaining the circumstances of the emotional event or by describing theiremotional reactions to this event or both In Experiment 2 this issue wasexamined by monitoring participants verbal interactions immediately afterexposure to the emotional stimulus A content analysis of the transcripts of theseverbal interactions showed that participants exposed to the intense emotionalmovie talked both about the eliciting circumstances and about their emotionalreactions This suggests that social sharing serves both informational andemotional goals

Alternative variables and residual effects Analyses of covariance indicatedthat controlling for interest vividness of mental images ease of verbalisationand the appraisals of certainty clarity and importance did not affect the strengthof the relationship between movie type and duration of social sharing Whenemotional intensity was held constant however the effect of movie type on theduration of social sharing was substantially weaker albeit still significantTogether these findings constitute evidence that the three excerpts differedspecifically with respect to emotional intensity

680 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

However further analyses found that the appraisal of novelty differedsignificantly across conditions and that using novelty as a covariate also reducedthe amount of variance in duration of social sharing explained by movie typeThis could be seen as showing that variations in novelty provide an alternativeexplanation for the observed variations in social sharing Our preferredinterpretation is that novelty and emotional intensity are closely related in thepresent context Both empirical and theoretical considerations support thisposition Thus the high correlation between emotional intensity and novelty (r =71) indicates a substantial overlap in the constructs captured by these variablesIn addition several authors have argued that emotions tend to occur whenplanned behaviour is interrupted (Mandler 1975 Miller Galanter amp Pribram1960 Simon 1967) or more specifically when the evaluation of the likelysuccess of a plan changes (Carver amp Scheier 1990 Oatley amp Johnson-Laird1987) According to Oatley and Johnson-Laird (1987 p 35) the function ofemotion is precisely to organise a transition to a new phase of planned activityFrom these theoretical perspectives it is hardly surprising to find that emotionalintensity is related to novelty Although these two constructs are highly relatedto one another they are still differentiable It will be the task of future work toattempt to disentangle the respective contributions of emotional intensity andnovelty to the social sharing of emotion

About 50 of the variance of extent of social sharing was explained byvariations in emotional intensity Even if one bears in mind that factors such asmeasurement error preclude the possibility of accounting for 100 of thevariance in social sharing there remains a reasonable proportion of the varianceto be accounted for by other variables In this connection it is worth noting thatin two recent studies we observed that part of the variance of social sharingabout negative events (around 10) could be accounted for by individualdifferences in alexithymia (Luminet Zech Rime amp Wagner in press) Thesalient features of alexithymia include (1) difficulty identifying feelings anddistinguishing between feelings and the bodily sensations of emotional arousal(2) difficulty describing feelings to other people (3) a reduced capacity toengage in fantasy and other imaginal activities and (4) a stimulus-boundexternally oriented cognitive style (Nemiah Freyberger amp Sifneos 1976Taylor 1994 Taylor Bagby amp Parker 1991) Collectively these traits arethought to reflect a deficit in the cognitive processing and regulation ofemotional states (Taylor et al 1991 Taylor Bagby amp Parker 1997) Furtherstudies should probably explore more in details the impact of personalityvariables related to the (non)expression of emotion on the variations of socialsharing

Relation between emotional intensity and amount of social sharing Thepresent experiments intended to test the prediction that negative emotion elicitssocial sharing A positive linear relation was thus expected between the intensity

SOCIAL SHARING OF EMOTION 681

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

of the emotion and the extent of sharing (see Rime et al 1992) Althoughintensity is probably a central aspect of emotional experience it has generallybeen neglected as a specific object of investigation in research on emotion (forexceptions see Frijda Ortony Sonnemans amp Clore 1992 Sonnemans amp Frijda1994) Frijda et al (1992 p 64) proposed that emotional intensity refers to thetotal emotional impact of a given event of which the magnitude of variousobjective and subjective parameters are aspects or manifestations According tothese authors intensity more than anything else determines whether emotionslead to socially consequential behaviour (Frijda et al 1992 p 62) Thereasoning adopted at the beginning of the present investigation is thus clearly inline with the theoretical views expressed by Frijda et al (1992)

Although the reported results were highly consistent with the hypothesis thatemotion elicits social sharing they were also consistent in failing to show adifference between the moderate and the nonemotional conditions This opensthe possibility that the relation between emotional intensity and extent of socialsharing is not a linear one but rather a step-function (see Rime et al 1998)Unless a critical threshold of emotional intensity is exceeded social sharingwould not occur Recent data are supportive of the suggested nonlinear relationbetween emotional intensity and resulting social behaviours (Christophe ampRime 1997)

