The “more is less” effect in equifinal structures: Alternative means reduce the intensity and...

10
The more is lesseffect in equinal structures: Alternative means reduce the intensity and quality of motivation Jocelyn J. Bélanger a, , Noa Schori-Eyal b , Gennaro Pica c , Arie W. Kruglanski b , Marc-André Lafrenière b a Université du Québec à Montréal, Canada b University of Maryland, United States c Sapienza University of Rome, Italy HIGHLIGHTS Increasing the number of means to a goal reduces their perceived instrumentality. Means distinctiveness (vs. similarity) reduces perceived means instrumentality. Increasing meansgoal association reduces the instrumentality of other means. Increasing the number of means impacts decision making and reduces the motivation intensity. Increasing the number of means to a goal reduces intrinsic motivation. abstract article info Article history: Received 19 July 2014 Revised 6 May 2015 Accepted 11 May 2015 Available online 13 May 2015 Keywords: Goal-systems Self-regulation Associative network Dilution Prior research in a multinality context has demonstrated a dilution effect whereby the number of goals connected to a means is inversely connected to the perceived instrumentality of each means (Zhang, Fishbach, & Kruglanski, 2007). In the present research, six studies tested whether the dilution effect would also occur in an equinality context where more than one means serves a single goal. The results corroborated this hypothesis. Presenting additional means (Study 1) and having participants self-generate multiple means (Study 2) led, as expected, to a reduction in perceived effectiveness of the means. The effect was stronger when the means within the set were seen as distinct rather than similar (Study 3). Increasing the strength of association between one means and the goal led to a reduction in the associative strength of the other means within the set and to a decrease of its perceived effectiveness (Study 4). The dilution effect was found to inuence means selection and the magnitude of means engagement (Study 5). Lastly, presenting additional means to attain a goal reduced the perceived effectiveness of the means, which in turn reduced participants' intrinsic motivation to pursue it (Study 6). Overall, the present work demonstrates that the presence of alternative means to a goal reduces the quality and intensity of engagement in a given means. Broad implications of these ndings are discussed. © 2015 Elsevier Inc. All rights reserved. There's more than one way to bake a cake, so the saying goes, and being able to pursue one's goals through a multitude of means is a com- mon experience. To attain tness one can jog, cycle, swim, or participate in an aerobics class, to name just a few options; to quench one's thirst one may imbibe a glass of water, an energy drink, or a fruit juice, and to invest one's funds one can select stocks, bonds, or buy precious metals, for example. Attainment means to the same goal are referred to as equinal (Heider, 1958; Kruglanski, Pierro, & Sheveland, 2011; Kruglanski et al., 2002) and they afford the possibility of choice and of substitution of one means for another in case of initial failure. Choice and substitution appear advantageous things to have. Indeed, prior re- search shows that the presence of numerous means to a goal (compared to only a few) increases one's subjective likelihood of goal attainment (Kruglanski et al., 2011). But having multiple means to a goal may have drawbacks as well. According to recent research, it may undermine each mean's impor- tance and reduce individuals' commitment to each means. Kruglanski et al. (2011, Study 3) demonstrated this phenomenon by asking state employees to generate two work related goals, and to generate either one or several means to each of those goals. Participants were then asked how irritated or upset they would be if they failed to attain the goal via a particular means. The results (replicated in the remaining studies of the Kruglanski et al.'s, 2011 research) revealed that partici- pants reported being signicantly less irritated/upset in the multiple Journal of Experimental Social Psychology 60 (2015) 93102 Corresponding author at: Département de Psychologie, C.P. 8888, Succursale Centreville, Montréal, Québec H3C 3P8, Canada. E-mail address: [email protected] (J.J. Bélanger). http://dx.doi.org/10.1016/j.jesp.2015.05.005 0022-1031/© 2015 Elsevier Inc. All rights reserved. Contents lists available at ScienceDirect Journal of Experimental Social Psychology journal homepage: www.elsevier.com/locate/jesp

Transcript of The “more is less” effect in equifinal structures: Alternative means reduce the intensity and...

Journal of Experimental Social Psychology 60 (2015) 93–102

Contents lists available at ScienceDirect

Journal of Experimental Social Psychology

j ourna l homepage: www.e lsev ie r .com/ locate / j esp

The “more is less” effect in equifinal structures: Alternativemeans reducethe intensity and quality of motivation

Jocelyn J. Bélanger a,⁎, Noa Schori-Eyal b, Gennaro Pica c, Arie W. Kruglanski b, Marc-André Lafrenière b

a Université du Québec à Montréal, Canadab University of Maryland, United Statesc Sapienza University of Rome, Italy

H I G H L I G H T S

• Increasing the number of means to a goal reduces their perceived instrumentality.• Means distinctiveness (vs. similarity) reduces perceived means instrumentality.• Increasing means–goal association reduces the instrumentality of other means.• Increasing the number of means impacts decision making and reduces the motivation intensity.• Increasing the number of means to a goal reduces intrinsic motivation.

⁎ Corresponding author at: Département de PsychCentreville, Montréal, Québec H3C 3P8, Canada.

E-mail address: [email protected] (J.J. Bélan

http://dx.doi.org/10.1016/j.jesp.2015.05.0050022-1031/© 2015 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 19 July 2014Revised 6 May 2015Accepted 11 May 2015Available online 13 May 2015

Keywords:Goal-systemsSelf-regulationAssociative networkDilution

Prior research in amultifinality context has demonstrated a dilution effectwhereby the number of goals connectedto a means is inversely connected to the perceived instrumentality of eachmeans (Zhang, Fishbach, & Kruglanski,2007). In the present research, six studies tested whether the dilution effect would also occur in an equifinalitycontext where more than one means serves a single goal. The results corroborated this hypothesis. Presentingadditional means (Study 1) and having participants self-generate multiple means (Study 2) led, as expected, toa reduction in perceived effectiveness of the means. The effect was stronger when the means within the setwere seen as distinct rather than similar (Study 3). Increasing the strength of association between one meansand the goal led to a reduction in the associative strength of the other means within the set and to a decrease ofits perceived effectiveness (Study 4). The dilution effect was found to influence means selection and themagnitude of means engagement (Study 5). Lastly, presenting additional means to attain a goal reduced theperceived effectiveness of the means, which in turn reduced participants' intrinsic motivation to pursue it(Study 6). Overall, the present work demonstrates that the presence of alternative means to a goal reduces thequality and intensity of engagement in a given means. Broad implications of these findings are discussed.

© 2015 Elsevier Inc. All rights reserved.

There's more than one way to bake a cake, so the saying goes, andbeing able to pursue one's goals through amultitude of means is a com-mon experience. To attain fitness one can jog, cycle, swim, or participatein an aerobics class, to name just a few options; to quench one's thirstone may imbibe a glass of water, an energy drink, or a fruit juice, andto invest one's funds one can select stocks, bonds, or buy preciousmetals, for example. Attainment means to the same goal are referredto as equifinal (Heider, 1958; Kruglanski, Pierro, & Sheveland, 2011;Kruglanski et al., 2002) and they afford the possibility of choice and ofsubstitution of one means for another in case of initial failure. Choice

ologie, C.P. 8888, Succursale

ger).

and substitution appear advantageous things to have. Indeed, prior re-search shows that the presence of numerousmeans to a goal (comparedto only a few) increases one's subjective likelihood of goal attainment(Kruglanski et al., 2011).

