the experience of regret

7/27/2019 The Experience of Regret

1/17

Psychological Review1995, Vol. 102, No. 2, 379-395Copyright 1995 by the American PsychologicalAssociation, Inc.0033-295X/95/S3.00

The Experience of Regret: What, When, and W hyThomas Giloviqh and Victoria Husted Medvec"Cornell University

This article reviews evidence indicating that there is a temporal pattern to the experience of regret.Actions, orerrors of commission, generate more regret in the short term; but inactions, or errors ofomission, produce more regret in the long run. The authors contend that this temporal pattern ismultiply determined, and present a framework to organize the divergent causal mechanismsthat areresponsible for it. In particular, this article documents the importance of psychological processesthat (a) decrease the pain of regrettable action over time, (b) bolster the pain of regrettable inactionover time, and(c) differentially affect the cognitive availability of these two types ofregrets. Both thefunctional and cultural origins of howpeople think about regret are discussed.

We have left undone those things which we ought to have done; an dwe have done those things which w e ought not to have done. American Book of Common PrayerMany people say they "regret nothing" or they "wouldn't doanything differently" if they could live over again. Althoughsuch statements are offered with sincerity, they can be difficultto accept at face value. Living a life with no mistakes and with-out any regrets isextraordinarily hardto accomplish. A lifetimeof making choices brings with it the knowledge tha t at leastsome actions were ill-considered, some failures to act unwise.For most of us, it also b rings with it the realization that some ofthese un for tun ate outcomes could have been avoided. To live, itseems, is to accu mula te at least some regrets.Given that few of us are likely to be strangers to the experi-ence of regret, one question tha t arises is how we can minimize

the depth and frequency of this unpleasant emotional experi-ence. How might we live our lives so as to keep the num ber andintensity of our future regrets close to some optimal level? Toanswer this question requires an understanding of what it is thatpeople regret most in their lives and why. These issues are thefocus of the present article.Previous Approaches to the Study of Regret

Philosophical ApproachesUntil recently, nearly all scholarly w ork on the subject of re-gret involved philosophical efforts to define the term (Bedford,Mu c h of the research reported here was supported by Grants

MH45531 from th e National Institute of Mental Health an dSBR9319558 from the National Science Foundation. W e would like tothank Todd Bickford, Kirsten Blau, Ann Charlton, Theresa Buckley,Deborah Fidler, Nina Hattiangadi, Allison Himelfarb, Elena Jeffries,Danielle Kaplan, Dorie Katzer, Talia Korenbrot, Jennifer Lowe, KenSavitsky, M arshall Schacht, Sarah Sirlin, Shane Steele, and Robin Win-itsky for collecting much of the data reported here. Daryl Bern an d Den-nis Regan provided helpful comments on an earlier version of thisarticle.Correspondence concerning this article should be addressed toThomas Gilovich, Department of Psychology, Uris Hall, Cornell Uni-versity, Ithaca, Ne w York 14853, or to Victoria Husted Medvec, John-son Graduate School of Management, Malott Hall, Cornell University,Ithaca, New York 14853.

1957; deSousa, 1987; Ham pshire, 1960; Landma n, 1987a;Rorty, 1980; Taylor, 1985; see also Landman, 1993). This workhas tried to elucidate the nature of the internal state that ac-companies or constitutes the experience of regret, specify thecircumstances in which regret can properly be said to be expe-rienced, and distinguish regret from related emotional statessuch as disappointment, guilt, remorse, and sadness.These philosophical efforts have met with m ixed success. Onthe plus side, there is general consensus that regret is an unus u-ally cognitively-laden or cognitively-determined emotion. AsHampshire (1960) states, "Do youregret that decision?' is aquestion that requires me to think and to think practically,about the decision, and not merely to inspect my feelings" (p.241) . Judgment is more central, in other words, to the experi-ence of regret than, say, the experience of jealousy or anger. Atthe same time, regret is more than a simple appraisal or judg-ment; it is typically loaded with feeling and therefore qualifiesas a true emotion. W hen on e contemplates a life of unfulfilledpromise, the destructive things said in anger, or the failure toheed the "cries for help" from someone who later committedsuicide, on e feels "... a particular sort of painful feeling, a pang,a stab, waves of stabs" (Ro rty, 1980, p. 496).There are other definitional issues, however, that are harderto resolve. For example, there appears to be little consensus asto w hether one can properly be said to regret something that onewould continue to do under similarcircumstances in the future.Ham pshire (196 0) asserts that

If a man continuesto make the kind ofdecisions that heclaims thathe regrets, he could not properly continue to describe his distressas regret. He would be compelled to describe it as a vague sense ofguilt or anxiety, or perhaps as an unhappy wish that he had greaterpowers, or that he wasplaced in othercircumstances (p. 2 4 1 ) .Others disagree (L andm an, 1993; Rorty, 1980; Taylor, 1985),and it is unclear how these rival claims can be decided. Hamp-shire's contentionthat we should differentiate between "re-grettable" acts a person would an d would no t repeatcertainlyhas some appeal. At the same time, it is easy to generate exam-ples in which someone could rightfully be said to regret an ac-tion that is repeated m any times. Fo r example, if Richard Nixonwere the type to experience regret, he might genuinely regretRed-baiting his opponents and yet continue to do so in electionafter election in order to win votes.

