This article was downloaded by: [University of Groningen]On: 12 August 2013, At: 02:46Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: MortimerHouse, 37-41 Mortimer Street, London W1T 3JH, UK

MemoryPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/pmem20

Differential correlates of autobiographicalmemory specificity to affective and self-discrepant cuesIneke Wessel a , Ineke R. Postma b , Rafaële J.C. Huntjens a , Catherine Crane c ,Jorien Smets d , Gerda G. Zeeman b & Thorsten Barnhofer ea University of Groningen , Groningen , The Netherlandsb University Medical Center Groningen , Groningen , The Netherlandsc University of Oxford , Oxford , UKd University of Leuven , Leuven , Belgiume Institute of Psychiatry, King's College , London , UKPublished online: 26 Jul 2013.

To cite this article: Memory (2013): Differential correlates of autobiographical memory specificity to affective andself-discrepant cues, Memory, DOI: 10.1080/09658211.2013.811255

To link to this article: http://dx.doi.org/10.1080/09658211.2013.811255

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy, completeness, or suitabilityfor any purpose of the Content. Any opinions and views expressed in this publication are the opinionsand views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy ofthe Content should not be relied upon and should be independently verified with primary sources ofinformation. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands,costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial orsystematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distributionin any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions

Differential correlates of autobiographical memoryspecificity to affective and self-discrepant cues

Ineke Wessel1, Ineke R. Postma2, Rafaele J.C. Huntjens1, Catherine Crane3,Jorien Smets4, Gerda G. Zeeman2, and Thorsten Barnhofer5

1University of Groningen, Groningen, The Netherlands2University Medical Center Groningen, Groningen, The Netherlands3University of Oxford, Oxford, UK4University of Leuven, Leuven, Belgium5Institute of Psychiatry, King’s College, London, UK

According to the CaRFAX model (Williams et al., 2007), several processes may result in overgeneralautobiographical memory. The present study examined whether the type of cue used in theAutobiographical Memory Test (AMT) is important for illuminating relationships between autobio-graphical memory specificity and variables pertinent to the Functional Avoidance (FA) and Capture andRumination (CaR) aspects of the model. Sixty-one women varying in their experience of a potentiallytraumatic event and previous depression completed two versions of the AMT: one containing affectivecues and the other containing cues representing idiosyncratic self-discrepancies. Consistent with the FAhypothesis, avoidance of the potentially traumatic event was associated with fewer specific memories onthe affective, but not the self-discrepant AMT. Furthermore, in line with the CaR hypothesis,performance on the self-discrepant, but not the affective AMT was related to ruminative self-reflectionin women reporting previous depression, even after controlling for current depression and educationlevels. Together the results suggest that varying cue type may increase the sensitivity of the AMT,depending on the aspect of the CaRFAX model of overgeneral memory that is to be addressed.

Keywords: Autobiographical memory specificity; CaRFAX model; Avoidance; Self-discrepancies; Reflectiverumination; Centrality of Event Scale.

Overgeneral memory refers to the phenomenonof being relatively unable to come up withmemories of specific, personally experiencedevents (Williams et al., 2007). Specific eventsare tied to a particular place and time. Since thefirst demonstration of the phenomenon in suicideattempters (Williams & Broadbent, 1986), re-search on reduced autobiographical memoryspecificity (rAMS) has been accumulating rapidly.

In this literature, rAMS is typically assessed withthe Autobiographical Memory Test (AMT). TheAMT provides participants with cue words (e.g.,happy, angry) and participants have to respondwith specific memories. rAMS has been found tobe a characteristic of patients suffering fromdepression and posttraumatic psychopathology(Williams et al., 2007). Importantly, rAMS doesnot seem to be a mere epiphenomenon of

Address correspondence to: Ineke Wessel, Department of Clinical Psychology, University of Groningen, Grote Kruisstraat 2-1,

9712 TS Groningen, The Netherlands. E-mail: J.P.Wessel@rug.nl

We thank Janneke Carmelia, Judith van Rijn, Suzanne van der Heijden, Rolienke Stuursma, Esther Canrinus, and Rita Timmer

for their assistance in this study. Parts of this study were supported by an Innovational Research Incentive VIDI grant (452*03*329) of the Foundation for Behavioural and Educational Sciences of the Netherlands Organization for Scientific research (NWO)

awarded to IW.

Memory, 2013http://dx.doi.org/10.1080/09658211.2013.811255

# 2013 Taylor & Francis

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

psychopathology, but is predictive of its course(Sumner, Griffith, & Mineka, 2010) and may evenconstitute a premorbid vulnerability factor (e.g.,Bryant, Sutherland, & Guthrie, 2007; Hauer,Wessel, Engelhard, Peeters, & Dalgleish, 2009;van Minnen, Wessel, Verhaak, & Smeenk, 2005).

Explanations (Williams et al., 2007; see alsoSumner, 2012) for rAMS rest on assumptionsspecified in Conway’s general account of autobio-graphical memory (i.e., Self-Memory System the-ory; Conway & Pleydell Pearce, 2000). Conway’saccount conceptualises autobiographical knowl-edge as a hierarchical cognitive structure, withlayers running down from a highly abstract level ofself-knowledge (e.g., life-time periods, such as‘‘When I was a university student’’) throughgeneral event knowledge (e.g., ‘‘Holidaysabroad’’) to the level of specific episodic details(e.g., ‘‘That spectacular sunset at the beach in X’’).The experience of remembering (i.e., episodicrecall) involves a spreading of activation through-out all levels of this autobiographical knowledgebase. A further assumption is that a specificmemory may be retrieved by navigating throughthe autobiographical knowledge base in differentways. Direct retrieval occurs when a cue (e.g., astunning sunset) activates specific features at theepisodic detail level (white sand, palm trees, soundof waves breaking). Activation of these detailswould then spread upwards to the more abstractknowledge levels (0 That particular holidayabroad in X; 0 when I was a university student),thus providing a context for the event details,resulting in the emergence of a full-blown recol-lective experience. This cue-driven, bottom-upretrieval is relatively automatic and effortless. Bycontrast, generative retrieval is a deliberate mem-ory-search, which is effortful and draws on cogni-tive resources. It is this kind of retrieval that isthought to be instantiated by cues in the AMT(Williams et al., 2007). A word cue (e.g., happy)would activate a semantic associate (holiday),which would trigger a categoric descriptor at theintermediate level containing general eventknowledge (holidays abroad). The search wouldthen be completed by reconstructing a specificinstance at the episodic detail level (that specta-cular sunset). According to this account, over-general memories arise when the top-down searchstops at the intermediate level (Conway & PleydellPearce, 2000; Williams et al., 2007).

In their CaRFAX model, Williams and collea-gues (2007) summarise three main factorsthat relate to rAMS. Two factors of this model,

functional avoidance (FA) and capture and rumi-nation (CaR), are relevant for the present pur-pose. The FA hypothesis has its origins in therelation of rAMS with trauma-related psycho-pathology. It assumes that recalling general events(the intermediate level of the hierarchy) elicits lessintense affect than recalling specific memories oftraumatic events, and thus that aborting the searchprocess at the intermediate level enables people tocognitively avoid strong negative affect accompa-nying a specific episodic memory. This avoidantretrieval style is thought to generalise, resulting inrAMS observable across a broad range of cues.Capture and rumination (CaR) refers to the ideathat a memory search may fail to produce aspecific memory when attention is captured bytask-irrelevant information at the intermediatelevel of the autobiographical knowledge hierarchy.Moreover, it is suggested that attentional capturewill be particularly likely when this irrelevantinformation reflects abstract self-descriptors. Theidea is that, in depressed patients, such abstractself-related information in the autobiographicalmemory knowledge base is often well-rehearsedand highly interconnected, and therefore easilyactivated.