But having multiple means to a goal may have drawbacks as well.According to recent research, it may undermine each mean's impor-tance and reduce individuals' commitment to each means. Kruglanskiet al. (2011, Study 3) demonstrated this phenomenon by asking stateemployees to generate two work related goals, and to generate eitherone or several means to each of those goals. Participants were thenasked how irritated or upset they would be if they failed to attain thegoal via a particular means. The results (replicated in the remainingstudies of the Kruglanski et al.'s, 2011 research) revealed that partici-pants reported being significantly less irritated/upset in the multiple

94 J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

versus single means condition. Kruglanski et al. (2011) interpretedthese findings as attesting to a decreased dependence on (or commit-ment to) a given means in the presence of alternative means (cf.Rusbult, 1980; Thibaut & Kelley, 1959; van Lange, 2007). The foregoingexplanation implies that in the situation at hand, the perception of eachmeans' effectiveness remained constant, and the number of alternativemeans only reduced the extent to which each means was adjudged asessential or indispensible to goal attainment. But there might be anadditional drawback associated with multiple vs. single means contextthat has not been investigated thus far; a phenomenon that may havedeep ramifications for our understanding of motivation and goal-pursuit. Specifically, it is possible that the presence of multiple meansto the same goal dilutes the strength of association between any onemeans and the goal, which in turn is construed as a lessened instrumen-tality of the means with respect to goal attainment. This possibility isconsistent with research on the dilution model (Zhang, Fishbach, &Kruglanski, 2007) described next.

1. The dilution model

The dilution model is based on the concept of spreading activation(Anderson, 1983; Anderson & Bower, 1973) whereby the greater thenumber of associations attached to a mental construct the weakereach of those associations. Consistent with this logic, Zhang et al.(2007) proposed that when several goals are attached to a givenmeans, the activation by the means of any one of those goals would beimpeded, meaning that the goal would not come to mind as readilywhen the means is presented. In turn, the decrease in strength of themeans–goal association is subjectively interpreted as implying thatthemeans is less instrumental to the goal\\a phenomenon that occursbecause effective means are likely to be used often and thus to bestrongly associated with the goal (Shah & Kruglanski, 2003, see alsoMeyer & Schvaneveldt, 1971, 1976). Consequently then, meansattached to several goals (as opposed to a single goal) are perceived asless effective for goal-attainment.

Zhang et al. (2007) obtained support for these notions in a series ofstudies. Specifically, it was found that a means which serves two (vs.one) goals is seen as less effective for attaining any one of the goals:for example, learning that aerobic exercise could prevent heart diseaseas well as help maintain healthy bones resulted in it being perceivedas less effective for preventingheart disease. The effectwas found acrossdomains and independently of the content of the means or of the goals(Zhang et al., 2007). The dilution model postulates that the sum ofconnection strengths to a given construct is constant; that is why, infact, the greater the number of connections the lesser the strength ofeach connection. The constant sum assumption leads to a furtherprediction supported by Zhang et al. (2007) namely, that the strongerare some of the links (i.e., the quicker the activation of a given goal bythe means) the weaker are the others (i.e., the slower the activation ofalternative goals by that means).

Another premise of the dilution model is that the strength of themeans–goal association depends on the extent to which the goals com-monly associated with a given means subjectively differ from eachother. The more distinctive, or different from each other, those goalsare perceived to be, the stronger their tendency to undermine eachother's associationwith the sharedmeans. In contrast, themore similar,overlapping, ormutually facilitating the goals appear to be, the less theirjoint presence is likely to weaken their associative strength with theshared means. The underlying reason is that if the goals are similar,then one of them activates the other, and as a result, if one of thegoals is activated by the means, then the other would be activated aswell. This tendency counteracts the dilution effect and reduces itsimpact. This prediction too received support in the Zhang et al. (2007)research.

Though the studies by Zhang et al. (2007) provide consistentsupport for the dilution model, they suffer an inherent limitation

owing to the fact that in all instances the factor varied in their experi-ments was the number of goals attached to a given means. This allowsa possible alternative interpretation that the observed findings in theirresearch were at least partially mediated by a heuristic of the kind“Jack of many trades master of none,” suggesting that a meanspurporting to serve multiple goals serves none of them effectively. Ifthe underlying logic of the dilution model holds, however, we wouldexpect that it is the number of connections rather than the contents ofthe nodes to which the connections are made that determines the de-gree of dilution. In other words, the same dilution effect concerningthe reduced perceived instrumentality of means should obtain notonly with several goals attached to a single means, but also vice versa,with several means attached to a given goal. Should this prediction besupported we would have strong evidence that the dilution effectresides indeed in the number of cognitive connections betweenmotiva-tional constructs (goals andmeans) rather than in the specific directionof those connections (that is, whether they run from a single means tomultiple goals or from a single goal to multiple means). Moreover, if itwas found that the perceived effectiveness of ameans is inversely relat-ed to the number of availablemeans thiswould allow a new insight intothe found reduction of commitment to a givenmeans in the presence ofalternativemeans (Kruglanski et al., 2011). Specifically, itwould suggestthat such undermining of commitment might stem from perceivedeffectiveness reduction of each means where alternative means arepresent.

Cast in a broader context, this could mean that the availability ofmultiple alternatives does not merely reduce individuals' commitmentto each alternative (Kruglanski et al., 2011) via reduced dependence onit (Thibaut & Kelley, 1959), but instead it may do so via reducedperceived effectiveness of each of those alternatives to the goal it isexpected to serve. Thus, for instance, having multiple possibilities to agiven end, saymultiple job offers, or multiple possibilities for a romanticrelationship, may reduce the perceived allure of each of those alterna-tives and its ability to attain the desired end (say, of landing a satisfactoryjob, or having a fulfilling relationship). As a consequence, one mightultimately feel less satisfied with a chosen alternative the larger the setof alternatives from which that alternative was selected.

The present set of studies examined such implied effectivenessreduction in a set of equifinal means to a single goal; a vital point forthe validation of goal-systems theory. We expected to replicate theeffects obtained by Zhang et al. (2007) while reversing, relative totheir research, the directionality of the connection between goals andmeans. Specifically, we predicted that the number of equifinal meansconnected to a given goal will be negatively related to the perceivedinstrumentality of those means to the goal in question. We further ex-pected that strengthening the connection between one of the meansand the goal will increase its perceived instrumentality to the goal anddecrease the perceived instrumentality of the remaining means. Wepredicted that the dilution effect will be reduced if the equifinalmeans are perceived as similar to (versus distinct from) each other.Lastly, we examined the implications of the dilution effect fordecision-making, and its impact on the quality and intensity of engage-ment in the means.

2. The present research

We carried out six experimental studies to test the dilutionmodel inan equifinality paradigm, that is, with several differentmeans connectedto the same goal, rather thanwith several different goals being connect-ed to the same means as investigated previously (Zhang et al., 2007;Orehek et al., 2012). Our studies were conducted in a variety of contextsaswell as of content domains in order to determine the generality of thedilution effect. In Study 1, participants were presented with one or twomeans to a goal and they assessed the degree of effectiveness of the firstof these means. Study 2 employed a similar design, but it includedself-generated rather than experimenter-generated means. In Study 3,

1

2

3

4

5

6

7

per

ceiv

ed e

ffec

tive

nes

s

One-means

Two-means

95J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

wemanipulated the degree of perceived similarity between means andexamined the effect it has on perceived instrumentality of those meansto their intended goal. In Study 4, we experimentally strengthened thedegree of association of one of the means to the goal to observe a pre-dicted decline in effectiveness of alternative means. Novel implicationsof the dilution phenomenon were examined in Studies 5 and 6. InStudy 5, the relevance of dilution for individuals' decision-making andintensity of means engagement was documented. Lastly, Study 6examined the implications of the dilution effect for intrinsic motivation.The description of our specific studies follows.

3. Study 1: adding means to a goal

The purpose of Study 1 was to test the main prediction of the dilu-tion model in the equifinality context: specifically, we investigatedwhether the perceived instrumentality of a means diminishes as func-tion of the number of means assumed to serve that goal, presumablyby weakening the means–goal association.1 If the number of links iswhat accounts for the dilution effect then in the sameway that dilutionwas produced by adding goals to a given means it should be similarlyproduced by adding means to a given goal. This basic prediction wastested in the experiment described below.