379


2/17

380 THOMAS GILOVICH AND VICTORIA HUSTED MEDVECBecause debates such as this one about the nature of regret

have not always been decisive, the conceptual edges of regretare not sharp, and so a rather inclusive definition is in order.Landman's definition seems to be the most suitably inclusiveand appropriate, both in general andwith respect to the empir-ical research we review in this article. According to Landman(1993),

Regret is a more or less painful cognitive and emotional state offeeling sorry for misfortunes, limitations, losses, transgressions,shortcomings, or mistakes. It is an experience of felt-reason or rea-soned-emotion. The regretted matters may be sins of comm issionas well as sins of omission; they may range from the voluntary tothe uncontrollable and accidental; they may be actually executeddeeds or entirely mental ones committed by oneself or by anotherperson or group; they may be moral or legal transgressions or mor-ally an d legally neutral. . . .(p. 36 )Economic Approaches

Anotherapproach to the subject of regret, one that combinesconceptual analysis with empirical investigation, comes fromthe field of economics. The impetus for this work was the ideathat certain violations of the axioms of rational choice, such asthat revealed in Allais' paradox, can be explained by postulat-ing that people will sacrifice monetary gain to ensure that theywill not experience subsequent regret. Thus, the anticipation offuture regret affects current choices (Bell, 1981, 1982, 1983,1985; Loomes & Sugden, 1982, 1987a). These core ideas haveconsiderable intuitive appeal, and they have received some sup-port from empirical tests (Loomes, 1988; Loomes & Sugden,1987a, 1987b;Simonson, 1992).

Unfortunately, economic theorists havedefined or operation-alized regret so narrowly that the applicability of their work ismore limited than perhaps it could be. In particular, regret hasbeen conceptualized as "the difference in value between the as-sets actually received and the highest levelofassets producedbyother alternatives" (Bell, 1982, p. 963). Although this differ-ence is certainly an important determinant of the amount ofregret experienced, other elements are also critical. Forexam-ple, the path by which a particular alternative was decided oncan have a tremendous impact on whether the choice is regret-ted, as can the way the alternatives or outcomes are framed(Harless, 1992; Thaler, 1980; Tversky & Kahneman, 1981).Bydefining regret as a reaction to the simple difference between theoutcomes of a chosen and rejected alternative, economic regrettheories cannot account for the fact that some rejected alterna-tives generate more regret than others because they are particu-larly salient, because they were "almost" chosen, or becausethey are the choices that others would make.A numberof economic theorists have also argued that regretdoes not arise if the outcomes of the rejected alternatives arenever revealed (Bell, 1982, 1983; Kelsey & Schepanski, 1991;Sage & White, 1983). Thus, no allowance is made for the factthat individuals might be tormented by what they imagine tobe the consequences of roads not taken.Counterfactual Thinking an d Regret

A number of the shortcomings of the economic models ofregret have been directly addressed by psychological researchon counterfactual thinking (Kahneman & Miller, 1986; Kah-

neman & Tversky, 1982b; Miller, Turnbull, & McFarland,1990). This research begins with the observation that eventsare not evaluated in isolation, but are compared to alternativeevents that "could have,""might have,"or "should have" hap-pened. The work on counterfactual thinkinghas focused on twosubtopics:(a) the rules by which counterfactual alternatives aregenerated (i.e., some alternatives to reality are more likely to beimagined than others) and (b) the consequences of comparingactual eventswith imagined events that might have happened.

Much of the work on the consequences of variouscounterfac-tual comparisons has focusedon the phenomenon of emotionalamplification, or the tendency for people to react more stronglyto those events for which it is easy to imagine a different out-come occurring. For example, the fate of someone who dies inan airplane crash after switchingflightsseems more tragic thanthat of a fellow traveler who was booked on the flight all along(Miller et al., 1990). The former induces a stronger reactionbecause it is so easy to imagine the person surviving "if only"he or she had not switchedflights.This phenomenon underliesa favorite technique of filmdirectors. Whenevera character inan action film announces that "This is my last mission," "I'mquitting the force," or "I think we should turn back," experi-enced moviegoers know that the character is unlikely to surviveto the final scene. The death of such a character evokes particu-larly strong emotions because he or she "almost" made it out ofharms way.

The work on counterfactual thinking overcomes two of thebiggest shortcomings of the economic models of regret. First,because the study of counterfactual thinking is concerned withpeople's imagined alternatives to reality, regret is not thought tobe restricted to those circumstances in which the outcomes ofrejected alternatives are known. Second, in contrast to the eco-nomic models, the exact path by which a decision is made isseen as critically important: Different paths to the same out-come can lead to the consideration of very different counterfac-tual alternativesand thus induce very different levels of regret.

Regrets of Action and InactionOne way hat individuals can arrive at the same outcome viadifferent paths is through action or inaction. The research on

counterfactual thinking demonstrates that the distinction be-tween omission and commission has considerable hedonic con-sequences. In what is perhaps the clearest and most frequentlyreplicated finding in the entire literature on counterfactualthinking, it appears that people experience more regret overnegative outcomes that stem from actions taken than fromequally negative outcomes that result from actions foregone(Gleicheretal., 1990; Kahneman & Tversky, 1982a; Landman,1987b). Consider an oft-cited scenario experiment by Kahne-man and Tversky (1982a):

M r. Paul owns shares in company A. During the past year heconsidered switching to stock in company B, but he decidedagainst it. He now finds out that he would have been better off by$1,200 if he had switched to the stockof company B. Mr.Georgeowned shares in company B. During the past year he switched tostock in comp any A. He now finds that he would have been betteroff by $1,200 if he had kept his stock in company B. Who feelsgreater regret? (p. 173)A rather stunning 92% of the respondents thought that Mr.