Future research should thus examine this notion of a critical level of emotionalintensity more closely Our understanding of the relation between emotionalintensity and amount of social sharing might benefit from adopting amathematical nonlinear approach For instance catastrophe theory allows oneto conceptualise both physical and psychological processes in which a dependentvariable does not change smoothly in response to incremental changes inindependent variables (Stewart amp Peregoy 1983) Psychological processeshowever are still usually described and explained in terms of linear relationshipsor bell-shaped curves It is still possible however that with alternative sets ofstimuli one would have observed a monotonic effect of emotional intensity on theamount of social sharing Such an issue needs to be addressed in future studies

Motives underlying the cognitive and social consequences of emotionalevents Why do people engage in social sharing of emotion As suggested inthe introduction social sharing may be the consequence of two well-documented effects of emotion Emotion has been shown to induce bothrumination about the emotional material (for reviews see Horowitz 19751992) and the seeking out of the company of others (Schachter 1959 forreviews see Cottrell amp Epley 1977 Shaver amp Klinnert 1982) Putting thesetwo together intense emotional materials are very likely to be discussed in theaffiliation situation Thus it may be that the motives underlying the socialsharing of emotion overlap quite extensively with those underlying emotion-elicited rumination and affiliation

682 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

As regards the development of rumination following emotional events twobroad classes of motives have been considered A first one is the cognitiveprocessing of emotional information It is assumed that intense emotionchallenges a personrsquos mental model of the world People are thereby compelledto assess the meaning or implication of the event (cf Tait amp Silver 1989) toreconstruct basic assumptions about the self and the world (cf Janoff-Bulman1992) and to assimilate event-related information and accommodate existingschemata (cf Horowitz 1992) A second type of motive for rumination in thepresent situation consists of reassessing goals and goal-related informationAccording to this view intrusive thoughts result from perceived discrepanciesbetween the individualrsquos actual state and the goals heshe is trying to attain(Martin amp Tesser 1989 1996) Thoughts aimed at restoring the motivationalprocess therefore develop

Turning to the propensity to affiliate when exposed to threat or stress twotypes of motives can account for the lsquolsquostress and affiliationrsquorsquo effect in the presentstudies (Schachter 1959) First direct anxiety reduction may be achievedthrough the presence of others such that participants affiliate in the hope thatsimply being with others would make them feel less anxious or afraid Second itis possible that participants affiliate in order to gain cognitive clarity about thestimulus situation by obtaining information from others who know about thenature and dangerousness of the stressor

To summarise the motives for social sharing are likely to include one ormore of the following cognitive processing of emotional information cognitiveprocessing of goal-related information anxiety reduction through socialpresence and achieving cognitive clarity about emotion stimuli or responsesAs noted by Rime et al (1998) further motives can be identified on the basis ofrecent empirical findings First it has been observed that following highlyemotional situations people favour nonverbal behaviours such as huggingkissing or touching rather than verbalisation (Christopher amp Rime 1997) Thussharing of a markedly intense emotional experience may be instrumental indecreasing interpersonal distance and in strengthening social bonds Secondsocial sharing may contribute to the spreading of social knowledge aboutemotions in the social network of the person who shares (Rime 1995 Rime ampChristophe 1997) Social sharing provides the listener with useful knowledge onwhat may happen in life on how one can react to it and on what theconsequences may be It is a task for future research to explore the extent towhich these motives underlie the social sharing of emotion and whether thesemotives apply equally to different kinds of emotional events

Generalisability of the present findings The issue of the generalisability ofthe present findings warrants some comment What was elicited by ourexperimental inductions was probably far from approaching the intensity ofemotional reactions provoked by strong real-life situations The movie excerpt

SOCIAL SHARING OF EMOTION 683

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

used in the intense condition lasted only about three minutes It was displayed ina laboratory situation which is certainly not an optimal milieu to createspontaneous emotional responses Further ethical considerations clearly preventthe use of any strong emotional manipulations in the context of a trueexperiment Therefore further investigation will entail examination of the natureand content of social sharing of those who have been exposed naturally tointense emotional events

Generalisability of the present findings would be strengthened if the resultscould be replicated using different emotional stimuli thereby ruling out thepossibility that the results are due to some idiosyncratic features of the stimulusmaterials used here However consistency between the present findings andresults from previous studies do provide some evidence of generalisability withrespect to type of stimulus