Especially with regard to the notion of CaR,self-discrepancy theory (Higgins, 1987) is relevant.According to this theory, people’s cognitive repre-sentations of their selves incorporate self-guides.These self-guides are relatively stable cognitivestructures reflecting norms or standards that mo-tivate behaviour. A crucial feature of this theory isthat a perceived discrepancy between one’s presentcharacteristics and a particular type of self-guidewould be related to a specific type of negativeaffect, depending on the availability and accessi-bility of the discrepant representation. Indeed,there is evidence (see Higgins, 1987; Strauman,1992) that depressed mood is associated with aperceived gap between current characteristics andideal self-guides (‘‘how I ideally want to be’’), andanxiety is linked to a perceived inconsistencybetween the current self and ought self-guides(‘‘how I think I ought to be, or others think I oughtto be’’). As for autobiographical memory, Strau-man (1992, 1996) found that childhood memorieswere more efficiently retrieved in response to cuesthat mapped on to an individual’s self-guides thanself-descriptive control cues (including guides notrelevant to the particular participant). These re-sults suggest that abstract self-referent cues have asignificant capacity to activate mnemonic content,and potentially speak to the idea of ‘‘capture’’

2 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

during generative retrieval processes. However,this early work did not examine the specificity ofretrieved memories or explicitly differentiate be-tween cue words that represented discrepant self-guides from those that represented self-concordantself-guides. Later studies addressing these issuesindicated that cues reflecting self-discrepancies ordysfunctional schematic content are more likelyto produce retrieval of overgeneral memoriesin depressed or formerly depressed individuals(Barnhofer, Crane, Spinhoven, & Williams, 2007;Crane, Barnhofer, & Williams, 2007; Schoofs,Hermans, & Raes, 2012; Spinhoven, Bockting,Kremers, Schene, & Williams, 2007; Van den Broeck,Claes, Pieters, & Raes, 2012).

The CaRFAX model posits multicausality ofovergeneral memory, assuming that differentprocesses may interact or may be relevant todifferent degrees in different groups of indivi-duals. The preceding findings indicate that differ-ent cue words may be used to preferentially tapinto FA or CaR processes. In general, research onrAMS employs cues that are mainly affective(positive/negative; Williams et al., 2007). Suchcues would be particularly suitable to elicitavoidance of negative affect (FA) in traumatisedpeople. In contrast, for studies investigating CaRin (formerly) depressed individuals, self-discre-pant cues might be more appropriate. Varying cuetype should enhance the sensitivity of the AMTfor detecting the differential processes as speci-fied by the CARFAX model and associations withrelated variables (rather than show absolutedifferences in terms of presence or absence ofovergenerality).

The present study examined the use of differ-ent cues and their potential to tap into differentprocesses underlying rAMS in a sample of womenwho varied in their experience of a potentiallytraumatic event and previous depressive episodes.More specifically, part of the sample had a historyof (pre)eclampsia, which is a complication ofpregnancy that is potentially life-threatening toboth the mother and (unborn) child (Steegers,von Dadelszen, Duvekot, & Pijnenborg (2010).1

We administered two versions of the AMT: one

consisting of (typical) affective cues and the othercontaining idiosyncratic self-discrepant descrip-tors as cues. These self-discrepant descriptorspertained to actual-ideal and actual-feared self-discrepancies. Although initially anxiety wasfound to be associated with ideal-ought discre-pancies (e.g., Higgins, 1987; Strauman, 1992), laterwork introduced the concept of the feared self tocapture avoidant motives. Discrepancies betweenfeared and actual selves have been found to be abetter predictor of anxiety and guilt than discre-pancies between ought and actual selves, whichseem to become relevant only when avoidantmotives are low (Carver, Lawrence, & Scheier,1999). The current study therefore focused onideal and feared self-discrepancies assuming thatboth of these were likely to be important in ourgroup of participants. We explored correlationsbetween performance on either AMT and differ-ent variables pertinent to either the FA or CaRhypotheses.

As for variables conceptually related to FA, weincluded a measure of avoidance of memories ofthe target pregnancy and related affect. In addi-tion, we were interested in seeing how overgener-ality would relate to the centrality of this event tothe participants’ life story or identity. The central-ity of adverse events is associated with posttrau-matic stress responses (Berntsen & Rubin, 2006;Brown, Antonius, Kramer, Root, & Hirst, 2010)and prolonged grief (Boelen, 2009, 2012).Although there is some evidence that ratings ofcentrality and specificity are negatively related(Rubin, Dennis, & Beckham, 2011), to the best ofour knowledge no study to date has examinedwhether high centrality of one particular event islinked to a reduced number of specific memoriesin general. It has been suggested that a preoccu-pation with the memory of a single central eventinterferes with the accessibility of memories ofother autobiographical events, resulting in rAMS(Boelen, 2012). Furthermore, there is evidencethat the centrality of adverse events is related toavoidance of reminders (Berntsen & Rubin, 2006;Boelen, 2009). Apart from exploring a straightfor-ward relationship with event centrality, we weretherefore also interested in examining whetherevent centrality moderated the relation betweenavoidance and rAMS, i.e., whether this relationwould be more evident in those showing strongerpreoccupation with a central event.

Pertinent to CaR, we looked at the relation-ship of rAMS with trait rumination. In itself,rumination can be viewed as self-discrepant

1 Preeclampsia is a complication of pregnancy or the direct

postpartum period. It is characterized by new onset high blood

pressure and protein in the urine, and is potentially life

threatening as it may involve several organ systems, such as

kidneys, liver, or brain. When the brain is involved and a

woman develops seizures, the condition evolves into eclamp-

sia. The only way to reverse this condition is delivery of the

baby, often resulting in the birth of a premature infant.

CORRELATES OF MEMORY SPECIFICITY 3

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

processing (e.g., Crane et al., 2008). Trait rumina-tors frequently focus on discrepancies betweenactual and desired states and this would strengthena network of global self-descriptors in memory.Thus, trait ruminators would be especially proneto capture errors during retrieval, resulting inrAMS, when exposed to self-discrepant content.Although this hypothesis has not been testeddirectly before, there is some evidence addressingrelated issues that is broadly supportive. Forexample Papadakis, Prince, Jones, and Strauman(2006) explored the combined effect of self-discrepancy and ruminative coping on predictionof depressive symptoms in adolescent girls andfound that ideal self-discrepancy was only closelyassociated with depression in the context of highbrooding, the more detrimental form of rumina-tion. Similarly, Jones, Papadakis, Hogan, andStrauman, (2009) found that ideal-self discre-pancy was associated with greater mood distur-bance only in individuals with high ruminationand/or low reflection (i.e., the more adaptive formof rumination). Since negative mood is thought toresult from an individual’s ineffective attempts toresolve discrepancies, these findings suggest thatmaladaptive preoccupation with self-discrepan-cies is more pronounced in habitual ruminators.There is also direct evidence that such capture,indexed directly by rAMS, is more pronounced inthose high in rumination. For example, although arelationship between trait rumination and rAMShas rarely been found in nonclinical samples (seeRaes, Schoofs, Griffith, & Hermans, 2012; Smets,Griffith, Wessel, Walschaerts, & Raes, 2013),nonclinical participants with a history of depres-sion appear to be an exception. For example,Raes and colleagues (2012) showed that inremitted depressed patients, trait rumination wasrelated to memory specificity after a self-discre-pancy induction. Similarly, Crane, Barnhofer,Visser, Nightingale, and Williams (2007) demon-strated that previously depressed patients whoreported high levels of trait rumination showed anincrease in rAMS after an analytical ruminationinduction, whereas those who were low in traitrumination showed no change. Finally, Crane,Barnhofer, and Williams (2007) found that cuesreflecting self-discrepancies elicited overgeneralmemories in remitted depressed patients, but notin never-depressed persons. Taken together, itmay be that, in previously depressed patients, anelaborated network of general self-descriptorsneeds to be reactivated either by thinking aboutself-discrepancies, by rumination or by a close