3/17

TH E EXPERIENCE OF REGRET 381George, whose misfortune stems from an action taken, wouldexperience more regret. And who could argue with them? It ismuch easier to put oneself in Mr.George's position and imaginebeing tormented by thoughts such as "This need not have hap-pened" or "I brought this on myself." It is harder to imaginegetting equally workedup if one were in Mr.Paul's position. Mr.George seems more likely to be plagued by thoughts of whatmight have or should have been in part because "it is usuallyeasier to imagine oneself abstaining from actions that one hascarried out than carrying out actions that were not in fact per-formed" (Kahneman & Miller, 1986).Nevertheless, as powerful and intuitively appealing as theseresults are, they do conflictat least on the surfacewith acommon observation from everyday life. When people areasked about their biggest regrets in life, it seems that they tendto focus on things they failed to do in their lives: "I wish I hadbeen more serious in college"; "I should have told my father Iloved him before he died"; "I regret that I never went to Eu-rope." As troubling as regrettable actions might be initially,when people look back on their lives, it seems to be their regret-table failures to act that stand out and cause greater grief.1How can this apparent conflict between the research oncounterfactual thinking and the lessons of everyday life be re-solved? We believe that both observations are valid reflectionsof important phenomena and that both reveal something sig-nificant about the experience of regret. In particular, webelievethat together the two findings indicate that there is a temporalpattern to the experience of regret over actions and inactions(Gilovich & Medvec, 1994, in press). As the literature oncounterfactual thinking suggests, people may be more upset bytheir unfortunate actions in the short run. Initially, regrettableactions may generate more counterfactual thoughts about"what might have been." Over time, however, it may be thosethings that people have failed to do that stand out andcause themost grief. Regrettable failures to act, in other words, mayhavea longerhalf-life than regrettable actions. The interesting ques-tion, then, is why might such a temporal pattern exist. What isit about the waypeople think about actions and inactions thatmakes failures to act more cognitively available in the long runand preserves their power to cause grief when the impact of re-grettable actions has faded?

Our examination of these questions proceeds in three parts.First, we investigate whether, with time, people do indeed regrettheir failures to act more than their actions. Second, we reviewevidence indicating that this tendency is part of an overall timecourse in which actions are regretted more in the short term butfailures to act are regretted more in the long run. Finally, wepresent a framework to organize the various psychologicalmechanisms that give rise to this temporal pattern to the expe-rience of regret.

What People Regret Most in TheirLivesW e are aware of no research that has directly examined theextent to which people's real-life regrets stem from action ver-

sus inaction. Nevertheless, some suggestive evidence can befound in the literature. Forexample, KinnierandMetha (1989)asked samples of young (ages 20 to 29 years), middle-aged (35to 55), and older (64 and above) respondents what they woulddo differently if they could live their lives over again. The most

common regret across all three samples was an inactionawish that they had taken their "education more seriously andworked harder on it." The same result wasobtained in two Gal-lup polls conducted in 1953 and 1965 (Erskine, 1973). Inac-tions predominated in the other most frequently mentioned re-grets reported by Kinnier and Metha's respondents as wellnot being more assertive, not taking more risks, and not spend-ing more time with family (see also Metha, Kinnier, &McWhirter, 1989).

This same question was also asked of the intellectually giftedindividuals studied by Lewis Terman. In particular, 381 menand 339 women from his original sample were asked this ques-tion in 1986 when they were mostly in their 70s (mean age = 74years). Their responses were coded by the Terman group intonumerous categories, including the 50 shown in Table 1.2 Weobtained the coded data from the Terman archives and exam-ined the extent to which the respondents listed an action or in-action as the first thing they would do differently if they had theopportunity (Hattiangadi,Medvec, & Gilovich, in press). Eightjudges, unaware of the study's hypotheses, evaluated whethereach of the 50 categories represented a regret of action, a regretof inaction, or was "indeterminate." As the data in Table 1make clear, Terman's respondents reported many more regretsof inaction than regrets of action. Overall, 54% of the regretsappeared to be regrets of inaction, whereas only 12% appearedto be regrets of action (34% were ambiguous with respect to theaction-inaction dichotomy).

1 Much has been written in the philosophical literature about whatmakes an action an action, bu t most of it involves the distinction be-tween actionsand intentions on the one hand, and actionsand conse-quences on the other (Black, 1958;Hampshire, 1960;Oldenquist,1967 ). The issue of what distinguishes actions from inactions has beenvirtually ignored. Thus, the discipline appears to have taken the samestance to the action-inaction distinction that Supreme Court JusticePotter Stewart had toward pornography: "I cannotdefine it, but I knowit when I see it." We were willing to take largely the same approach inthe research reported here. For example, in all of our ow n experimentsdescribed in this article, we never gave our subjects a precise definitionof action or inaction beyond the injunction to consider "something yo udid but wish you hadn't done, and something you didn't do but wishyou had done." It is telling that no one was ever stymied or expressedconcern about this issue.On e distinction between regrets of action and inaction that emergedfrom our data is worth noting. Regrets of action more often involveddecisions made at a specific choice point than did inactions, which weremore likely to result from an accum ulated, unfocused pattern of inac-tion. This difference wa s established by having two judges who wereunaware of our hypothesis assess whether each of the 21 3 regrets sum-marized in Table 2 appeared to arise from a decision made at a particu-lar choice point. Seventy-eight percent of the actions appeared to havebeen initiated as a resu lt of a decision made at a specific point in time,as opposed to only 34% of the inactions, x2( 1, N = 145) = 24.2, p