First previous studies have been consistent in showing the pervasiveness ofsocial sharing across negative emotional states These studies encompassed (1)retrospective procedures in which respondents recalled emotional episodes oftheir past (for a review see Rime et al 1992) (2) follow-up procedures inwhich participants were contacted immediately after events eliciting differentnegative emotions (eg car accident bereavement first experience of donatingblood anxiety-arousing exam being present at the dissection of a human corpseperforming a dissection for a review see Rime et al 1998) and (3) diarystudies in which people report on the most emotional episode of the day (Rimeet al 1994) All these studies provided strong evidence that negative emotionalepisodes are shared regardless of the type of basic emotion involved with thesingle exception of shame

Second incidental observations from recent research on affiliation alsosuggested that experiencing negative emotion results in the social sharing ofemotion In an experiment conducted by Kulik and colleagues (Kulik et al1994) participants arrived at the laboratory in pairs They were placed indifferent rooms and were then each led to expect that heshe would later engagein a cold pressor task In one condition (similarity) both participants of the dyadanticipated the stress in another condition (dissimilarity) one participant of thedyad actually experienced the stress whereas the other only anticipated it Thedyads were then asked to wait together for a few minutes The time spent bydyads in talking about the stress was used as the dependent variable Resultsindicated that in the dissimilarity condition the mean duration of talks related tothe stressor was significantly longer than it was in the similarity condition Thissuggests that actually experiencing stress leads to more verbalisation thanmerely anticipating it A content analysis also revealed that those exposed to astress talked about both the situation and their feelings related to the stressfulsituation The experiment conducted by Kulik et al (1994) differed from thepresent ones on two major aspects First the type of emotional elicitor (coldpressure task vs movie) was different Second Kulik and his colleagues were

684 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

specifically interested by affiliation which involves to focus more on theanticipation of the stress rather than on its direct experience Despite thesedifferences both showed that negative emotion leads to the verbalisation ofemotion

Thus in total data issued from a variety of emotional contexts consistentlysuggest that negative emotional events are followed by social and cognitiveconsequences independent of the nature of the emotional stimulus used

Third the possibility that the frequency of social sharing differs for positiveand negative emotions has also been investigated In two separate studies Rimeet al (1991a) asked college students and older adults to describe four specificevents which were selected according to (a) their potential for eliciting emotion(b) their precise location in time and (c) their short duration The events wereclassified a priori as having a positive or a negative valence The proportions ofpositive and negative events which were shared were remarkably similar(positive events 89 in study 1 95 in Study 2 negative events 89 in study1 94 in study 2) Results reported by Christophe and DiGiacomo (1995)suggest that exposure to positive emotional episodes accelerates and enhancesthe social sharing effect

To conclude being exposed to an emotional episode has a marked impact onseeking social contacts with others and on the cognitive activity related to theeliciting situation Together these two processes lead to the social sharing ofemotion in which individuals communicate openly with one or more othersabout the circumstances of the emotion-eliciting event and about their ownfeelings and emotional reactions The three experiments reported here areconsistent in showing (a) that emotional intensity provides a good (if notcomplete) explanation for variations in extent of social sharing and (b) that therelation between emotional intensity and social sharing is not a linear function

Manuscript received 31 March 1999Revised manuscript received 16 February 2000

REFERENCES

Ainsworth MD Blehar MC Waters E amp Wall S (1978) Patterns of attachment Assessed inthe strange situation and at home Hillsdale NJ Erlbaum

Archer RL amp Berg JH (1978) Disclosure reciprocity and its limits A reactance analysis Journalof Experimental Social Psychology 14 527ndash540

Bales RF (1951) Interaction process analysis Cambridge MA Addison-WesleyBownes IT OrsquoGorman EC amp Sayers A (1991) Assault characteristics and posttraumatic stress

disorder in rape victims Acta Psychiatrica Scandinavica 88 27ndash30Bowlby J (1969) Attachment and loss Vol 1 Attachment New York Basic BooksBowlby J (1973) Attachment and loss Vol 2 Separation Anxiety and anger New York Basic

BooksCarver CS amp Scheier MF (1990) Origins and functions of positive and negative affect A

control-process view Psychological Review 97 19ndash35

SOCIAL SHARING OF EMOTION 685

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Christophe V amp Di Giacomo JP (1995) Contenu du partage social secondaire suite a un episodeemotionnel negatif ou positif [Content of the secondary social sharing following a positive or anegative emotional event] Unpublished manuscript University of Lille III France

Christophe V amp Rime B (1997) Exposure to the social sharing of emotion Emotional impactlistener responses and secondary social sharing European Journal of Social Psychology 27 37ndash54