match of AMT cues with those descriptors inmemory. Therefore, we also investigated whetherthe relationship between AMT performance andrumination differed between previously depressedand never-depressed participants. In addition, evi-dence is accumulating that rumination may havemaladaptive and more adaptive forms (Schoofs,Hermans, & Raes, 2010; Treynor, Gonzalez, &Nolen-Hoeksema, 2003; Watkins, 2008). Specifi-cally, brooding is thought to involve unproductivefocusing on one’s current undesired state, whereasreflection is thought to involve active cognitiveproblem-solving strategies intended to change thatstate. We explored the association of both types ofrumination with AMT performance.

In summary, the current study examined theidea that varying the type of memory cue used toassess rAMS may increase sensitivity of the AMTto different factors of the CaRFAX model ofovergeneral memory. Pertinent to FA, we antici-pated that performance on the standard affectiveAMT would be especially related to the sequelaeof an adverse experience, that is, the avoidance ofmemories and the centrality of the event. Germaneto CaR, we anticipated that performance on anAMT containing self-discrepant cues would beassociated with ruminative tendencies, especiallyin people who reported previous depression.

METHOD

Participants

The present sample included 64 women whoparticipated in a larger project examining thesequelae of the serious pregnancy complicationof (pre)eclampsia at the Department of Obstetricsand Gynecology of the University Medical CenterGroningen (UMCG). For selection and exclusioncriteria, see Postma, Wessel, Aarnoudse, andZeeman (2010). The project was approved by theUMCG Institutional Review Board, and all parti-cipants signed informed consent. Participation wasvoluntary and travel expenses were reimbursed.

Assessment

Variables referring to target pregnancy. TheImpact of Event Scale (IES; Horowitz, Wilner,& Alvarez, 1979) was used to measure thefrequency and avoidance of intrusions of a stress-ful event during the past seven days. Again, the

4 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

instructions explicitly referred to the target preg-nancy. The IES consists of 15 items that arescored on a four-point scale (0 �‘‘not at all’’,1 �‘‘rarely’’, 3 �‘‘sometimes’’, 5 �‘‘often’’; totalscore 0�75, Cronbach’s a�.84). The IES has twosubscales: intrusion (seven items, range 0�35,Cronbach’s a�.80) and avoidance (eight items,range 0�40, Cronbach’s a�.73).

The Centrality of Event Scale (CES; Berntsen &Rubin, 2006) measures the extent to which thememory of an event is a reference point for personalidentity. We used the short, seven-item scale andrewrote the instruction such that it explicitlyreferred to the target pregnancy. Items are scoredon a five-point scale (anchors 1 �‘‘totally dis-agree’’, 5 �‘‘totally agree’’). Total scores rangefrom 7 to 35, with higher scores indicating higherevent centrality. Internal consistency was high(Cronbach’s a�.91).

Rumination and self-discrepancies. The Rumi-native Response Scale (RRS) measures traitrumination. Participants are asked how theytypically respond when they feel depressed. Weused the Dutch translation (Raes, Hermans, &Eelen, 2003) of the original version (Nolen-Hoeksema & Morrow, 1991), that includes 22items scored on a four-point scale (0 �‘‘almostnever’’ to 3 �‘‘almost always’’; range total score0�66, Cronbach’s a�.87). Later versions of theRRS (Treynor et al., 2003; see also Schoofs et al.,2010) were slightly altered and incorporate aBrooding and a Reflection subscale. The originalversion of the RRS used in the present studycontains all five reflection items of the adaptedscale. Reflection refers to ‘‘cognitive problemsolving’’ (Treynor et al., 2003, p. 256) and thescale contains items such as ‘‘I analyse recentevents to try to understand why I am depressed’’.The original version of the RRS used herecontains three out five brooding items (e.g.,I think ‘‘Why do I always react this way?’’).Hence, in the current study, RRS�Reflectionscores range from 0 to 15 (Cronbach’s a�.65)and RRS�Brooding scores range from 0 to 9(Cronbach’s a�.61).

The Self-Description Questionnaire (adaptedfrom Carver et al., 1999; Crane, Barnhofer, &Williams, 2007) was used for identifying self-guides. Participants were instructed to writedown seven traits that they would like to ideallypossess (ideal self) and seven traits that theywould not like to have or would be afraid ofacquiring (feared self). Next, these traits were

scored on a seven-point scales with respect tosimilarity to their actual self (i.e., to what extentparticipants thought they possessed that trait rightnow; 1 �‘‘I am the opposite’’ to 7 �‘‘I am exactlylike’’).

Autobiographical memory. Two versions of theAutobiographical Memory Test (AMT) were usedto assess autobiographical memory specificity.The first version used a Dutch translation of thefive positive (happy, safe, interested, successful,surprised) and five negative words (sad, angry,clumsy, hurt, lonely) that are typically employedin overgeneral memory studies (e.g., Williams &Broadbent, 1986). Because these frequently usedwords predominantly pertain to affect, we refer tothis version as affective AMT. The second version(self-discrepant AMT) was constructed for eachparticipant separately and used idiosyncraticwords from their own self-description question-naire. A research assistant selected the five ideal-self traits with the lowest and the five feared-selftraits with the highest ratings of similarity withparticipants’ actual self.

The two AMT versions were administered incounterbalanced order. Participants were shownthe cue words one by one on separate cards andwere instructed to recall a specific memory inresponse to each of them within 30 s. In theinstruction, no explicit reference was made to thetarget pregnancy. Prior to testing, it was explainedto them that a specific memory refers to apersonally experienced event that happened ona specific place and time (i.e., within one day).In addition, participants were told that the eventin the memory should have happened more thanone week ago. There was an extensive practicephase using 10 neutral cues (e.g., grass, car)during which the experimenter and participantdiscussed why responses were specific or not. Theactual test was started after the participant hadresponded with three specific memories to neutralpractice cues in a row. A research assistant codedthe responses as specific, extended (i.e., eventsthat lasted longer than a day), categorical (i.e.,series of events), no memory (i.e., a semanticassociation), or as an omission. Specific memoriesthat referred to an event that happened less thanone week ago or to an event that was alreadymentioned, were counted as omissions. The num-ber of first responses to the cue words on eitherAMT that were specific memories was used as thedependent variable. A second rater (IW) codedthe memories of 25 participants (i.e., 250 mem-

CORRELATES OF MEMORY SPECIFICITY 5

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

ories for each AMT version). Interrater reliabilitywas excellent for coding memories as specific ornot (affective AMT: k�.87; self-discrepant AMT:k�.89).

In order to construct cue-word self-discrepancyratings, participants rated the extent to which the20 cue words from both AMT versions fit withthe person they were now (actual self), and withthe person they wanted to be (ideal self) on aseven-point scale (1 �‘‘totally’’ to 7 �‘‘not atall’’). Scores for negative (affective AMT) andfeared self (self-discrepant AMT) words werereversed. Next, for each cue word self-discre-pancy scores were calculated by subtracting theactual self-scores from the ideal self-scores. Thus,higher scores indicate higher self-discrepancy,that is, a larger distance between one’s presentstate and how one would like to be. Finally, self-discrepancy scores were summed to obtain totalscores on the affective and self-discrepant AMT,as well as negative, positive, feared, and idealsubscores.