4/17

382 THOMAS GILOVICH AND VICTORIA HUSTED MEDVECTable 1What Terman's "Geniuses" Would Do Differently If They Could Live Their Lives Over

RegretRegrets of inaction

Should have completed college or graduate school; not have interrupted educationShould have attended college; needed more e ducationShould have worked harder; not wasted college time;been more motivatedShould have pursued a career or a professional interest; aimed high er in careerShould have been more assertive; more selfish in developing ow n abilitiesShould have emphasized social relationshipsShould have prepared for a professional career or avocationShould ha ve tried harder to be married and/or have a familyShould have had a goal fo r self; should have ha d more choices of ownShould have spent more time in family relationshipsShould have married earlier; married too lateShould have taken up a sport or exercise regim enShould have been more involved in cultural pursuitsand communi ty affairsShould have married again after first husband's death or after divorceShould have pursued a careerwhe n children were olderShould have traveled m oreShould have paid mo re attention to finances; should have saved moreShould have had more ch ildren; had too few childrenShould have spent m ore t ime w ith children/grandchildren when they were youngShould have moved to a preferred locationShould have found an occupation/career; should have ha d paid employmentShould have divorcedShould have been more assertive in marriageShould have paid more a ttention to social/nonwork lifeShould have had c hildren earlier; had childre n too lateShould have insisted children finish collegeShould have been m ore affectionate to relatives

Total

3921171615131 110773332222222

1111111

186(54%)Regrets of action

Shouldn't have married so earlyShouldn't have smoked; should have conquered alcoholism earlierShouldn't have stressed work so m u c hShou ldn't have participated in Terman studyShouldn't have managed finances badly; made bad investmentsShouldn't have divorcedShouldn't have made love and sex so important in choice of partnerShouldn't have marriedShouldn't have retired soearlyShouldn't have returned to parents ' home when marriage failed

Total

12764332211

41 ( 1 2 % )Indeterminate

Should have chosen different occupation/career or different company 37Should h ave studied different subjects; should have ha d different major 29Should ha ve chosen different mate; chose badly, poor marriage 14Should have been less self-centered; m ore g iving, understanding, and outgoing 9Should have gained more self-confidence; better self-image 9Should have delayed admission to university or high school until older 7Should have had more fu n, been more flexible; too serious; too much housework 4Shouldhave chosen different college 3Should hav e reared children differently 2Should have prepared for a career in college; needed different kind of education 1Should have retained own or family home 1Should have continued to work when children were youn g 1Should have recognized mother/son relationship of husband before marriage 1

Total 118(34%)Note. From " Failing to Act: Regrets of Terman's Geniuses," N. Hattiangadi, V. H. Medvec, and T.Gilov-ich, in press, International Journal of Aging and Human Development. Copyright by Baywood PublishingCompany, Inc. Reprinted with permission.


5/17

TH E EX P ERI EN CE O F REG RET 383Although all of these results support ou r claim that people'sbiggest regrets involve their failures to act more than their ac -tions, the data are limited in a number of respects. For onething, the respondents were nev er actually asked a bout their re-grets. Althou gh asking people what they w ould do differently ifthey had another chance is likely to tap into some regrets, it

need not. A person can look back and seize on something thatshould have been done differently without feeling any sense ofregret or remorse. One might look back and wish one hadbought shares of Microsoft stock instead of investing in TIAA-CREF, but one might do so without much emotion, or withouta sense of self-recrimination that is characteristic of regret. Inaddition, asking people what they would do differently if theyhad the chance may have predisposed them to think of inac-tions. After all, inactions can only be overturned by doing some-thing, whereas ma ny actions are overturned by not doing some-thing. Thus, to de termine w hether people really do tend to re-gret their inactions more than their actions, further research isneeded.Forced-Choice Responses

To overcome the limitations of previo us research, we con-ducted telephone surveys that directly asKed people about theirregrets of action and inaction (Gilovich & Medvec, 1994, Study1). In our initial survey, a random sample of 60 adults (M age= 40.3 years) selected fro m the Syracuse, New " Y o r k , telephonedirectory was asked the following question:When you look back on your experiences in life and think of thosethings that yo u regret, what would you say you regret more, thosethings that yo u di d bu t wish yo u hadn't , or those things that yo udidn '1 do bu t wish you had?