4. Method

4.1. Participants

Two hundred and eleven participants (103 women, Mage = 34.17,SDage = 11.15) recruited on Amazon's Mechanical Turk (MTurk) tookpart in the study in exchange for a small monetary compensation.MTurk allows researchers to post questionnaires that are completedby users who participate in exchange for small contributions towardsan Amazon.com gift voucher. The platform records participants' IP ad-dress to prevent them from completing the same questionnaire morethan once. Researchers who have compared data from MechanicalTurk vis-à-vis data obtained in university laboratories or elsewhere onthe web (e.g., discussion forums) have concluded that (1) MTurk pro-vides more diverse and more representative samples (Buhrmester,Kwang, & Gosling, 2011), (2) “the quality of data provided by MTurkmet or exceeded the psychometric standards associated with publishedresearch” (Buhrmester et al., 2011, p. 5) and that (3)MTurk is a “reliablesource of experimental data” (Paolacci, Chandler, & Ipeirotis, 2010,p. 416, cf. Berinsky, Huber, & Lenz, 2012). Participants' gender showedno significant effects on our dependent variables here and in followingstudies, therefore, it is omitted in subsequent discussions.

4.2. Procedure and tools

Participantswere presentedwith the goal of experiencing popularitydefined as “Feeling that you are liked, respected, and have influence overothers rather than feeling like a person whose advice or opinions no-body is interested in” (Sheldon, Elliot, Kim, & Kasser, 2001). Participantswere randomly assigned to one of four experimental conditions. In thefirst two conditions, participants were assigned either to a one-meansor a two-means condition. In the one-means condition, participantswere presented with “hanging out with other people” as a means toachieve popularity. In the two-means condition participants were pre-sentedwith “hanging outwith other people” and “helping other people”as two alternative means to achieve popularity. Participants were thenasked to rate the extent to which they thought hanging out with otherpeople is an effectivemeans to experience popularity (1— very ineffective,

1 An early hint that the number of means is negatively related to association strengthbetween themeans and the goal is contained in a study by Fishbach et al. (2004) inwhichhaving two versus onemeans associatedwith the goal significantly decreased the transferof affect from the goal to the means.

to 7— very effective). In the remaining two experimental conditions, par-ticipants were also randomly assigned to a one or two-means condition.However, instead of rating the effectiveness of “hanging out with otherpeople”, participants rated the effectiveness of “helping other people”for the goal of popularity.

5. Results and discussion

A t-test was conducted with the first two experimental conditionsrating the effectiveness of “hanging out with other people” for thegoal of popularity. Consistent with our hypotheses, participants whowere presented with two means (M = 4.98, SD = 1.61) for experienc-ing popularity rated “hanging out with other people” as less effectiveto the goal of popularity than participants presented with the lattermeans alone (M=5.76, SD=1.61), t (100) = 2.91, p= .004, r2 = .07.

A t-test was also performed to compare the remaining two experi-mental conditions. Mirroring the findings above, participants in thetwo-means condition (M = 4.75, SD = 1.70) rated “helping otherpeople” (this time) as less effective than participants presented withthe latter means on its own (M = 5.42, SD = 1.16), t (99) = 2.31,p = . 02, r2 = .04. Fig. 1 displays the results.

These results provide initial support for the hypothesis that increasingthe number ofmeans to a goal decreases the perceived instrumentality ofa specificmeans to goal attainment. Especially noteworthy is the fact thatthe perceived reduction in means effectiveness in the two (vs. one)means condition was replicated with each of the presently employedmeans. Thus, whereas “helping other people” might be seen as alreadyimplying “hanging outwith other people,” reducing theperceived effec-tiveness of the latter means in the context of the former means (versusalone), the obverse, “hanging out with other people” does not seem toreciprocally imply “helping other people.” The fact that the perceivedeffectiveness of each of those means was dimmed by presence of theother isn't, therefore, readily explicable by a change of meaning(i.e., the redundancy) introduced by the additional means.

Nonetheless, our results are open to a possible alternative interpre-tation grounded in a conversational norm whereby if the speakerfinds it necessary to mention more than one means, neither meansalonemay be seen as very effective in achieving the goal. A yet differentalternative interpretation would be that the additional information inthe two-means condition was distracting which in turn, reducedthe perceived effectiveness of the focal means. Both concerns wereaddressed in our next study described below.

6. Study 2: self-generated means

The aim of Study 2 was to address the limitations of Study 1 byasking participants to self-generate means for goal-pursuit. Specifically,participants were presented with two goals (Goal 1: keeping in shape;

Hanging out with otherpeople

Helping others

Fig. 1. Perceived effectiveness of the first means presented as a function of the number ofmeans (Study 1).

1

2

3

4

5

6

7

One means Three means

per

ceiv

ed e

ffec

tive

nes

s

Fitness goal

Culture goal

Fig. 2. Perceived effectiveness of self-generated means as a function of the number ofmeans (Study 2).

96 J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

Goal 2: enhancing one's cultural knowledge) andwere instructed towritedown one means for achieving Goal 1 and three means for achievingGoal 2, or the opposite (i.e., three means for Goal 1 and one means forGoal 2). The purpose of this experimental design was (1) to controlfor the amount of information being processed in all conditions (i.e., 4means) and (2) to allow participants to generate effective means ontheir own which would rule out the possibility that the experimenterinfluenced participants' perceived means instrumentality. In keepingwith Study 1, we expected that for both goals, generating additionalmeans would decrease the perceived instrumentality of the focalmeans (i.e., the first generated means) to the goal as predicted by thedilution model.

7. Method

7.1. Participants

Eighty undergraduates (60 women, Mage = 25.09, SDage = 3.30)participated in the study on a voluntary basis.

7.2. Tools and procedure

Participants were presented with the goals of keeping in shape (Goal1) and enhancing one's cultural knowledge (Goal 2), and were asked towrite down one means to achieve Goal 1 (e.g., “doing sports”) andthree means to achieve Goal 2 (e.g., “traveling”, “reading”, “going tothe museum”), or vice-versa, counterbalancing the presentation orderand the type of goal presented in the two conditions. Participantswere then asked to rate the extent to which the means they generatedwere effective to achieve their goals (1 — very ineffective, to 7 — veryeffective).

8. Results and discussion

To test our hypothesis we performed a repeated measure ANOVAwith the ratings of means instrumentality as a within-subjects factorand the experimental conditions (1 means for keeping fit and 3 meansfor cultural knowledge vs. 3means for keeping fit and 1means for culturalknowledge conditions) as a between-subjects factor. Results indicatedthat the within-subjects factor was not significant, F (1, 78) = .12,p = .72. Similarly, the between factor was not significant, F (1, 78) =.19, p= .66. However, as expected, the results indicated a significant in-teraction, F (1, 78) = 17.17, p b .001, pη2 = .18.

Simple effects revealed that participants who generated one means(M=6.53, SD= .75) for the goal of keeping fit perceived it as more ef-fective than participantswho generated 3means (M=6.03, SD=1.02)for the same goal, F (1, 78) = 6.19, p= .01, pη2 = .07. Similarly, simpleeffects revealed that participants who generated onemeans (M=6.50,SD= .59) for the goal of enhancing their cultural knowledge perceivedit as more effective than participants who generated 3 means (M =6.13, SD= .72) for the same goal, F (1, 78) = 6.38, p = .01, pη2 = .08.Fig. 2 displays the results.

Results of Study 2 thus provide further support for the dilutionmodel. Just like the experimenter-provided means in Study 1, the self-generated means, familiar to participants, also displayed a decrease inperceived effectiveness when their greater number was associatedwith a goal. The different domains in which the effect occurred furtherattests to the generality of the dilution process, hence augmentingconfidence in its validity. By asking participants to generate the samenumber of items in both conditions, we also ruled out the alternativeexplanation that it is merely the number of informational itemsthat creates the observed effect due to a distractionmechanism. Finally,by requiring participants to generate their own means, we reduced theplausibility of a conversational interpretation of the obtained effect thatrests on the suggestion of the means by an outside interlocutor.

9. Study 3: means distinctiveness

According to the dilution hypothesis, the degree of dilution of themeans–goal association is related to uniqueness of the association,and should be greater if the additional means are dissimilar from theoriginal focal means. When two means are similar (vs. dissimilar),dilution should be weaker because the additional means could thenactivate the focal means, strengthening its connection to the goal, andcounteracting the dilution effect. Study 3 investigated this possibility.Specifically, we manipulated the perceived distinctiveness of twomeans to the same goal by asking participants to elaborate on whatmakes these means similar or different. We then measured theperceived efficacy of eachmeans to achieving the goal. By asking partic-ipants to elaborate on how the same two means are either similar ordifferent,we sought to demonstrate the effect of perceiveddistinctivenessindependently of means' content (the amount of actual overlap betweenthe mean). As noted earlier, we expected that a greater distinctionbetween the means would foster a greater dilution and thus lead tolower perceived effectiveness of the means in question.