Background and psychopathology question-naires. Information on background variables wasobtained with a questionnaire asking for age andeducational level. Highest level of completededucation was rated using a seven-point scaleranging from 1 �‘‘elementary school’’ to 7 �‘‘university’’ (Engelhard, van den Hout, & Schou-ten, 2006). In addition, participants completed twoscreening questions for remitted depression (‘‘Wasthere ever a time when . . . (1) . . . you felt depressedfor most of the day?; (2) . . . you were not interestedin activities that you usually enjoy?’’) based on theSCID-I (First, Spitzer, Gibbon, & Williams, 1997).These questions were answered with yes/no.

The Beck Depression Inventory II (BDI-II;Beck, Steer, Ball, & Ranieri, 1996; van der Does,2002) measures symptoms of depression in thepast two weeks. It consists of 21 items containingfour statements indicating increasing severity(range 0�3). Total scores range from 0 to 63(Cronbach’s a�.86), with higher scores indicatingmore severe depressive symptoms.

Procedure

Participants received a questionnaire package inthe post, containing an informed consent form,the self-description questionnaire and backgroundquestions. After completion, they returned thepackage in the mail. They were invited to visit the

university for an individual assessment session. Inaddition, they received another questionnairepackage (containing the BDI-II, PSS-SR, CES,IES, and RRS) by post to complete at home onthe evening prior to their visit. At the university,all participants were tested individually. First,participants engaged in unrelated neurocognitivetests (see Postma et al., 2010). Next, they com-pleted both the affective and self-discrepant AMT(in counterbalanced order) and provided AMTcue word ratings. Finally, participants were de-briefed.

Statistical analysis

The data from three participants were excludedfrom the analyses. Due to experimenter failure,one participant had missing data on crucial vari-ables (educational level and previous depressionscreening questions) in the regression analyses (seelater). Two participants emerged as multivariateoutliers from the regression analyses. Inspection ofthe experimenter’s notes revealed that, in one case,the participant had become upset and tearfulduring the AMT and produced no specific mem-ories. For the other multivariate outlier, inspec-tion of the background variables suggested thatshe was an atypical participant in this samplebecause she had had her target pregnancy at age16. In all, the final sample consisted of 61 women.

As for obstetric diagnosis, the final sampleconsisted of 24 formerly eclamptic, 19 formerlypreeclamptic, and 18 parous women who haduncomplicated pregnancies. Initial analyses re-vealed no significant differences between the ob-stetric diagnostic groups on any of the variables,highest F(2, 58) �2.33, p�.11, other than preg-nancy duration, birth weight, and impact of thedelivery at the time of the event. These differencesare to be expected given the nature of thecomplication of (pre)eclampsia. Controlling forpatient status in the regression analyses (see later)did not affect the outcome. Therefore, correlationalanalyses are reported for the group as a whole.

For simple correlations, we computed Spear-man’s rank correlations as some of the variablesshowed nonnormal distributions. For purpose ofHierarchical Multiple Regression (HMR) analyses,IES-avoidance scores were square root transformedbecause of skewness of the distribution. After thistransformation, inspection of residual plots sug-gested that assumptions were adequately met.

6 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

RESULTS

Avoidance and event centrality of targetdelivery

Mean age in the present sample was 39.4 years(range 24�64 years). Overall, time since the targetpregnancy was M�8.64 years (SD�4.83). Table1 provides the descriptives of the variables con-cerning the target delivery and AMS, as well astheir intercorrelations. Overall, participants gen-erated more specific memories to affective cuesthan to self-discrepant cues, t(60) �4.68, pB.001,d�0.60. Level of education correlated positivelywith the number of specific memories to bothaffective and self-discrepant cues, i.e., highereducational levels were associated with morespecific memories. IES avoidance was signifi-cantly and negatively related to memory specifi-city to affective cues, but not self-discrepant cues.The centrality of the target delivery, as reflectedby CES scores, did not display significant correla-tions with AM specificity.

In order to see whether event centrality wouldmoderate the association between IES avoidanceand memory specificity, separate hierarchical mul-tiple regression (HMR) analyses were conductedwith affective and self-discrepant AMT perfor-mance as dependent variables. First, all relevantpredictors were centred (West, Aiken, & Krull,1996). Next, the predictors were entered in threesteps in the HMR. The first step controlled forlevel of education. On the second step, IESavoidance and CES scores were entered aspredictors. The final step contained the productof the centred IES-avoidance and CES scores.

Table 2 summarises the HMRs with thenumber of specific memories to affective as wellas self-discrepant cues as dependent variables. Itcan be seen in this table that the control variableof educational level explained a borderline sig-nificant portion of the variance in performance onthe affective AMT. The next step added asignificant 11% to explained variance. This waspredominantly due to IES avoidance and not CESscores. The third step, containing the interactionbetween IES avoidance and CES, did not sig-nificantly contribute to explained variance. Forthe HMR of self-discrepant AMT performance,only the first step containing level of educationexplained a significant portion of the variance.

Overall, then, the HMR analyses for affectiveand self-discrepant cues differed in that avoidanceof reminders of the target pregnancy was relatedto fewer specific memories in response to affec-tive but not self-discrepant cues. Centrality of thetarget pregnancy to the life story did not moder-ate this effect nor did it contribute to variance inmemory specificity, irrespective of cue type.

Past depression and rumination

Next, we turn to correlates of memory specificitythat are related to depression and rumination.Crane, Barnhofer, and Williams (2007) reportedthat correlational patterns differed in previouslydepressed and never depressed participants.Therefore, we created groups based on thescreening questions for remitted depression. Par-ticipants who answered one of the two questionsaffirmatively were allocated to the previously

TABLE 1

Descriptives and Spearman’s rank correlations between number of specific memories and variables related to the target pregnancy

(N�61)

Education CES

IES

intrusion

IES

avoidance IES total

No. of specific

affective AMT

No. of specific

self-discrepant AMT

Mean (SD) 4.90 (1.43) 20.47 (7.6) 4.72 (5.29) 1.92 (3.44) 6.64 (7.73) 6.9 (2.03) 5.7 (2.35)

Median (range) 5 (1�7) 21 (7�35) 3 (0�23) 0 (0�15) 4 (0�30) 7 (2�10) 6 (1�9)

CES �.05

IES intrusion �.05 .48**

IES avoidance .10 .40** .67**

IES total �.02 .47** .97** .80**

No. of specific

affective AMT

.27* �.10 �.12 �.26* �.18

No. of specific self-

discrepant AMT

.40** �.10 �.08 �.18 �.12 .56** *

$pB.10, *pB.05, **pB.01. CES �Centrality of Events Scale; IES �Impact of Events Scale; AMT �Autobiographical

Memory Test.

CORRELATES OF MEMORY SPECIFICITY 7

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

depressed group (PD, n�28). The never de-pressed group (ND, n�33) consisted of womenwho answered no to both questions.

Table 3 summarises both background and vari-ables related to self-discrepant processing for boththe PD and ND groups. It can be seen in this tablethat the PD group had lower education levels andreported more depressive symptoms and totalrumination than the ND group. There were nodifferences between the groups with respect todiscrepancy ratings and memory specificity.