Cohen J (1960) A coefficient of agreement for nominal scales Educational and PsychologicalMeasurement 20 37ndash46

Cottrell NB amp Epley SW (1977) Afiliation social comparison and socially mediated stressreduction In JM Suls amp RL Miller (Eds) Social comparison processes Theoretical andempirical perspectives (pp 43ndash68) Washington DC Hemisphere

Frijda N (1986) The emotions Cambridge UK Cambridge University PressFrijda N Ortony A Sonnemans J amp Clore GL (1992) The complexity of intensity Issues

concerning the structure of emotion intensity Review of Personality and Social Psychology 1360ndash89

Gump BB amp Kulik JA (1997) Stress affiliation and emotional contagion Journal ofPersonality and Social Psychology 72 305ndash319

Horowitz MJ (1975) Intrusive and repetitive thoughts after experimental stress Archives ofGeneral Psychiatry 32 1457ndash1463

Horowitz MJ (1992) Stress response syndromes (3rd ed) Northvale NJ Jason AronsonHorowitz MJ amp Redibord SP (1992) Memory emotion and response to trauma In SA

Christianson (Ed) The handbook of emotion and memory Research and theory (pp 343ndash357)Hillsdale NJ Erlbaum

Janoff-Bulman R (1992) Shattered assumptions Towards a new psychology of trauma New YorkFree Press

Kirkpatrick LA amp Shaver P (1988) Fear and affiliation reconsidered from a stress and copingperspective The importance of cognitive clarity and fear reduction Journal of Social andClinical Psychology 7 214ndash233

Kulik JA amp Mahler HIM (1987) The effects of preoperative roommate assignment onpreoperative anxiety and postoperative recovery from coronary by-pass surgery HealthPsychology 6 525ndash543

Kulik JA amp Mahler HIM (1989) Stress and affiliation in a hospital setting Preoperativeroommate preferences Personality and Social Psychology Bulletin 15 183ndash193

Kulik JA Mahler HIM amp Earnest A (1994) Social comparison and affiliation under threatGoing beyond the affiliate-choice paradigm Journal of Personality and Social Psychology 66301ndash309

Kulik JA Moore PJ amp Mahler HIM (1993) Stress and affiliation Hospital roommates effectson preoperative anxiety and social interaction Health Psychology 12 118ndash124

Lepore SJ (1997) Expressive writing moderates the relation between intrusive thoughts anddepressive symptoms Journal of Personality and Social Psychology 73 1030ndash1037

Lepore SJ Silver RL Wortman CB amp Wayment HA (1996) Social constraints intrusivethoughts and depressive symptoms among bereaved mothers Journal of Personality and SocialPsychology 70 271ndash282

Leventhal H amp Scherer K (1987) The relationship of emotion to cognition A functional approachto a semantic controversy Cognition and Emotion 1 3ndash28

Luminet O Zech E Rime B amp Wagner HL (in press) Predicting cognitive and socialconsequence s of emotional episodes The contribution of emotional intensity the Five FactorModel and alexithymia Journal of Research in Personality

Mandler G (1975) Mind and emotion New York WileyMarshall GD amp Zimbardo PG (1979) Affective consequence s of inadequately explained

physiologica l arousal Journal of Personality and Social Psychology 37 970ndash988

686 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Martin LL amp Tesser A (1989) Toward a motivational and structural theory of ruminativethought In JS Uleman amp JA Bargh (Eds) Unintended thought (pp 306ndash326) New YorkGuilford Press

Martin LL amp Tesser A (1996) Some ruminative thoughts In RS Wyer Jr (Ed) Advances inSocial Cognition Vol 9 Ruminative thoughts (pp 1ndash48) Mahwah NJ Erlbaum

Maslach C (1979) Negative emotional biasing of unexplained arousal Journal of Personality andSocial Psychology 37 953ndash969

McCammon S Durham TW Allison E amp Williamson JE (1988) Emergency workersrsquocognitive appraisal and coping with traumatic events Journal of Traumatic Stress 1 353ndash371

Miller GA Galanter E amp Pribram K (1960) Plans and the structure of behavior New YorkHolt Rinehart amp Winston

Nemiah JC Freyberger H amp Sifneos PE (1976) Alexithymia A view of the psychosomaticprocess In OW Hill (Ed) Modern trends in psychosomatic medicine (pp 430ndash439) LondonButterworth

Oatley K amp Johnson-Laird PN (1987) Towards a cognitive theory of emotions Cognition andEmotion 1 29ndash50

Ortony A Clore GL amp Collins A (1988) The cognitive structure of emotions New YorkCambridge University Press