The question was counterbalanced across respondents interms of the order in wh ich the two types of regret were men-tioned. The results were both clear an d supportive of our hy-pothesis: Of the 60 respondents, 45 (7 5% ) indicated that theyexperienced more regret over those things they did not do butwish they ha d done (binomial p < .001). Th e order in whichthe two alternative s were listed m ade no difference.By themselves, these data do not establish unequivocally thatpeople regret their inactions more than their actions. Indeed, itmight even be the case that people actually feel more p ain overtheir regrettable actions, bu t there may be fewer of them thanregrettable failures to act. After all, the set of things one doeshas to be smaller than the set of things one does not do becausefor every action taken there are many options that are passedover. Thus, when summed over a larger number of regrettableinactions, people may report greater regret for their failures toact, even though individually they are not as potent. Althoughasymmetries such as this may contribute to any differences inregrets over actions and inactions (and to any changes in relativeregret with the passage of t ime) , it is important to determinewhether this is the sole explanation of our survey findings.To examine this issue, we selected a random sample of 30adults from the Chicago telephone directory (M age = 40.1years) an d asked them to think of their greatest regret of actionand their greatest regret of inaction. W e told each respondentnot to tell us the content of each regret, but to b e sure to have aspecific instance of each type in mind. Once they had done so ,

we asked them wh ich they regretted more . Twenty-one of the 30respondents ( 7 0 % ) expressed greater regret over their biggestfailure to act (binomial p < .05).3 Thus, when directly asked,people report that they regret their inactions more than theiractions.Free-Response Data.What happens when people are asked to list specific regretsrather than to make a summary judgment about what they re-gret most? Do they report more regrets of inaction? To find out,we asked several groups of respondents to describe the biggestregrets in their lives (Gilovich & Medvec, 1994, Study 2) . W edid not mention the action-inaction distinction; instead, wehad raters score each regret on this dimension a fterwards. Be-cause we were interestedin obtaining the regrets of a wide rangeof people, we interviewed four groups of participants. Three ofthe group s were interviewed face-to-face: 10 professors emeritiat Cornell University, 1 1 residents of various nursing homes inupstate New York, and 40 Cornell U niversity undergradu atestudents. The responses of individuals in each of these threegroups were tape-recorded and later transcribed. In addition, agroup of 16 clerical and custodial staff members at Cornell weregiven questionnaires that they returned anonymously throughthe campus m ail.All respondents were asked (either in person or byquestionnaire), "When yo u look back on your life to this point,what are your biggest regrets?" Those interviewed in personwere prompted after each response, "I s there a nything else youregret?" Those filling out written questionnaires found spacemarked off for as many as five regrets.Overall, the 77 respondents described 21 3 regrets. Each ofthese regrets was scored by two judges who were unaware of ourhypothesis. The judges determined whether each regretstemmed from an action taken, an action foregone, or circum-stances beyond the person's control (e.g., "Having polio as achild"). Th e judges agreed on the scoring of 204 of the 213regrets. The scoring of the remaining 9 was resolved by havinga third judge, also unaware of the hypothesis, cast a decisivevote."It is interesting that only 10 of the 213 regrets involved out-comes caused by circumstances beyond the person's control.Thus, a sense of personal respo nsibility appears to be central tothe experience of regret. People might bemoan or curse their

3 There were no consistent sex differences in our respondents' regretsover actions and inactions. Across the two surv eys, 43 of the 61 femalerespondents ( 7 0 % ) expressed more regret over their failures to act, asdid 23 of the 29 m ale respondents ( 7 9 % ) .4 W e initially thought that it might be difficult to decide whether aspecific regret belonged unambiguously to one of the tw o categories be-cause every action implies a corresponding inaction and vice versa.When people say, "I regret that I joined the Navy when I was 18," dothey regret the action itself or not doing all the things that joining theNavy prevented them from doing? We decided to code each regret ac-cording to what th e respondents themselves emphasized ( in this case,being in the Navy, not the lost opportunities caused by joining theNavy), and by doing so the scoring was not at all difficult, as the veryhigh inter-rater reliability indicates. Follow-up questions addressed to asubset of our respondents indicated th at whatever they chose to empha-size in their initial statem ent w as indeed the tru e source of their regret.


6/17

384 THOMAS GILOVICH AND VICTORIA HUSTED MEDVECTable 2Most Common Regrets

Failures to actMissed educational opportunityFailure to seize the mom entNo t spending enough time with friends an drelativesMissed romantic opportunityNo t pursuing interest in "X"Missed career opp ortunity; insufficienteffortNo t making financial transactionMiscellaneous inaction

Total

n212115131 17238

128

ActionsBad educational choiceRushed in too soonSpent time badlyUnwise romantic adventureWasted time on "X"Bad career decision; wasted effortUnwise financial actionMiscellaneous action

Total

n3174

1003632

75Note. An additional 10 regrets dealt with events outside the person's control and thus lie outside the action-inaction dichotomy. From "TheTemporal Pattern to the Experience of Regret," by T. Gilovich and V . H. Medvec, 1994, Journal of Personality and Social Psychology, 67 , p. 360.Copyright 1994 by the A merican Psychological Association. Adapted by permission.