Tables 4 and 5 show Spearman’s rank correla-tions between depression, rumination, cue self-discrepancy ratings, and AMT performance for PDand ND participants separately. In the ND group,there were no significant correlations betweenmemory specificity and any of the other variables.In PD participants, there was a significant positivecorrelation between RRS reflection and memoryspecificity in response to self-discrepant cues. Thus,the higher their ruminative reflection scores, themore specific memories PD participants retrieved.For affective cues, this correlation showed a trendtowards significance. There was also a significantnegative correlation between BDI-II scores andmemory specificity following self-discrepant cues.Therefore, we subjected memory specificity to self-discrepant cues to a HMR analysis controlling forBDI-II and education level on the first step. Thisstep explained a borderline significant 19% of thevariance, F(2, 24) �2.85, p�.08, with BDI-IIscores as the only significant predictor, b � �.41,p�.04. The second step containing RRS reflectionscores added a significant 15% to explained var-

TABLE 2

Summary of hierarchical multiple regression analyses for pregnancy-related predictors of specific memories

Dependent variable Set of predictors DR2 df F-change B SE b t

Affective AMT 1. Education .06 1, 59 3.88$ 0.35 0.18 .25 1.97$

2. .11 2, 57 3.58*

SqIES-av �0.61 0.25 �.33 �2.51*

CES �0.01 0.04 �.02 �0.16

3. .03 1, 56 1.72 0.04

CES * SqIES-av �0.05 �.18 �1.31

Self-discrepant AMT 1. .12 1, 59 8.38**

Education 0.58 0.20 .35 2.89**

2. .05 2, 57 1.72

SqIES-av �0.45 0.28 �.21 �1.60

CES �0.01 0.04 �.03 �0.21

3. .00 1, 56 0.03

CES * SqIES-av 0.01 0.04 .02 0.17

$pB.10, *pB.05, **pB.01. AMT �Autobiographical Memory Test; sqIES-av �Impact of Event Scale, avoidance subscale,

square-root transformed; CES �Centrality of Event Scale.

TABLE 3

Comparisons between previously depressed (PD) and never

depressed (ND) participants of background variables,

self-discrepant processing, and autobiographical memory

specificity

PD

(n �28)

ND

(n �33) Test statistic

Age 38.71 (7.23) 40.0 (7.70) t(59) �0.67

Education 4.5 (1.55) 5.24 (1.25) t(59) �2.07*

BDI-II 9.89 (7.29) 5.17 (3.89) t(39.64)2� �3.08**

RRS

Reflection1 3.37 (2.54) 2.55 (2.22) t(58) � �1.34

Brooding1 2.74 (1.81) 2.09 (1.74) t(58) � �1.41

Total1 14.81 (8.00) 9.94 (7.40) t(58) � �2.45*

Discrepancy affective cues

Positive 4.86 (5.16) 4.34 (4.80) t(59) � �0.40

Negative 6.18 (4.34) 5.85 (4.96) t(59) � �0.27

Total 11.04 (8.43) 10.19 (7.05) t(59) �0.67

Discrepancy self-discrepant cues

Ideal self1 5.96 (5.53) 5.67 (3.57) t(58) � �0.25

Feared self1 5.93 (4.84) 5.67 (4.98) t(58) � �0.20

Total1 11.89 (9.14) 11.61 (6.41) t(58) �0.89

Specificity affective AMT

Positive 3.18 (1.36) 3.64 (1.06) t(59) �1.48

Negative 3.54 (1.20) 3.42 (1.15) t(59) � �0.37

Total 6.71 (2.32) 7.06 (1.77) t(59) �0.66

Specificity self-discrepant AMT

Ideal self 2.71 (1.46) 3.18 (1.26) t(59) �1.34

Feared self 2.68 (1.39) 2.79 (1.34) t(59) �0.31

Total 5.39 (2.42) 5.97 (2.28) t(59) �0.96

$pB.10, *pB.05, **pB.01. 1Data for one participant

missing. 2Adjusted DF due to unequal variances. PD �previously depressed; ND �never depressed; CES �Centrality of Events Scale; IES �Impact of Events Scale;

PSS-SR �Posttraumatic Symptoms Scale, Self-Report; BDI-

II �Beck Depression Inventory, 2nd edition; RRS �Ruminative Responses Scale; AMT �Autobiographical

Memory Test.

8 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

iance, F(1, 23) �5.02, p�.04. A similar analysiswith the number of specific memories to affectivecues as the dependent variable yielded no signifi-cant results, DR2�.11, F(2, 24) �1.41, p�.026, andDR2�.08, F(1, 23) �2.18, p�.15. Thus, it appears

that in participants reporting previous depressionmore ruminative reflection was associated withmore specific memories to self-discrepant cues,even after controlling for current depressive symp-toms and educational level. This relation was notfound for memory specificity in response to affec-tive AMT cues.

DISCUSSION

The main results of the present study can besummarised as follows. To begin with, in line with

expectations, rAMS elicited by affective AMT

cues was related to the avoidance of memories of

an adverse event. The association between avoid-

ance and memory specificity in response to self-

discrepant cues was less evident. However, with

regard to discrepancy-based processing, a differ-

ent picture emerged. That is, more specific

memories to self-discrepant cues were related to

more ruminative reflection in previously de-

pressed participants, even after controlling for

current depression. With regard to affective cues,

this pattern of results was less pronounced.By and large, the current findings are consistent

with the idea that varying cue type may increase

the sensitivity of the AMT, depending on what

aspect of the CaRFAX model of overgeneral

memory is to be tested. The finding that avoidance

of memories was predominantly associated with

TABLE 4

Spearman’s rank correlations between number of specific memories and variables related to self-discrepant processing in

previously depressed women (n�28)

BDI-

II

Discrepancy

affective cues

Discrepancy self-

guides1 RRS1RRS�

reflection1RRS�

brooding1No. of specific

affective AMT

Discrepancy affective

cues

.39*

Discrepancy self-guides1 .38* .47*

RRS1 .43* .07 .25

RRS�reflection1 .05 .04 .04 .68**

RRS�brooding1 .42* .07 .28 .72** .26

No. of specific affective

AMT

�.19 �.09 .10 .10 .34$ .08

No. of specific self-

discrepant AMT

�.40* .01 .04 .05 .39* �.04 .60**

$pB.10, *pB.05, **pB.01. 1Data for 1 participant missing. RRS �Ruminative Responses Scale; AMT �Autobiographical

Memory Test.

TABLE 5

Spearman’s rank correlations between number of specific memories and variables related to self-discrepant processing in never

depressed women (n�33)

BDI-II

Discrepancy

affective cues

Discrepancy self-

guides RRS

RRS�reflection

RRS�brooding

No. of specific

affective AMT

Discrepancy affective

cues

.36*

Discrepancy self-guides .33$ .48**

RRS .48** .35* .20

RRS�reflection .29$ .02 .00 .71**

RRS�brooding .36* .37* .27 .82** .52**

No. of specific affective

AMT

.11 �.09 .10 .05 .04 �.06

No. of specific self-

discrepant AMT

�.02 �.28 �.17 �.16 .08 �.28 .51**

$pB.10, *pB.05, **pB.01. RRS �Ruminative Responses Scale; AMT �Autobiographical Memory Test.

CORRELATES OF MEMORY SPECIFICITY 9

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

specificity of the affective AMT fits well with theempirical literature on functional avoidance (FA;see, for overviews, Sumner, 2012; Williams et al.,2007). Most of the research in this area was donewith these typical AMT cues (Williams & Broad-bent, 1986), which mainly are cues indexingaffective states. It is assumed that rAMS originatesfrom the avoidance of the negative affect asso-ciated with recalling adverse experience, general-ising in the longer run to the avoidance of affect ingeneral. To the extent that affective cues increasethe probability of triggering memories associatedwith affective states, employing this type of cuesshould maximise the sensitivity of tests of FA-related hypotheses.