10. Method

10.1. Participants

One hundred and twelve participants (57 women, Mage = 35.00,SDage = 12.45) recruited on MTurk took part in the study in exchangefor a small monetary compensation.

10.2. Procedure

Participants were asked to take part in an experiment on fitness.Specifically, they were provided with information about two means toachieving that goal: swimming and running. Participants in the similarcondition, were asked to describe in their own words how runningand swimming are similar to each other (theywrote, for example, “Run-ning and swimming are similar to each other in the sense that they bothwork on the cardiovascular aspect of fitness as well as maintaining themuscle and fat ratio needed for the body. They both require a lot ofwork from all parts of the body and they both help you keep fit”). Inthe distinct condition, participants were asked to describe how thetwo means differed from each other (they wrote, for example, that“Swimming is a full-body exercise, running is not. Swimming is lowimpact, while running has a high impact on joints. Running takesplace on land, while swimming is inwater”). At the end of the question-naire, participants rated the effectiveness of running and swimming forthe goal of fitness (0 — not at all and 100 — extremely).

97J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

11. Results and discussion

We performed a multivariate analysis of variance (MANOVA) forcondition (similar vs. distinct) on participants' perceived instrumentalityof running and swimming. The MANOVA yielded a significant effect forcondition, F (1,110) = 13.61, p b .001, pη2 = .11, indicating that partic-ipants in the similar condition (M = 88.40, SD = 11.37) rated runningas more instrumental for the goal of fitness than participants in thedistinct condition (M = 77.89, SD = 17.82). Similarly, participants inthe similar condition (M = 87.37 SD = 10.74) rated swimming asmore effective for the goal of fitness than participants in the distinctcondition (M = 82.29, SD = 14.83), F (1, 110) = 4.24, p b .05, pη2 =.03. Fig. 3 displays the results.

The results of support our hypothesis that increasing the distinctive-ness of the means causes the dilution effect to be more pronounced,resulting in decreased effectiveness. Moreover, these results indicatedthatwhen bothmeanswere seen as distinct, the dilution effect occurredon bothmeans simultaneously replicating Studies 1 and 2. This suggeststhat the link strengths of the differentmeans to the goal are interdepen-dent. Study 4 aimed to further investigate this phenomenon.

12. Study 4: increasing the means–goal association within theequifinal Set

As noted earlier, the dilution effect implies that the associationstrength of a means to a given goal construct is a fixed quantity distrib-uted over the number of links. This directly suggests that strengtheningthe link of a given means to the goal will proportionately weaken thelinks of alternative means. To test this hypothesis, we relied on animplicit technique to directly manipulate the strength of the means–goal association. We expected that strengthening such association willnot only increase the perceived effectiveness of the strengthenedmeans, but also decrease the perceived effectiveness of the remainingmeans.

Using an implicit technique to increase the means–goal associationwas also an opportunity for testing an alternative explanation to ourfindings. Specifically, it could be argued that our experimental instruc-tions prompted different cognitive elaboration strategies which influ-enced how individuals differentially construed the means presented ineach experiment. By attributing different meanings to the means, indi-viduals' perceived means instrumentality may have been affectedthrough a process different from that suggested by the dilution model.For example, in Studies 1 and 2, asking people to reflect on, or generate,multiple (vs. a single) means for a goal might have suggested to them anarrower (vs. broader), conception of what the means entail, thus re-ducing somehow the perceived effectiveness of the means. In Study 3,asking participants to explain why twomeans are different (vs. similar)may have indicated that the means have less (vs. more) commonmean-like features and thus are less effective for attaining the goal.

50

55

60

65

70

75

80

85

90

95

100

Running Swimming

per

ceiv

ed e

ffec

tive

nes

s

Similar

Distinct

Fig. 3. Perceived effectiveness as a function of goal distinctiveness (Study 3).

Three methodological caveats need to be observed to prevent themeans from being construed differently: (1) the number of means forattaining the goal must be held constant, (2) the informational contentwith regard to the means must also be kept constant, and (3) theinstructions provided to participants should not elicit different elabora-tion strategies vis-à-vis the means and the goal. To implement thesenotions, participants were presented with exactly the same number ofmeans for a given goal, while keeping the informational content of themeans constant. Finally, unconsciously increasing the strength of themeans–goal association via a subliminal priming task was meant toprevent a different conscious elaboration on the means and hence toprevent their differential construal.

13. Method

13.1. Participants

Seventy University of Maryland students (45 women,Mage = 20.45,SDage = 2.23) participated in the study in exchange for course credit.

13.2. Procedure

Participants were invited to take part in a study about comprehen-sion and attention. The studywas completed on a computer and includ-ed three sections: reading a short scientific text, performing a sequentialpriming task (described as an “attention task”), and rating of means'effectiveness.

Participants were first asked to carefully read a short text describingthe concept of health. The text mentioned two specific foods as meansfor achieving good health. Specifically, participants read the followingpassage:

“Health is a state of complete physical, mental, and social well-beingand not merely the absence of disease or infirmity (World HealthOrganization, 1946). Health is a positive concept emphasizing socialand personal resources, as well as physical capacities. Health is main-tained and improved not only through the advancement and applica-tion of health science, but also through the efforts and intelligentlifestyle choices of the individual and society. An effective means formaintaining good health is a balanced diet, such as the Mediterraneandiet, which includes foods like olives and fish.”

Participants then proceeded to complete an alleged “attention” task,which was designed to manipulate the strength of the association be-tween the goal (health) and one of the means (olives). Participantswere told that in this word judgment task, they would be presentedwith some letter strings and their task was to determine as quickly aspossible whether each letter stringwas aword or not. The task included100 trials, with an equal number of words and nonwords as target letterstrings. Each target letter stringwas pairedwith another word, present-ed subliminally before the appearance of the target letter string. In 12 ofthe 100 trials, the target letter string was the goal described in the text(health). In the strengthened-association condition, the word “olives”(i.e., the first means described in the text) was subliminally primedeach time before health was presented as the target letter string. Inthe control condition, the target word was paired with unrelatedwords so that on 12 trials, participants responded to the letter stringhealth following a subliminal presentation of irrelevant words (e.g.,chair). The number of trials using the goal as a means was equal inboth conditions. The remaining 88 trials were identical in both condi-tions and used irrelevant words or nonword letter strings as targets. Asimilar repetition of target letter strings was included among the fillertrials to further conceal the purpose of the task.

On each trial, a fixation point (the letter X) appeared at the center ofthe screen for 1000ms. Participants were asked to focus their attentionon this point. It was then replaced by a string of Xs, presented for 50msand serving as a forward mask. A prime word was then presented for abrief period of 34 ms, followed by a backward mask (a string of &s),

98 J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

which was replaced after 50 ms by the target word. Participants' taskwas to categorize the target letter strings as either words or nonwords,using the Z and “slash” keys, respectively. Each response was followedby a 700-ms pause before the next trial. Participants completed threepractice trials before beginning the main part of the task.

After completing the sequential priming task, that manipulated thestrength of association between the “olives” means and the goal ofhealth, participants engaged in the final part of the study, in whichthey answered questions about the text they had read earlier. Partici-pants assessed how effective was eachmeans for attaining the specifiedgoal (“maintaining good health”). The order of these questions wascounterbalanced, and they were embedded among five filler questions(e.g., “How relevant do you think ‘social well-being’ is to the definitionof health?”). All answers were recorded on a 9-point scale rangingfrom 1 (not at all) to 9 (extremely). After completing the survey, partic-ipants were debriefed and dismissed. No participant reported beingaware of the subliminal primes in the sequential priming task.