bad fate, but they rarely regret it in the sense that the term istypically understood. As for the events people do regret, the re-sults provided strong support for our hypothesis: Regrettablefailures to act outnumbered regrettable actions by nearly a 2 to1 margin (63%vs . 3 7 % ) .A number of subsidiary issues were also examined. First,male and female respondents did not differ in terms of how fre-quently they mentioned regrets of action versus regrets of inac-tion. Second, there was some (not statistically significant) evi-dence that older individuals were more likely than younger re-spondents to regret things they failed to do. For instance, 74%of the regrets listed by our two oldest samples (the professorsemeriti and the nursing home residents) involved things theydid not do, as compared to 61% for our two youngest samples(the students and the staff members).5A third issue concerns the precise content of individuals' re-grets beyond the action-inaction dichotomy. What do peopleregret most in their lives? Table 2 presents the regrets most fre-quently mentioned by our respondents. The taxonomy pre-sented in Table 2 was designed to be balanced with respect tothe action-inaction dichotomy. Forexample, because "not pur-suing interest in 'X' "was a commonly mentioned regret, it ap-pears as an inaction category along with its complementary ac-tion category "wasted time on 'X'"(see Gilovich & Medvec,1994, Study 2, fo r details on scoring).A n examination of Table 2 indicates that the most commonregrets involved such inactions as missed educational opportu-nities, the failure to "seize the moment," and not devotingenough time and effort to personal relationships. The mostcommon regret of action was to "[rush] in too soon." Sadly, itseemed to us that people's regrets reflect a trade-off betweeneducational and career pursuits on the one hand and interper-sonal relationships on the other: Those who spent time on inter-personal relationshipsregrettednot achieving more profession-ally; those who spent time in professional pursuits regretted notdevoting enough attention to friends and family.Table 2 also reveals that no one regretted spending time de-veloping a skill or hobby, even when the skill lies dormant andthe hobby is no longer pursued. No one reported anymisgivingsabout spending their adolescence learning ballet or collecting

coins, even though, as adults, they have given up dance and thecoin collection no longer holds interest. Compare this to the 11entries in the corresponding inaction category, "Not pursuinginterest in 'X'." The Difference within this pair of action-inac-tion categories captures our main finding with particular clar-ity: When people look back on their lives, it is the things theyhave not done that generate the greatest regret.

The Temporal Aspects ofRegretHaving established that people regret their failures to act

more than their actions, we return to the conflicting patternof results with which webegan. Scenario experiments like thestock market example of Kahneman and Tversky (1982a)indicate that commission looms larger than omission in theexperience of regret. However, surveys of people's greatestreal-life regrets reveal that people are more troubled by theiromissions. Of course, there are important differences in thetypes ofdata on which these contradictory findings are based,and it may be these differences that are responsible for thedivergent results. People who have been asked about specific,short-term regrets may be evaluating fundamentally differentkinds of events than those queried about the biggest regrets

5 On e reason the effect of age may have fallen short of statistical sig-nificance is that our respondentsdid not represent the entire life span.No one under college age was included. Our results ma y have beendifferent had we included younger respondents. Adolescents and chil-dren may be less likely to regret the things they have not done becausethe future can seem limitless at that age and things undone may simplybe assigned to a less troublesome category of things "yet to be done." Inaddition, it seems to be part of the "job" ofbeing a childto act in waysthat lead to trouble in order to ascertain the limits of the social andphysical world. Negative outcomes that stem from actions taken aretherefore likely to be highly available. To test these suppositions, weasked junio r high school and high school students the same forced-choice question we asked our telephone survey respondents. Unlike ouradult subjects, a majority of whom expressed greater regret over theirfailures to act, the adolescents were equally divided between those whofelt worse about their actions and those wh o felt worse about their fail-ures to act.


7/17

TH E EXPERIENCE OF REGRET 385of their lives. In addition, there may be various subtle butimportant differences in what is meant by regret when ap-plied to these two temporal perspectives.

Our thesis, however,is that these two different sets of resultsare manifestations of a temporal pattern to the experience ofregret. Actions tend to generate more regret in the short term,but inactions tend to be more troubling in the long run. Wehave obtained evidence consistent with this idea from two verydifferent kinds of experiments (see Gilovich & Medvec, 1994,Studies 3-5).Intuitions About the Short-Term and Long-Term Reg retsofOthers

Consider the following scenario:Dave and Jim do not know each other, but both are enrolledat the same elite East Coast university. Both are only moderatelysatisfied where they are, and both are considering transferring toanother prestigious school. Each agonizes over the decision, goingback and forth between thinking he is going to stay and thinking hewill leave. They ultimately make different decisions: Dave opts tostay where he is and Jim decides to transfer.Suppose their decisions turn out badly for both of them: Davestill doesn't like it where he is and wishes he had transferred, an dJim doesn't like his new environment and wishes he had stayed.

Students who were presented with this scenario were thenasked (a) "Who do youthink would regret hisdecision the mostupon learning that it was a mistake?" and/or (b) "Who do youthink would regret his decision the most in the long runl"

Whether participants answered both questions in a within-subjects designor only one of the two in a between-subjects for-mat, most ( 7 6 % ) thought that Jim, the person who regrets doingsomething, would experience more regret in the short termthat is, upon learning his action was a mistake. When askedabout the long run, however,the majority (63%) thought thatDave, who regretted not doing something, would experiencemore regret (Gilovich&Medvec, 1994, Studies 3 and 4).

These data suggest that people are familiar with both sides ofthe temporal profile we haveproposed. People have apparentlybeen exposed to the intense, immediate pain that stems from aregrettable action; at the same time, they are aware that in thelong run it is often the failure to act that causes more distress.Recent and Life-Long Regrets

To move beyond the scenario methodology and examine thetemporal profile of people's own real-life regrets, werecruited agroup of adults to fill out a questionnaire that asked them torecall (but not write down) their single most regrettable actionsand inactions both from the past week and from their entirelives. Then, for each time period, they were asked to indicatewhich they regrettedmore, the action or the inaction. As antic-ipated, participants' responses depended on the time period un-der consideration. When focused on the last week, the respon-dents were rather evenly split between those who most regrettedtheiractions ( 5 3 % ) and those who most regretted their failuresto act. However, when looking back over their entire lives, a sub-stantial majority (84%) reported greater regret for what theyfailed to do (Gilovich &Medvec, 1994, Study 5). Although peo-

ple's actions may be troublesome initially, it is their inactionsthat plague them most with long-term feelings of regret.