Our results also provide further insights regard-ing the effects of self-discrepant cues (e.g., Crane,Barnhofer, & Williams, 2007; Spinhoven et al.,2007; Van den Broeck et al., 2012). In previousresearch, rAMS in response to self-discrepant cueshas been found to be tied to (past) depression(Crane, Barnhofer, & Williams, 2007; Van denBroeck et al., 2012). The present finding thatruminative reflection was associated with betterperformance on the self-discrepant AMT in for-merly depressed, but not in never-depressedparticipants is consistent with these results. Speci-fically, we found that the more formerly depressedparticipants habitually engage in purposeful self-analysis in response to sad mood, the more specificmemories they retrieved. This finding fits nicelywith accumulating evidence that reflection may bean adaptive form of rumination. For example,Jones et al. (2009) found that higher self-reflectiondecreased the link between ideal self-discrepan-cies and depressive symptoms in college students.Other evidence includes results from Treynor et al.(2003), who found that, although reflective rumi-nation was associated with more concurrent de-pression in community volunteers, it predictedfewer symptoms in the long run. Moreover,Arditte and Joormann (2011) found that in cur-rently depressed participants, reflection predictedrecovery from that episode six months later. Theidea is that reflection represents a concrete, instru-mental problem-solving approach rather than arepetition of abstract ‘‘why me?’’ type of questions(Nolen-Hoeksema, Wisco, & Lyubomirsky, 2008).Interestingly, better social problem-solving skillshave also been related to more autobiographicalmemory specificity (e.g., Goddard, Dritschel, &Burton, 1996). The idea is that retrieval of specificmemories helps the problem-solving process, e.g.,by facilitating the generation of various alternative

solutions. Based on the present finding it may bespeculated that the habitual use of a relativelyconcrete self-analytic problem-solving approachprovides ample opportunity for practising theretrieval of specific memories that are relevant toself-descriptors. In terms of CaR, an increasingability to exit the abstract level of self-descriptors orself-discrepant representations in the autobiogra-phical knowledge base would prevent furtherelaboration of those abstract self-labels and, in-stead, strengthen links between those self-descrip-tors and episodic detail. Of course, the presentfinding is correlational and we cannot be sure aboutthe causal direction of this association. It may bethat the use of reflective strategies in response tosad mood promotes the retrieval of specific mem-ories, that a tendency to be specific enhanceshabitual reflective rumination or that some otherprocess during the lifting of a depressive episode(e.g., increasing executive capacity to overridecapture of attention) might be responsible.

Other findings may also be relevant to studieson depression, rumination, and self-discrepancies.For example, there was a correlation betweendepression severity and memory specificity toself-discrepant cues in formerly depressed partici-pants, but not in never-depressed participants. Forstandard affective cues, this correlation was non-significant in both groups. In the literature, a directlink between rAMS and depression severity ap-pears to be an elusive phenomenon (see Williamset al., 2007). The present pattern of results suggeststhat the self-discrepant AMT may be more sensi-tive in detecting relationships with clinical vari-ables that are relevant to CaR, such as depressivesymptoms. This also fits with suggestions thatrAMS needs to be activated in nonclinical popula-tions, for example by cues that closely matchperceived self-discrepancies, or by the inductionof state rumination (Crane, Barnhofer, Visser,et al., 2007; Raes et al., 2012; Smets et al., 2013).Another finding that is of interest is that, overall,self-discrepant cues elicited fewer specific mem-ories than affective cues, irrespective of depressionhistory. This is reminiscent of findings that abstractcues elicit more overgenerality than concrete,highly imageable cues (Hauer, Wessel, Geraerts,Merckelbach, & Dalgleish, 2008; Williams et al.,1996). By definition, self-discrepancies would beabstract*at least to the rememberer. Thus, thisfinding indicates that regardless of (past) depres-sion, using self-discrepant cues might provide amore suitable method to elicit overgenerality in

10 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

nonclinical samples than affective cues (Schoofset al., 2012).

Some findings raise further questions. To beginwith, we did not find a correlation between rAMSand either trait rumination in general or broodingin particular. Although such a relationship wouldbe expected given the CaRFAX model andfindings in clinically depressed samples, the evi-dence in samples that are not clinically depressedis scarce (Raes et al., 2012). Thus, our findings areconsistent with the broader literature. However,some studies (Crane, Barnhofer, Visser, et al.,2007; Raes et al., 2012) demonstrated that highertrait rumination was linked to less memoryspecificity in formerly depressed participants aftera rumination induction. Hence, Raes and collea-gues (2012) argued that overgeneral memory mayneed to be reactivated by state rumination itsassociation with rAMS to become visible. Appar-ently, by themselves, our self-discrepant AMTcues were not powerful enough to induce such astate. Future studies might determine whetherusing a state-rumination induction on top of self-discrepant AMT cues reveals a stronger associa-tion between trait rumination and rAMS thanusing standard affective cues in combination withsuch an induction.

Furthermore, discrepancy ratings of the cuewords were not associated with memory specifi-city in an important way. At first glance, thisseems to be at odds with findings that cues withhigher discrepancies were negatively correlated tomemory specificity (Van den Broeck et al., 2012).However, this earlier study used the same cues forall participants, leaving room for cues that maynot have been self-relevant at all to a particularparticipant. By contrast, we selected people’smost self-discrepant cues. Thus, all cues wouldbe self-discrepant to our participants, renderingthe actual degree of self-discrepancy less impor-tant for memory retrieval. Self-discrepancy rat-ings were associated with depressive symptoms.

As for the variables pertinent to FA, thesuggestion that the centrality of the event to theparticipants’ life stories would be associated withmemory specificity was not confirmed. In addi-tion, we did not find an interaction between eventcentrality and avoidance of reminders of the targetpregnancy. We explored the possible involvementof event centrality because a preoccupation withthe memory of a single event might interfere withthe retrieval of other autobiographical memories(Boelen, 2012), and because of previous evidencefor relations between centrality of adverse events

and avoidance. From this latter perspective, ourresults are somewhat surprising given positivecorrelations between event centrality and intru-sion and avoidance subscales of the IES, whichreplicate previous findings in this literature (e.g.,Berntsen & Rubin, 2006; Boals, 2010; Boelen,2009). However, it could be argued that the birthof a child is a transitional event, which would bejudged as identity changing by almost everyone.Thus, high scores on the CES would come fromvarious sources, obscuring any impact on memoryspecificity. To our knowledge, this is the first studyto look into the relation between centrality of onetarget event and specificity of memories for otherevents, and it will be interesting to see whethersignificant associations emerge in other popula-tions and for unambiguously negative events.