14. Results and discussion

We expected that strengthening the association between a meansand the goal would increase the perceived instrumentality of thatmeans to that goal, but decrease the perceived instrumentality of alter-native means to the same goal. Accordingly, a repeated measuresANOVA performed on the perceived instrumentality data yielded anAssociative Strength Goal interaction, F (1, 68) = 18.08, p b .01. Specif-ically, enhancing the means–goal (Olives-Health) association enhancedthe perceived instrumentality of that means for that goal (M = 7.22,SD = 1.55), compared with the baseline condition (M = 6.29, SD =1.64), t (68) = 2.48, p b .05, r2 = .07. Moreover, as shown in Fig. 4,enhancing the means–goal (Olives-Health) association impaired theperceived instrumentality of fish for the goal of health (M = 6.03,SD = 1.52), compared with the control condition (M = 6.71, SD =1.43), t (68) = −1.98, p = .05, r2 = .05.

The results provided further support for the dilution model anddemonstrate the role of means–goal associative strength. As the associ-ation between two nodes (goal-means A) goes up, the association of theother nodes (goal-means B) goes down, affecting their perceivedeffectiveness in opposite directions. This confirms the fixed nature ofassociation strength to a given goal that may be distributed over itsassociated means. Providing additional means to the goal thus dilute allmeans' associations to the goal, decreasing their perceived effectiveness.

Notably, results of Study 4 addressed the differential construal-hypothesis as an alternative explanation to the dilution model. Specifi-cally, Study 4 kept constant the informational content and the numberof means associated with the goal while using an implicit proceduretask, which akin to conditioning, did not require the presence ofconscious awareness (see Olson & Fazio, 2001) and thus was unlikelyto promote different cognitive elaborations and construal with regard

1

2

3

4

5

6

7

8

9

Olives Fish

per

ceiv

ed e

ffec

tive

nes

s

Increased association

Baseline association

Fig. 4. Perceived effectiveness of means as a function of means–goal association strength(Study 4).

to the presently provided means. An interesting corollary of Study 4 isthat individuals' means preference can be automatically construedwithout any objective correlations between the salient features of themeans and its effectiveness in the world.

15. Study 5

So far, we have accumulated evidence demonstrating that thedilution effect reduces the inferred instrumentality of the means withrespect to a specific goal.While thesefindings provide valuable insights,they bring to the fore the question of what could be the impact of dilu-tion on behavior? In that regard, research indicates that, by and large,individuals select and engage (e.g., by investing time, effort, andresources) in means perceived to be instrumental to their goal(Kruglanski et al., 2012; 2013b; Vroom, 1964). In line with this reason-ing, we expected that the dilution effect would reduce the perceived in-strumentality of the means, which in turn would influence whetherindividuals choose and actually engage in that means. In testing forthese predictions, we controlled for the subjective importance of thegoal at hand, given that prior research has indicated that importantgoals render any means to attain them as justifiable (i.e., “the end jus-tifies the means”, Köpetz, Faber, Fishbach, & Kruglanski, 2011).

16. Method

16.1. Participants

Eighty-eight participants (45 women,Mage = 35.68, SDage = 12.92)recruited on MTurk took part in the study in exchange for a smallmonetary compensation.

16.2. Procedure and tools

Participants were presented with a short blurb on reducing childrenhunger andworld poverty. Akin to Study 1, participantswere presentedwith either one means (“giving to charity”) or two means (“giving tocharity” and “signing a petition”) to achieve this goal. Participantsrated 1) the extent to which “giving to charity” is effective for reducingchildren hunger andworld poverty and 2) the extent to which this goalis important to them, on a 7-point Likert scale ranging from1 (Not at all)to 7 (Extremely). Then, participants read about a humanitarian organiza-tion (UNICEF) tackling this issue by using “donations to provide food tomillions of malnourished children”. After being informed that theycould allocate part of their remuneration ($0.00 – $0.40) to supportthis organization, participants decided either to give, or not, to this orga-nization. If they chose to donate, using a slider scale, participantsindicated how much they wanted to give. This paradigm has beenused in prior research (Bélanger, Caouette, Sharvit, & Dugas, 2014).

17. Results and discussion

Path analyses were conducted to investigate the mediating role ofinstrumentality between the experimental conditions (coded 1 forthe two-means condition and 0 for the one-means condition) andchoosing to donate to charity, controlling for the subjective importanceof reducing children hunger and world poverty. The model was testedwith Amos (Arbuckle, 2007) using maximum likelihood estimationprocedure. Four paths were specified: one path from the experimentalconditions to perceived means instrumentality, one path from instru-mentality to choosing to donate to charity, and two paths linking goalimportance to perceived instrumentality and choosing to donate tocharity. We display means, standard deviations, and correlations forall measures in Table 1. Results revealed that the hypothesized modelfit the data well (χ2 [df = 1, N = 88] = 1.18, p = .27, CFI = .99,IFI = .99, RMSEA = .04, AIC = 27.18).

Table 1Means, standard deviations, and correlations involving all variables from Study 5 (N=88).

Variable M SD 2 3 4 5

Conditionsa (1) 0.55 0.49 − .24⁎ .00 .00 − .03Instrumentality (2) 4.84 1.51 .35⁎⁎⁎ .24⁎ .08Accepting to donate (3) 0.47 0.50 .69⁎⁎⁎ .32⁎⁎⁎

Money donated (4) 7.72 11.95 .22⁎

Goal importance (5) 73.56 26.39

⁎ p b .05.⁎⁎⁎ p b .001.

a 0 = one-means condition; 1 = two-means condition.

99J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

As shown in Fig. 5, the experimental conditions were negativelyrelated to perceived instrumentality (β = − .24, p = .01), whereasgoal-importance was unrelated to it (β = .07, p = .49). Perceivedinstrumentality (β = .33, p b .001) and means importance (β = .29,p = .002) were both related to choosing to donate to charity.

Indirect effects were investigated to further test the mediating roleof instrumentality in the relation between the experimental conditionsand choosing to donate to charity. Consequently, bootstrapped confi-dence interval estimates of the indirect effect (see Preacher & Hayes,2008) were calculated to confirm the significance of mediation. In thepresent study, the 95% confidence intervals of the indirect effect wasobtained with 5000 bootstrap resamples (Preacher & Hayes, 2008).Results confirmed the mediating role of instrumentality in the relationbetween the experimental conditions and choosing to donate to charity(β = − .08; CI = − .16 to − .02).

An identical model was tested using our second dependent variable,that is, how much money participants actually gave to the charity.Results revealed that the hypothesized model fit the data well (χ2

[df = 1, N = 88] = 0.50, p = .48, CFI = 1.00, IFI = 1.00,RMSEA = .00, AIC = 26.50).

As shown in Fig. 6, the experimental conditions were negativelyrelated to perceived instrumentality (β = − .24, p = .01), whereasgoal importancewas unrelated to it (β=.07, p=.49). Perceived instru-mentality (β = .33, p b .001) and goal importance (β = .29, p = .002)were both related to the amount of money participants donated.

Indirect effects were examined using the same bootstrappingtechnique previously described. Results confirmed the mediating roleof instrumentality in the relation between the experimental conditionsand the amount of money participants donated (β=− .05; CI =− .12to − .008).

The results provide further support for the dilution model and dem-onstrate its influence on decision-making and behavior. Specifically,Study 5 found that providing additional means to a goal, reduces themeans' perceived instrumentality for attaining that goal. This reductionof instrumentality has two consequences. First, individuals are less like-ly to choose the diluted means. Second, the extent to which individualsengage in the means is also substantially reduced. The effects occurredabove and beyond the role of goal-commitment, which increasedindividual's willingness to select and engage in the means provided,which is consistent with prior findings (Köpetz et al., 2011).