Pennebaker JW (1989) Confession inhibition and disease Advances in Experimental SocialPsychology 22 211ndash244

Philippot P (1993) Inducing and assessing differentiated emotion-feeling states in the laboratoryCognition and Emotion 7 171ndash193

Rime B (1989) Le partage social des emotions [The social sharing of emotion] In B Rime amp KScherer (Eds) Les emotions (pp 271ndash303) Neuchatel Delachaux et Niestle

Rime B (1995) The social sharing of emotional experience as a source for the social knowledge ofemotion In JA Russell JM Fernandez-Dols ASR Manstead amp JC Wellenkamp (Eds)Everyday conceptions of emotions An introduction to the psychology anthropology andlinguistics of emotion (pp 475ndash489) Doordrecht Kluwer

Rime B amp Christophe V (1997) How individual emotional episodes feed collective memory InJW Pennebaker D Paez amp B Rime (Eds) Collective memories of political events Social andpsychologica l perspectives (pp 131ndash146) Mahwah NJ Erlbaum

Rime B Finkenauer C Luminet O Zech E amp Philippot P (1998) Social sharing of emotionNew evidence and new questions European Review of Social Psychology 9 145ndash189

Rime B Mesquita B Philippot P amp Boca S (1991a) Beyond the emotional event Six studieson the social sharing of emotion Cognition and Emotion 5 435ndash465

Rime B Noel MP amp Philippot P (1991b) Episode emotionnel reminiscences mentales etreminiscences sociales [Emotional episodes mental remembrances and social remembrances] Cahiers Internationaux de Psychologie Sociale 11 93ndash104

Rime B Philippot P Boca S amp Mesquita B (1992) Long-lasting cognitive and socialconsequence s of emotion Social sharing and rumination European Review of SocialPsychology 3 225ndash258

Rime B Philippot P Finkenauer C Legast S Moorkens P amp Tornqvist J (1994) Mentalrumination and social sharing in current life emotion Unpublished manuscript University ofLouvain Louvain-la-Neuve Belgium

Schachter S (1959) The psychology of affiliation Minneapolis MN University of Minnesota PressSchachter S (1964) The interaction of cognitive and physiologica l determinants of emotional state

Advances in Experimental Social Psychology 1 49ndash81Schachter S amp Singer JE (1962) cognitive social and physiologica l determinants of emotional

state Psychological Review 69 379ndash399Shaver P amp Klinnert M (1982) Schachterrsquos theories of affiliation and emotion Implications of

developmental research Review of Personality and Social Psychology 3 37ndash71

SOCIAL SHARING OF EMOTION 687

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL

Shortt JW amp Pennebaker JW (1992) Talking versus hearing about holocaust experiences Basicand Applied Social Psychology 13 165ndash179

Simon HA (1967) Motivational and emotional controls of cognition Psychological Review 7429ndash39

Smith CA (1993) Evaluations of whatrsquos at stake and what I can do In BC Long amp SE Kahn(Eds) Women work and coping A multidisciplinary approach to workplace stress (pp 238ndash265) Montreal McGill-Queenrsquos Press

Sonnemans J amp Frijda N (1994) The structure of subjective emotional intensity Cognition andEmotion 8 329ndash350

Sroufe LA amp Waters E (1977) Attachment as an organizational construct Child Development48 1184ndash1199

Stewart IN amp Peregoy PL (1983) Catastrophe theory modeling in psychology PsychologicalBulletin 94 336ndash362

Tait R amp Silver RC (1989) Coming to terms with major negative life events In JS Uleman ampJA Bargh (Eds) Unintended thoughts (pp 351ndash382) New York Guilford Press

Taylor GJ (1994) The alexithymia construct conceptualization validation and relationship withbasic dimensions of personality New Trends in Experimental and Clinical Psychiatry 10 61ndash74

Taylor GJ Bagby RM amp Parker JDA (1991) The alexithymia construct A potential paradigmfor psychosomati c medicine Psychosomatics 32 153ndash164

Taylor GJ Bagby RM amp Parker JDA (1997) Disorders of affect regulation Alexithymia inmedical and psychiatric illness Cambridge UK Cambridge University Press

Weisaeth L (1989) A study of behavioral responses to an industrial disaster Acta PsychiatricaScandinavica 80(Suppl 355) 13ndash24

Wilkinson CB (1983) Aftermath of a disaster The collapse of the Hyatt Regency Hotel skywalksAmerican Journal of Psychiatry 140 1134ndash1139

688 LUMINET ET AL