The present study looked at the value of usingdifferent types of cues for exploring autobiogra-phical memory specificity in relation to adverseexperience (FA) and rumination (CaR) in arather separate fashion. However, these corre-lates would be relevant to different types ofpsychopathology (i.e., posttraumatic stress disor-der [PTSD] and depression) that have a highcomorbidity (Kessler, Sonnega, Bromet, Hughes,& Nelson, 1995). In addition, the conceptualframework adopted in the present paper wouldpredict that the mechanisms linked to one type ofvulnerability would apply to the other type aswell. For example, the idea that some sort ofhighly accessible actual�ideal self-discrepant re-presentations (Higgins, 1987) capture memorialprocesses might apply to both depression andposttraumatic psychopathology. Perhaps the con-tent of such actual�ideal self-discrepancies woulddiffer slightly for each disorder. It is plausible thatdepressed people would perceive a discrepancybetween an actual and some unattainable futureself, whereas PTSD might be characterised by aperceived discrepancy between a posttraumaticself and their lost undamaged self from the past(see also, for related ideas, Berntsen & Rubin,2006; Conway, 2005; Dalgleish, 2004). In addition,the present study employed actual�ideal andactual�feared self-discrepancies because wethought these would both apply to our particularsample. Since actual�feared self-discrepancies arepredominantly affectively negative words, incor-porating them along with (positive) actual�idealself-descriptors in the AMT would provide anappropriate parallel to the more commonly usedpositively and negatively valenced words in theaffective AMT. However, because the AMTs

CORRELATES OF MEMORY SPECIFICITY 11

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

contained only five words of each self-descriptor

type and thus would provide rather insensitive

measures, we refrained from analysing them

separately. Yet, using larger sets of self-discrepant

words as AMT cues may shed light on the relative

role of discrepancy type (i.e., actual�ideal, actual�feared, and actual�ought; e.g., Carver et al., 1999)

in capturing memory retrieval processes in de-

pression and PTSD, respectively. Future studies

might explore these issues further.There are some limitations to note. The pre-

sent study was exploratory and relied on a

relatively small sample. In addition, we used

suboptimal measures of brooding and past de-

pression. Nevertheless, it is encouraging that the

groups obtained with such a crude measure of

depression history displayed correlational pat-

terns that were in line with both the theoretical

and empirical literature. If these patterns hold in

larger and more carefully selected groups of

previously and never-depressed people is for

future studies to determine. Additionally women

were assessed on average several years after their

target pregnancy and it is possible that we would

have identified different or stronger patterns of

association, especially with the trauma-related

variables, if we had sampled closer to the time

of occurrence of the target pregnancy.In sum, the current findings show that reduced

AMS to affective cues was related to avoidance of

reminders of an adverse event, whereas more

memory specificity to self-discrepant cues was

related to reflective rumination in formerly de-

pressed participants. Especially the latter finding

raises important new questions as to the role of

memory specificity in reflective rumination and

ultimately, recovery from depression. In general,

the present results suggest that varying cue type

may increase the sensitivity of the AMT, depend-

ing on the aspect of the CaRFAX model of

overgeneral memory that is addressed.

Manuscript received 26 November 2012

Manuscript accepted 30 May 2013

First published online 25 July 2013

REFERENCES

Arditte, K. A., & Joormann, J. (2011). Emotionregulation in depression: Reflection predicts recov-ery from a major depressive episode. CognitiveTherapy and Research, 35, 536�543. doi:10.1007/s10608-011-9389-4

Barnhofer, T., Crane, C., Spinhoven, P., & Williams,J. M. G. (2007). Failures to retrieve specific memoriesin previously depressed individuals: Random errorsor content-related? Behaviour Research and Therapy,45, 1859�1869. doi:10.1016/j.brat.2007.02.006

Beck A. T., Steer, R. A., Ball, R., & Ranieri, W. F.(1996). Comparison of Beck depression Inventories�IA and �II in psychiatric outpatients. Journal ofPersonality Assessment, 67, 588�597. doi:10.1207/s15327752jpa6703_13

Berntsen, D., & Rubin, D. C. (2006). The centrality ofevent scale: A measure of integrating a trauma intoone’s identity and its relation to post-traumaticstress disorder symptoms. Behaviour Research andTherapy, 44, 219�231. doi:10.1016/j.brat.2005.01.009

Boals, A. (2010). Events that have become central toidentity: Gender differences in the centrality ofevents scale for positive and negative events. Ap-plied Cognitive Psychology, 24, 107�121. doi:10.1002/acp.1548

Boelen, P. A. (2009). The centrality of a loss and its rolein emotional problems among bereaved people.Behaviour Research and Therapy, 47, 616�622.doi:10.1016/j.brat.2009.03.009

Boelen, P. A. (2012). A prospective examination of theassociation between the centrality of a loss and post-loss psychopathology. Journal of Affective Disorders,137, 117�124. doi:10.1016/j.jad.2011.12.004

Brown, A. D., Antonius, D., Kramer, M., Root, J. C., &Hirst, W. (2010). Trauma centrality and PTSD inveterans returning from Iraq and Afghanistan. Jour-nal of Traumatic Stress, 23, 496�499. doi:10.1002/jts.20547

Bryant, R. A., Sutherland, K., & Guthrie, R. M. (2007).Impaired specific autobiographical memory as a riskfactor for posttraumatic stress after trauma. Journalof Abnormal Psychology, 116, 837�841 doi:10.1037/0021-843X.116.4.837

Carver, C. S., Lawrence, J. W., & Scheier, M. F. (1999).Self-discrepancies and affect: Incorporating the roleof feared selves. Personality and Social PsychologyBulletin, 25, 783�792. doi:10.1177/0146167299025007002

Conway, M. A. (2005). Memory and the self. Journal ofMemory and Language, 53, 594�628. doi:10.1016/j.jml.2005.08.005

Conway, M. A., & Pleydell Pearce, C. W. (2000). Theconstruction of autobiographical memories in theself-memory system. Psychological Review, 107,261�288. doi:10.1037/0033-295X.107.2.261

Crane, C., Barnhofer, T., Duggan, D. S., Hepburn, S.,Fennell, M. V., & Williams, J. M. G. (2008). Mind-fulness-based cognitive therapy and self-discrepancy inrecovered depressed patients with a history of depres-sion and suicidality. Cognitive Therapy and Research,32, 775�787. doi:10.1007/s10608-008-9193-y

Crane, C., Barnhofer, T., Visser, C., Nightingale, H., &Williams, J. M. (2007). The effects of analytical andexperiential rumination on autobiographical mem-ory specificity in individuals with a history of majordepression. Behaviour Research and Therapy, 45,3077�3087. doi:10.1016/j.brat.2007.05.009

Crane, C., Barnhofer, T., & Williams, J. M. G. (2007).Cue self-relevance affects autobiographical memory

12 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

specificity in individuals with a history of majordepression. Memory, 15, 312�323. doi:10.1080/09658210701256530

Dalgleish, T. (2004). Cognitive approaches to posttrau-matic stress disorder: The evolution of multirepre-sentational theorizing. Psychological Bulletin, 130,228�260 doi:10.1037/0033-2909.130.2.228

Engelhard, I. M., van den Hout, M. A., & Schouten E.G. W. (2006). Neuroticism and low educational levelpredict the risk of posttraumatic stress disorderin women after miscarriage or stillbirth. GeneralHospital Psychiatry, 28, 414�417. doi:10.1016/j.gen-hosppsych.2006.07.001

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B.W. (1997). Structured clinical interview for DSM-IVaxis I disorders. New York, NY: Biometrics Re-search, New York State Psychiatric Institute.

Goddard, L., Dritschel, B., & Burton, A. (1996). Role ofautobiographical memory in social problem solvingand depression. Journal of Abnormal Psychology,105, 609�616. doi:10.1037/0021-843X.105.4.609

Hauer, B. J. A., Wessel, I., Engelhard, I. M., Peeters, L. L.,& Dalgleish, T. (2009). Prepartum autobiographicalmemory specificity predicts post-traumatic stresssymptoms following complicated pregnancy. Memory,7, 544�556. doi:10.1080/09658210902953836

Hauer, B. J. A., Wessel, I., Geraerts, E., Merckelbach,H., & Dalgleish, T. (2008). Autobiographical mem-ory specificity after manipulating retrieval cues inadults reporting childhood sexual abuse. Journal ofAbnormal Psychology, 117, 444�453. doi:10.1037/0021-843X.117.2.444

Higgins, E. T. (1987). Self-discrepancy: A theoryrelating self and affect. Psychological Review, 94,319�340. doi:10.1037/0033-295X.94.3.319

Horowitz, M., Wilner, N., & Alvarez, W. 1979). Impactof event scale: A measure of subjective stress.Psychosomatic Medicine, 41, 209�218.