ExperimentalConditionsa

PerceivInstru

Goal Importance

-.24**

.07-.03

Fig. 5. Results from path analysis (Study 5). Note. a0 =

18. Study 6

Study 6 aimed to demonstrate a novel implication of the dilutioneffect for a classic motivational phenomenon. Specifically, Study 6examined how the weakening of the means–goal association causedby the dilution effect influences the experience of intrinsic motivationwith respect to an activity. Intrinsic motivation refers to behaviorsderived from an internal desire to perform an activity in order to expe-rience pleasure and satisfaction afforded by the activity (Deci & Ryan,1985; Vallerand, 1997). Based on goal-systems theory, Shah andKruglanski (2000) postulated that the stronger the means–goal associ-ation, the more one should have the impression that the activity andthe goal are meshed together in an inseparable unit, hence thatperforming the activity amounts to accomplishing the goal. In otherwords, as the means–goal association increases so does the impressionof doing the activity for its own sake (i.e., greater intrinsic motivation).Support for these predictions has been reported by Shah and Kruglanski(2000), who found a positive correlation between the perceived instru-mentality of a means to a goal and proxy measures of intrinsic motiva-tion such as self-reported frequency of activity engagement andimportance. Although these findings provide important insights, thereliance on cross-sectional data (which prevents causal inferences)and the use of non-validated measures of intrinsic motivation do notafford sufficient evidence to substantiate the relation between means'perceived instrumentality and engaging in the means for its own sake.Consequently, Study 6 sought to conceptually replicate Shah andKruglanski's findings by experimentally manipulating the means–goalassociation strength using the previously shown dilution technique,while measuring intrinsic motivation using a validatedmeasure. Specif-ically, we hypothesized that addingmoremeans to a goal would reducethe means–goal association, which in turn would reduce one's intrinsicmotivation to engage in themeans. We tested for this mediation modelcontrolling for themeans' subjective importancewhich has been shownto be positively associated with intrinsically motivated actions (Deci &Ryan, 1985; Pelletier et al., 1995; Ryan & Deci, 2000).

19. Method

19.1. Participants

One hundred and four participants (50 women, Mage = 28.77,SDage = 7.47) recruited on MTurk took part in the study in exchangefor a small monetary compensation.

19.2. Procedure and tools

Participants were presented with the goal of relatedness defined ashaving “regular intimate contact with peoplewho care about you ratherthan feeling lonely and uncared for” (Sheldon et al., 2001). Akin to Study1, participants were presented with either one means (“hanging outwith other people”) or two means (“hanging with other people” and“helping others”) to achieve this goal. Participants rated the extent

ed Meansmentality

Accepting to Donate

.33***

.29**

one-means condition; 1 = two-means condition.

ExperimentalConditionsa

Perceived Means Instrumentality

Money Donated

Goal Importance

-.24** .22*

.20*.07-.03

Fig. 6. Results from path analysis (Study 5). Note. a0 = one-means condition; 1 = two-means condition.

100 J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

to which “hanging out with other people” is an effective means toexperience relatedness on a 7-point Likert scale ranging from 1 (Not atall) to 7 (Extremely). Then, participants' intrinsic motivation for“hanging out with other people” was measured with the IntrinsicMotivation subscale of the Situational Motivation Scale (SIMS) devel-oped by Guay, Vallerand, and Blanchard (2000). The intrinsic motiva-tion subscale of the SIMS is a 4-item scale and include items such as “Iwould hang out with other people because I think that this activity isinteresting” and “…because I think this activity is pleasant” (α = .86).Participants rated each statement on a 7-point scale ranging from 1(Not at all) to 7 (Extremely). Lastly, participants rated the extent towhich “hanging out with other people” is important on a 7-point itemranging from 1 (Not at all) to 7 (Extremely).

20. Results and discussion

Path analyses were conducted to investigate the mediating role ofinstrumentality between the experimental conditions (coded 1 for thetwo-means condition and 0 for the one-means condition) and intrinsicmotivation, controlling for the subjective importance of the means. Themodel was tested with Amos (Arbuckle, 2007) using maximum likeli-hood estimation procedure. Four paths were specified: one path fromthe experimental conditions to perceived means instrumentality, onepath from instrumentality to intrinsic motivation, and two paths linkingmeans importance to perceived instrumentality and intrinsic motiva-tion. We display means, standard deviations, and correlations for allmeasures in Table 2. Results revealed that the hypothesized model fitthe data well (χ2 [df = 1, N = 104] = 0.009, p = .92, CFI = 1.00,IFI = 1.00, RMSEA = .00, AIC = 18.00).

As shown in Fig. 7, all estimated paths were significant. The experi-mental conditions were negatively related to perceived instrumentality(β=− .16, p= .03), whereas means importance was positively relatedto it (β = .56, p = .001). Perceived instrumentality (β = .26, p b .01)and means importance (β = .44, p = .001) were both related tointrinsic motivation.

Indirect effects were investigated to further test the mediating roleof instrumentality in the relation of our experimental conditions tointrinsic motivation. Consequently, bootstrapped confidence intervalestimates of the indirect effect (see Preacher &Hayes, 2008)were calcu-lated to confirm the significance of mediations. In the present study, the95% confidence interval of the indirect effects was obtained with 5000

Table 2Means, standard deviations, and correlations involving all variables fromStudy 6 (N=104).

Variable M SD 2 3 4

Conditionsa (1) 0.45 0.50 − .19⁎ − .07 − .04Instrumentality (2) 5.27 1.31 .51⁎⁎⁎ .57⁎⁎⁎

Intrinsic motivation (3) 5.21 1.17 .59⁎⁎⁎

Means importance (4) 5.25 1.16

⁎ p b .05.⁎⁎⁎ p b .001.

a 0 = one-means condition; 1 = two-means condition.

bootstrap resamples (Preacher & Hayes, 2008). Results confirmed themediating role of instrumentality in the relation of the experimentalconditions to intrinsic motivation (β = − .04; CI = − .12 to − .00).

The hypothesized model was compared to an alternative model inwhich the experimental conditions predicted intrinsic motivation,which in turn predicted perceived means instrumentality, controllingfor means importance. Compared to the hypothesized model, resultsindicated that the alternative model (χ2 [df = 1, N = 104] = 3.90,p = .04, CFI = .96, IFI = .97, RMSEA = .16, AIC = 21.90) had a poorfit to the data.

The results provide further support for the dilution model anddemonstrate its implication for motivational phenomena. Specifically,Study 6 found support for the idea that when means to a goal areadded, the means–goal association becomes weaker as does themeshing of the means and the goal, which reduces one's intrinsicmotivation to engage in themeans. These results replicate those obtainedby Shah and Kruglanski (2000) with a stronger methodology whichinvolved experimentally manipulating means instrumentality andcontrolling for alternative explanations.

21. General discussion

In the present set of studies, we examined whether the dilutionmodel, proposed by Zhang et al. (2007) and tested in a multifinalitycontext, also pertains to the equifinality situation in which multiplemeans are associated with a single goal. Because the dilution effect isassumed to be mediated by the number of links between nodes in acognitive structure rather than by the nature or content of those nodes(e.g., whether they represent means or goals) we expected Zhanget al.'s (2007) multifinality findings to replicate within an equifinalgoal-means structure. Specifically,we expected that addingmeans serv-ing the same goal would lead to a decrease in perceived instrumentalityof eachmeanswithin the equifinal set. Reduction in perceived effective-ness was expected to depend also on the perceived distinctiveness ofthemeans connected to the goal. Based on a concept of fixed associativestrength, we predicted that the strengths of different links within thesame equifinal structure will be interdependent in that increasing thestrength of some of the links will foster a proportionate decrease instrength of the remaining links. The present work also examinedconcrete implications of the dilution phenomenon. We predicted thatreducing the instrumentality of a means to a given goal would reducethe likelihood of people choosing this means and engaging in it. Lastly,we anticipated that the dilution effect would reduce means' perceivedinstrumentality, which in turn would reduce individuals' intrinsicmotivation.

The six studies described above provide consistent support for ourhypotheses and for the dilution model more generally. In Studies 1and 2 we found that when more than one means was associated witha goal, the means were perceived as less effective vehicles for reachingthat goal. In Study 3, we found that thinking about how two meansare different (vs. similar) from each other decreased the perceivedeffectiveness of the first means presented. In Study 4, we directly

ExperimentalConditions

Perceived Means Instrumentality

Intrinsic Motivation

Means Importance

-.16* .26**

.44***.56***-.04

Fig. 7. Results from path analysis (Study 6). Note. a0 = one-means condition; 1 = two-means condition.