W hy Regrets Shift Systematically With TimeW e have argued that the discrepancy between past research

and our more recent findings indicates that there is a consistenttemporal pattern to the experience of regret. Actions producegreater regret in the short term; inactions generate more regretin the long run. Several tests of this contention were supportive.Why might such a temporal pattern exist? What is it about theway people think about their choices and their lives that dimin-ishes regrets of action but strengthens dismay over regrettablefailures to act?

As with many complex psychological phenomena, there isunlikely to be a single answer. The temporal pattern to the ex-perience of regret is no doubt multiply determined. Indeed, wepropose that there are three distinct categories of mechanismsthat together give rise to this temporal pattern. First, there arethose elements that decrease the pain ofregrettable actions. Sec-ond, there are those elements that bolster the pain of regrettableinactions. Finally, there are those elements that differentiallyaffect the cognitive availability (Tversky & Kahneman, 1973)of a person's regrettable commissions andomissions. These lat-ter elements do not affect the intensity ofregret over actions andinactions, but they do affect how often one is reminded ofsuchregrets and therefore howoften they are experienced.

Our explanatory framework is summarized in Table 3.Among the processes that reduce the pain of regrettable actionis a tendency for people to take steps to undo orameliorate theirregrets of action more than their regrets of inaction. A personwho regrets marrying Mr. Wrong will likely get divorced; some-one who regrets passing up Mr. Right typically must cope withthe fact that he is no longer available. Regrettable actions tendto be further diminished by the identification of "silver linings"that offset the pain they cause. Aperson whomarries Mr. Wrongwill often say, "But I have these wonderful children I would nothave had otherwise"; someone who passes up Mr. Right typi-cally finds less consolation. Finally, identifying silver linings isbut one way of rationalizing or reducing dissonance for a baddecision. As we demonstrate below, there is reason to believethat regrettable actions tend to prompt more vigorous efforts toreduce dissonance than do regrettable failures to act, furtherreducing the pain ofunfortunate actions relative to unfortunateinactions.

Ofcourse, the factors that serve to reduce the pain of regret-table action tend to do so for regrettable inaction as welltheyjust do so less consistently, less vigorously, or less successfully.For inactions, furthermore, such efforts at "damage control"are often undermined by several processes that conspire to bol-ster the pain of regrettable inaction. First, because the passageof time often brings with it increased confidence that one couldhave performed an earlier task successfully, the original reasonfor many inactionsa lack of confidencecan seem less thancompelling with the passage of time. Someone might decline tomarry Mr. Right out of fear that their religious differenceswould be insurmountable. Looking back, however, the personmay increasingly come to believe that "we could have workeditout." This tendency for inactions to seem inexplicable af-terwards is accentuated by a tendency for causal factors that


8/17

386 THOMAS GILOVICH AN D VICTORIA HUSTED MEDVECTable 3Causes of th e Temporal Pattern to the Experience of Regret

1 . Factors That Reduce the Pain of Regrettable Actions More Than the Pain of Regrettable InactionsBehavioral Repair Work: More com pensatory steps tend to be taken to ameliorate regrettable actions.Psychological Repair Work I: More silver linings tend to be associated with regrettable actions.Psychological Repair Work II : More dissonance reduction tends to be induced by regrettable actions.

2. Factors That Increase the Pain of Regrettable Inactions M ore Than the Pain of Regrettable ActionsInexplicable Inaction 1: The passage of time brings an increase in retrospective confidence that makesearlier failures to act inexplicable, and hence inexcusable.Inexplicable Inaction I I: Because factors that inhibit beh avior tend to b e less salient than those thatinduce behavior, failures to act can seem inexplicable, and increasingly so with the passage of time.What Was Versus What Might Have Been: The consequences of regrettable actions tend to be finite;the consequences of regrettable inaction tend to be psychologically infinite.

3. Factors That Prom ote the Cognitive Availability of Regrettable Inactions More Than the Availabilityof Regrettable ActionsThe Zeigarnik e f f e c t : Regrettable failures to act tend to be more memorable and enduring thanregrettable actions.

compel behavior to be more salient than those that inhibit be-havior (Dunning & Parpal, 1989; Hansen & Hall, 1985; Read,1985). The person who passes up Mr. Right often has troublein retrospect thinking of a truly compelling reason why sucha choice was ever made. In contrast, someone who mistakenlymarries Mr. Wrongcan nonetheless recall howmuch fun he wasat one time,howresponsible he seemed back then, orhowmucheveryone liked him. Finally, there is an important asymmetryin the perceived negative consequences of regrettable actionsand inactions: Regrets of action center around bad things thatactually happened; regrets of inaction involve good things thatone believes would have happened. Because of the open-endednature of the latter, the perceived consequences of missed op-portunities often grow in scope and importance with the pas-sage of time. Someone who chooses not to marry Mr. Rightmay developan unrealistic picture of the marital bliss that wasforegone.In addition to these mechanisms that alter the intensity ofregrettable actions andinactions, there is reason to believe that,over time, regrets of inaction occupy the mind more often thanregrettable actions. Many regrets of inaction involve unrealizedambitions and unfulfilled intentions that are more available inmemory than those goals or tasks that have been successfullycompleted (Seifert & Patalano, 1991; Zeigarnik, 1935).