Jones, N. P., Papadakis, A. A., Hogan, C. M., &Strauman, T. J. (2009). Over and over again:Rumination, reflection, and promotion goal failureand their interactive effects on depressive symp-toms. Behaviour Research and Therapy, 47, 254�259.doi:10.1016/j.brat.2008.12.007

Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M.,& Nelson, C. B. (1995). Posttraumatic stress disorderin the National Comorbidity Survey. Archives ofGeneral Psychiatry, 52, 1048�1060. doi:10.1001/arch-psyc.1995.03950240066012

Nolen-Hoeksema, S., & Morrow, J. (1991). A prospectivestudy of depression and posttraumatic stress symptomsafter a natural disaster: The 1989 Loma Prieta earth-quake. Journal of Personality and Social Psychology,61, 115�121. doi:10.1037/0022-3514.61.1.115

Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S.(2008). Rethinking rumination. Perspectives on Psy-chological Science, 3, 400�424. doi:10.1111/j.1745-6924.2008.00088.x

Papadakis, A. A., Prince, R. P., Jones, N. P., & Strauman,T. J. (2006). Self-regulation, rumination, and vulner-ability to depression in adolescent girls. Developmentand Psychopathology, 18, 815�829. doi:10.1017/S0954579406060408

Postma, I. R., Wessel, I., Aarnoudse, J. G., & Zeeman,G. G. (2010). Neurocognitive functioning in womenwith a history of eclampsia: Executive functioningand sustained attention. American Journal of Peri-natology, 27, 685�690. doi:10.1055/s-0030-1253099

Raes, F., Hermans, D., & Eelen, P. (2003). The Dutchversion of the ruminative response scale (RRS-NL)and the rumination on sadness scale (RSS-NL) [DeNederlandstalige versie van de Ruminative Re-sponse Scale (RRS-NL) en de Ruminationon Sad-ness Scale (RSS-NL)]. Gedragstherapie, 36, 97�104.

Raes, F., Schoofs, H., Griffith, J. W., & Hermans, D.(2012). Rumination relates to reduced autobiogra-phical memory specificity in formerly depressedpatients following a self-discrepancy challenge: Thecase of autobiographical memory specificity reactiv-ity. Journal of Behavior Therapy and ExperimentalPsychiatry, 43, 1002�1007. doi:10.1016/j.jbtep.2012.03.003

Rubin, D. C., Dennis, M. F., & Beckham, J. C. (2011).Autobiographical memory for stressful events: Therole of autobiographical memory in posttraumaticstress disorder. Consciousness and Cognition: AnInternational Journal, 20, 840�856. doi:10.1016/j.con-cog.2011.03.015

Schoofs, H., Hermans, D., & Raes, F. (2010). Broodingand reflection as subtypes of rumination: Evidencefrom confirmatory factor analysis in nonclinical sam-ples using the Dutch ruminative response scale.Journal of Psychopathology and Behavioral Assess-ment, 32, 609�617. doi:10.1007/s10862-010-9182-9

Schoofs, H., Hermans, D., & Raes, F. (2012). Effect ofself-discrepancy on specificity of autobiographicalmemory retrieval. Memory, 20, 63�72. doi:10.1080/09658211.2011.637934

Smets, J., Griffith, J. W., Wessel, I., Walschaerts, D., &Raes, F. (2013). Depressive symptoms moderate theeffects of a self-discrepancy induction on overgen-eral autobiographical memory. Memory, doi:10.1080/09658211.2012.756039

Spinhoven, P., Bockting, C. L. H., Kremers, I. P.,Schene, A. H., & Williams, J. M. G. (2007). Theendorsement of dysfunctional attitudes is associatedwith an impaired retrieval of specific autobiographi-cal memories in response to matching cues. Memory,15, 324�338. doi:10.1080/09658210701256555

Steegers, E. A., von Dadelszen, P., Duvekot, J. J., &Pijnenborg, R. (2010). Preeclampsia. Lancet, 376,631�644.

Strauman, T. J. (1992). Self-guides, autobiographicalmemory, and anxiety and dysphoria: Toward acognitive model of vulnerability to emotional dis-tress. Journal of Abnormal Psychology, 101, 87�95.doi:10.1037/0021-843X.101.1.87

Strauman, T. J. (1996). Stability within the self: Alongitudinal study of the structural implications ofself-discrepancy theory. Journal of Personality andSocial Psychology, 71, 1142�1153. doi:10.1037/0022-3514.71.6.1142

Sumner, J. A. (2012). The mechanisms underlyingovergeneral autobiographical memory: An evalua-tive review of evidence for the CaR-FA-X model.

CORRELATES OF MEMORY SPECIFICITY 13

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3

Clinical Psychology Review, 32, 34�48. doi:10.1016/j.cpr.2011.10.003

Sumner, J. A., Griffith, J. W., & Mineka, S. (2010).Overgeneral autobiographical memory as a predic-tor of the course of depression: A meta-analysis.Behaviour Research and Therapy, 48, 614�625.doi:10.1016/j.brat.2010.03.013

Treynor, W., Gonzalez, R., & Nolen-Hoeksema, S.(2003). Rumination reconsidered: A psychometricanalysis. Cognitive Therapy and Research, 27, 247�259. doi:10.1023/A:1023910315561

Van den Broeck, K., Claes, L., Pieters, G., & Raes, F.(2012). Memory specificity in borderline personalitydisorder: Associations with depression and self-discrepancy. Journal of Behavior Therapy and Ex-perimental Psychiatry, 43, S51�S59. doi:10.1016/j.jbtep.2011.02.001

van der Does, A. J. W. (2002). BDI-II-NL handleiding[Manual for the BDI-II-NL]. Enschede, The Nether-lands: Swets Test Publishers.

van Minnen, A., Wessel, I., Verhaak, C., & Smeenk, J.(2005). The relationship between autobiographicalmemory specificity and depressed mood following astressful life event: A prospective study. British

Journal of Clinical Psychology, 44, 405�415.doi:10.1348/014466505X29648\

Watkins, E. R. (2008). Constructive and unconstructiverepetitive thought. Psychological Bulletin, 134, 163�206. doi:10.1037/0033-2909.134.2.163

West, S. G., Aiken, L. S., & Krull, J. L. (1996). Experi-mental personality designs: Analyzing categorical bycontinuous variable interactions. Journal of Personal-ity, 64, 1�48. doi:10.1111/j.1467-6494.1996.tb00813.x

Williams, J. M. G., Barnhofer, T., Crane, C., Herman,D., Raes, F., Watkins, E., & Dalgleish, T. 2007).Autobiographical memory specificity and emotionaldisorder. Psychological Bulletin, 133, 122�148.doi:10.1037/0033-2909.133.1.122

Williams, J. M. G., & Broadbent, K. (1986). Autobio-graphical memory in suicide attempters. Journal ofAbnormal Psychology, 95, 144�149. doi:10.1037/0021-843X.95.2.144

Williams, J. M. G., Ellis, N. C., Tyers, C., Healy, H.,Rose, G., & MacLeod, A. K. (1996). The specificityof autobiographical memory and imageability of thefuture. Memory and Cognition, 24, 116�125.doi:10.3758/BF03197278

14 WESSEL ET AL.

Dow

nloa

ded

by [

Uni

vers

ity o

f G

roni

ngen

] at

02:

46 1

2 A

ugus

t 201

3