101J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

manipulated the means–goal association assumed to affect perceivedeffectiveness in equifinal states. We either enhanced the associativestrength between a means and a goal by repeatedly pairing the twostimuli in a sequential priming procedure, or left it intact. Participants inthe enhanced-association condition perceived the means whose relationto the goal was strengthened as more effective to achieving the goal ofhealth, and the remaining means as less effective. Novel implications ofthe dilution phenomenon were evinced in Studies 5 and 6. Study 5revealed that diluting the perceived instrumentality of ameans to a goal re-duced the likelihood of participants choosing that means and the extent towhich they pursued it (themagnitude of engagement). In Study 6, borrow-ing from the dilution procedure of Study 1, adding more means to reach agoal reduced the perceived instrumentality of themeans, which in turn re-duced individuals' intrinsic motivation (the quality of engagement).

Overall, these findings demonstrate that the dilution effect occurs inequifinal situations where several different means exist to a single goalin the same way as it does in multifinal situations studied by Zhanget al. (2007) where several different goals are attached to a singlemeans: the greater the number of means that are associated with agiven goal, the less is eachmeans perceived as effective.Whereas the re-duced perceived effectiveness of a single means attached to multiplegoals could be explained by the use of a heuristic (“Jack of many trades,master of none”) the reduced perceived effectiveness of a nonunique(vs. unique) means to a single goal could not. Thus, the fact that thedilution effect occurs in both equifinal andmultifinal situationsprovidesadditional support for premises of goal-systems theory (Kruglanskiet al., 2002, 2013b). The composition of the interconnections withinthe goal-system appropriately impacts the strength of those associa-tions, regardless of whether the added nodes in the configuration con-stitute goals or means. This supports the notion that the cognitivearchitecture of the motivational system has significant influence onthat system's properties, which in turn impacts relevant psychologicalexperiences and inferences. An example of such influencewas apparentin Study 5 where individuals refrained from choosing and investing re-sources in means diluted by the presence of alternatives. In Study 6, itwas the qualitative motivational experience that was impinge on bythe dilution effect when individuals' intrinsic motivation was reducedbased on the perceived means–goal associations.

Beyond supporting the dilution model and its generality across thecontents and directionality of connections, the present results offer im-portant novel insights into the context of equifinality where individualspossess at their disposal several different alternatives for getting to theirgoals. In prior analyses it has been noted that the presence of equifinalalternatives reduces the individual's dependence on each such alterna-tive (Thibaut & Kelley, 1959) as well as reducing the commitment toeach alternative (Kruglanski et al., 2011). The present analysis andfindings suggest that in addition the equifinality configuration maycommunicate the decreased effectiveness of each alternative tomediatethe end it is purported to serve. Thiswould suggest a decreased satisfac-tion with each alternative because of its diluted connection to the goal.In prior research Fishbach, Shah, and Kruglanski (2004) found that thepositive affect invested in the goal is transferred to the means assumed

to serve that goal in proportion to the strength of the connectionsbetween the goal and the means. Dilution reduces the strength ofthese connections hence it should proportionately reduce the extentof transfer. As a consequence one should feel less positive about themeans the greater the equifinality set of means.

These notions carry the counterintuitive implication that the morediverse possibilities one has for goal attainment, the less satisfied andhappy one might be, all else being equal. Paradoxically then, thesepositive and enviable attributes that engender multiple possibilitiesmay contribute under some conditions to lowering the individuals'states of satisfaction and life happiness. The scope of these implicationscovers an incommensurable variety of domains such as interpersonalrelationships, marketing, and terrorism, just to name a few. These arebriefly described in turn.

In the realm of interpersonal relationships, different people serve differ-ent functions in our lives, and in some respects serve as means to differentgoals (e.g., feeling loved; entertainment; physical pleasure; etc.). Researchby Fitzsimons and Shah (2008) found that when significant others are in-strumental to people's current goals they are cognitively more accessibleandperceivedmore favorably. The present research suggests that this effectcould be moderated by the principles of dilution herein described. To besure, romantic partners or friends would be evaluated positively whenthey serve people's goal, but less so when other individuals (e.g., friendsor a secret lover) fulfill this goal as well. Accordingly, the greater numbersof people serving the same function in one's life would result in each friendor acquaintance being perceived as less effective in satisfying one's goals,therefore less desirable, which could ultimately results in poorer interper-sonal relationships (e.g., less loyalty).

In the field of consumer research, Iyengar and Lepper (2000) havefound that when people were offered a large selection of productswithin a given product category, seen here as means to a given end(e.g., flavors of jams or different kinds of chocolate) they were lesssatisfied with their selections than those who offered a more limitedchoice. Schwartz (2009) described this phenomenon as “the paradoxof choice” and demonstrated it in different contexts. Nor is the effectlimited to relatively trivial choices, like snacks or detergents; it alsoappears to occur in regard to important life decisions, for instanceconcerning pension plans (Iyengar, Huberman, & Jiang, 2004). Examin-ing the issue of choice through lens of the dilution model may alsoaccount for the mixed results of research on the subject. In a recentmeta-analysis, Scheibehenne, Greifeneder, and Todd (2010) reviewthe choice overload effect described above. Their analysis of studiesconducted in the lab and in the field reveals an average effect sizevery close to zero, but with substantial variance between the studies.The mixed results are explained by a number of possible moderators,such as the assortment structure and decision strategies. However,factors that affect dilution (e.g., the degree of similarity betweenavailablemeans or choices)may also play a role in determiningwhetherchoice overload occurs. This may have interesting implications forbranding and marketing. Of note, launching a new product may be abalancing act between strategically differentiating the product andpreserving its efficacy.

102 J.J. Bélanger et al. / Journal of Experimental Social Psychology 60 (2015) 93–102

With regard to terrorism, one longstanding question scholarshave tried to elucidate is what motivates individual to support terroristactivities and engage in political violence? Recent social psychologicaltheorizing (see Kruglanski et al., 2013a, 2014) has put forward theidea that individuals' quest for personal significance plays an importantrole in the radicalization process.

According to the theory, individuals are generally motivated to feelsignificant (“to matter”, to be “someone”) and often turn to radicalgroups when: 1) their personal significance is lost (e.g., they feelhumiliated), 2) their significance is threatened, or 3) an opportunityfor considerable significance gain is foreseen (e.g., becoming a hero, amartyr). From this perspective, terrorism-justifying ideologies arealluring because they portray political violence as an effective meansto defend one's group and obtain significance. This is especially truewhen no alternative means are available to fulfill the significancequest. In line with this reasoning, the principles of dilution hereindescribed could be applied to deradicalization efforts conducted world-wide (e.g., Saudi Arabia, Singapore, Sri Lanka, Montreal) to reduceindividuals' support for violence. Offering individuals with alternativemeans to become significant (e.g., securing a job; getting involved ina peaceful political movement) could weaken the perceived instru-mentality of political violence, and thus the extent to which individuals1) select this means, 2) engage in it, and 3) derive pleasure from it(i.e., intrinsic motivation). Deradicalization programs that haveexplored this possibility, by providing vocational education as a meansto personal significance, have shown promising results with membersof the LTTE in Sri Lanka (Kruglanski et al., 2014), such as a significantdecline of support for armed struggle over time. However, more workis needed in order to understand if these effects are mediated by areduction of perceived instrumentality.

While the results of the present study set offer consistent support forthe dilution model, some limitations of this work should be noted andaddressed in future research. Themediating role ofmeans–goal associa-tive strength, although shown to affect perceived effectiveness, shouldbe examined as a mediator between number of means and perceivedinstrumentality in future studies. Examining how addition of meansaffects association strength (e.g., using reaction times) could provideadditional support for the model beyond extant findings.

In the present work, the addition of means sufficed to create a dilu-tion effect and to reduce means' perceived instrumentality (Studies 1and 2). The model predicts, however, that the relation betweennumerosity and perceived instrumentality ismonotonic, hence additionof further means to the equifinal set should further increase the effect:for example, the perceived effectiveness of each means in an equifinalset comprised of six means should be weaker than in set of threemeans. This hypothesis too should also be explored in future studies.

Finally, the dilution model raises a broad question worthy of pursu-ing in future research: is it the case that the more choices we have themore additional choiceswewant? If the addition of eachmeans reducesour satisfactionwith each option, thismay foster the desire for yet otheroptions, paradoxically decreasing satisfaction even further. The dilutionof means model implies that in some situations, such insatiability withoptions may indeed take place, and that it may grow with increase inone's set size of available options with which to pursue one's goals.

References

Anderson, J.R. (1983). The architecture of cognition. Cambridge, MA: Harvard UniversityPress.

Arbuckle, J. (2007). Amos 16.0 user's guide. Chicago, IL: Spss.Bélanger, J.J., Caouette, J., Sharvit, K., & Dugas, M. (2014). The psychology of martyrdom:

Making the ultimate sacrifice in the name of a cause. Journal of Personality andSocial Psychology, 107, 494.

Berinsky, A.J., Huber, G.A., & Lenz, G.S. (2012). Evaluating online labor markets forexperimental research: Amazon. com's Mechanical Turk. Political Analysis, 20, 351–368.

Buhrmester, M., Kwang, T., & Gosling, S.D. (2011). Amazon's Mechanical Turk a new sourceof inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6, 3–5.

Deci, E.L., & Ryan, R.M. (1985). Intrinsic motivation and self-determination in humanbehavior. New York: Plenum.

Fishbach, A., Shah, J.Y., & Kruglanski, A.W. (2004). Emotional transfer in goal systems.Journal of Experimental Social Psychology, 40, 723–738.

Fitzsimons, G.M., & Shah, J.Y. (2008). How goal instrumentality shapes relationshipevaluations. Journal of Personality and Social Psychology, 95, 319.

Guay, F., Vallerand, R.J., & Blanchard, C. (2000). On the assessment of situational intrinsicand extrinsic motivation: The Situational Motivation Scale (SIMS). Motivation andEmotion, 24, 175–213.

Heider, F. (1958). The psychology of interpersonal relations. New Jersey: John Wiley& Sons, Inc.

Iyengar, S.S., Huberman, G., & Jiang, W. (2004). How much choice is too much?Contributions to 401 (k) retirement plans. In O.S. Mitchell, & S.P. Utkus (Eds.),Pension design and structure: New lessons from behavioral finance (pp. 83–96).New York: Oxford University Press.

Iyengar, S.S., & Lepper, M.R. (2000). When choice is demotivating: Can one desire toomuch of a good thing? Journal of Personality and Social Psychology, 79, 995–1006.

Köpetz, C., Faber, T., Fishbach, A., & Kruglanski, A.W. (2011). The multifinality constraintseffect: How goal multiplicity narrows the means set to a focal end. Journal ofPersonality and Social Psychology, 100, 810–826.

Kruglanski, A.W., Bélanger, J.J., Chen, X., Köpetz, C., Pierro, A., & Mannetti, L. (2012).The energetics of motivated cognition: a force-field analysis. Psychological Review,119, 1.

Kruglanski, A.W., Bélanger, J.J., Gelfand, M., Gunaratna, R., Hettiarachchi, M., Reinares, F.,et al. (2013a). Terrorism: A (self) love story re-directing the significance quest canend violence. American Psychologist, 68(7), 559.

Kruglanski, A.W., Köpetz, C., Bélanger, J.J., Chun, W.Y., Orehek, E., & Fishbach, A. (2013b).Features of multifinality. Personality and Social Psychology Review, 17, 22–39.

Kruglanski, A.W., Gelfand,M.J., Bélanger, J.J., Sheveland, A., Hettiarachchi, M., & Gunaratna,R. (2014). The psychology of radicalization and deradicalization: How significancequest impacts violent extremism. Political Psychology, 35, 69–93.

Kruglanski, A.W., Pierro, A., & Sheveland, A. (2011). How many roads lead to Rome?Equifinality set‐size and commitment to goals and means. European Journal of SocialPsychology, 41, 344–352.

Kruglanski, A.W., Shah, Y.J., Fishbach, A., Friedman, R., Chun, W.Y., & Sleeth-Keppler, D.(2002). A theory of goal-systems. In M.P. Zanna (Ed.), Advances in experimental socialpsychology (pp. 331–376). San Diego, CA: Academic Press.

Meyer, D.E., & Schvaneveldt, R.W. (1971). Facilitation in recognizing pairs ofwords: Evidence ofa dependence between retrieval operations. Journal of Experimental Psychology, 90, 227.

Meyer, D.E., & Schvaneveldt, R.W. (1976). Meaning, memory structure, and mentalprocesses. Science, 192, 27–33.

Olson, M.A., & Fazio, R.H. (2001). Implicit attitude formation through classical conditioning.Psychological Science, 12, 413–417.

Orehek, E., Mauro, R., Kruglanski, A. W., & van der Bles, A. M. (2012). Prioritizing associa-tion strength versus value: The influence of self-regulatory modes on means evalua-tion in single goal and multigoal contexts. Journal of personality and social psychology,102, 22.

Paolacci, G., Chandler, J., & Ipeirotis, P.G. (2010). Running experiments on amazonmechanical turk. Judgment and Decision Making, 5, 411–419.

Pelletier, L.G., Fortier, M.S., Vallerand, R.J., Tuson, K.M., Briere, N.M., & Blais, M.R. (1995). Towarda newmeasure of intrinsicmotivation, extrinsicmotivation, and amotivation in sports: TheSport Motivation Scale (SMS). Journal of Sport & Exercise Psychology, 17, 35-35.

Preacher, K.J., & Hayes, A.F. (2008). Asymptotic and resampling strategies for assessingand comparing indirect effects in multiple mediator models. Behavior ResearchMethods, 40, 879–891.

Rusbult, C. E. (1980). Commitment and satisfaction in romantic associations:A test of theinvestment model. Journal of Experimental Social Psychology, 16, 172–186.

Ryan, R.M., & Deci, E.L. (2000). Self-determination theory and the facilitation of intrinsicmotivation, social development, and well-being. American Psychologist, 55, 68.

Scheibehenne, B., Greifeneder, R., & Todd, P.M. (2010). Can there ever be too many options?A meta‐analytic review of choice overload. Journal of Consumer Research, 37, 409–425.

Schwartz, B. (2009). The paradox of choice. New York: HarperCollins.Shah, J.Y., & Kruglanski, A.W. (2000). Aspects of goal networks. In M. Boekaerts, P.R.

Pintrich, & M. Zeidener (Eds.), Handbook of self-regulation (pp. 85–110). San Diego,CA: Academic Press.

Shah, J.Y., & Kruglanski, A.W. (2003). When opportunity knocks: Bottom-up priming ofgoals by means and its effects on self-regulation. Journal of Personality and SocialPsychology, 84, 1109–1122.

Sheldon, K.M., Elliot, A.J., Kim, Y., & Kasser, T. (2001). What is satisfying about satisfyingevents? Testing 10 candidate psychological needs. Journal of Personality and SocialPsychology, 80, 325–339.

Thibaut, J.W., & Kelley, H.H. (1959). The social psychology of groups. Oxford, England: Wiley.Vallerand, R.J. (1997). Toward a hierarchical model of intrinsic and extrinsic motivation.

In M.P. Zanna (Ed.), Advances in experimental social psychology. Vol. 29. (pp.271–360). San Diego, CA: Academic Press.

van Lange, P.A. (2007). The goal construct in social psychology. In A.W. Kruglanski, & E. T.Higgins (Eds.), Social psychology: Handbook of basicprinciples (pp. 490–515). NewYork: Guilford Press.

Vroom, V. H. (1964). Work and motivation. New York: Wiley.Zhang, Y., Fishbach, A., & Kruglanski, A.W. (2007). The dilution model: How additional

goals undermine the perceived instrumentality of a shared path. Journal ofPersonality and Social Psychology, 92, 389–401.

World Health Organization (1946). Preamble to the Constitution of the World HealthOrganization as adopted by the International Health Conference [Official Records ofthe World Health Organization, no. 2, p. 100] 1948 [cited 2012 May 4].