None of this is to suggest that the processes we have brieflydescribed always work in precisely this fashion. We view eachof these as tendencies that, in general, give rise to the observedtemporal pattern to the experience of regret. There are boundto be occasional exceptions, and no doubt many readers havebeen able to think of counterexamples in which, say, a particu-lar inaction is accompanied by a powerful silver lining or a spe-cific restraining force is unusually salient. Nevertheless, we be-lieve that there is ample theoretical and empirical support foreach of the mechanisms wehave put forward. We now turn to amore in-depth look at the theoretical and empirical basis ofeach of these three classes of mechanisms.Factors That Reduce the Pain of RegrettableActions

Behavioral repair work. When bad things happen, we some-times sit idly by and bemoan our bad judgment or bad luck.

Other times, however, we take decisive action to overcome,undo, or compensate for the negative outcome. If we sign onwith the wrong organization, we switch companies or even ca-reers; if we do something to hurt or offend someone, we apolo-gize and vow to do better in the future . When our actions get us"off course" in some way,we endeavor to set ourselves right.The claim, in other words, is that people do things to alleviatethe pain of their regrettable choices, and they are more likelyto do so for their actions than for their failures to act. Theirregrettable failures to act often just sit there, causing continuedgrief.

There are at least two reasons why people may be more in-clined to rectify their regrettable actions than their regrettablefailures to act. First, as our earlier data make clear, the pain ofregrettable action is often felt more quickly than that of regret-table inaction and, therefore, provides more of an impetus toimmediate change. And, as everyday experience demonstrates,that which is not dealt with immediately is often never dealtwith at all.A second, related reason to expect people to do more abouttheir regrettable actions than their regrettable inactions is bestunderstood by thinking of behavior in Lewinian terms (Lewin,1938, 1951) . Lewin argued that various psychological andphysical forces form a tension system that determines the be-havior of the individual. Because the forces in the tension sys-tem tend toward equilibrium, people are subject to a fairamount of inertia. From a Lewinian perspective, then, when-ever people act, they overcome whatever inertia had kept themin the position they were in beforehand, which upsets the bal-ance of forces that existed previously. Byacting, in other words,people alter the tension system in which they were embedded,creating a new situation in which the forces acting upon themare less likely to be in equilibrium. With the operative forcesin flux, it is relatively easy to take steps to overcome whatevermistakes were made. It is often easiest to make a move when ourbags are already packed, so to speak.

In contrast, when people fail to act, they are still heldbydefinitionin the grip of preexisting inertial forces. By not act-ing, in other words, individuals remain in the same situation as


9/17

TH E EXPERIENCE OF REGRET 387before with the forces acting upon them in equilibrium. What-ever it was that made it difficult to pull free of these forces ini-tially often makes it difficult to pull free later on. As a result, it isrelatively difficult to change from initial inaction to subsequentaction. Like many other things in life, behavior is subject tomomentum.

To test this contention, we asked 60 adults recruited fromvarious locations around Ithaca, New \brk, to recall (but notdescribe) their single most regrettable action in their lives andtheirsingle most regrettable failure to act. Once they had a spe-cific instance of each category in mind, we asked them,

Have you done anything in an effort to overcome or compensatefor your regrets? In particular, for which regret have you engagedin the most effective remedial action? By remedial action we meanany actions or l i f e changes you made to offset the regret; we do notmean any psychological changes or strategies you have employedto try to think differently about the regret or cope better with it.Asexpected, a substantial majority of the respondents (65%)

indicated that they had made more significant changes to dealwith their most regrettable action than their most regrettableinaction (binomial z = 2.20,p < .03).Because of this difference in steps taken to ameliorate regret-table actions and inactions, unfortunate actions should be lessbothersome over time. They become less troubling after the re-medial steps have been taken. As a result, commissions maygenerate the most regret initially, but omissions may prove moretroublesome in the long run.Psychological repair work I : Identifying silver linings. Thereare many ways to deal with negative events. Sometimes, one cancope by taking remedial action. Other times, however, there isno effective action to be taken, and so efforts to come to gripswith the problemare confined to the psychological realm. Peo-ple engage in "psychological work" designed to lessen the painof the unfor tuna te event. Once again, we argue that people aremore likely to engage in effective psychological work for theirregrettable actions than fortheir regrettable inactions.

Consider one of the most common ways people cope with neg-ative events, by noting how much they have learned from theexperience. In other words, people acknowledge that the out-come may have been regrettable, but they mentally offset theregret by identifying a silver lining that consists of how muchthey profited from the experience as well. Such a silver lining"But I learned so much"is obviously much more likely toapply to regrettable actions than regrettable inactions. Peopletypically learn more by doing newthings than by sticking to oldpatterns. In fact, for most inactions the claim that "I learned somuch" hardly makes sense. ("I did not do anything, but Ilearned so much?")

But taking stock of how much one learned from an unpleas-ant experience is not the onlyway o lessen the pain of a negativeoutcome. There are other silver linings that serve the same pur-pose. To examine whether people are generally more likely toidentify offsetting positive elements for their unfortunate ac-tions than fortheir unfortunate failures to act, weasked anothergroup of 60 adults to think of their single most regrettable ac-tion in theirlives andtheir single most regrettable failure to act.Once respondents had a specific instance of each category inmind, we asked them which regret was offset by the most sig-nificant silver lining.Asexpected, the vast majority ( 7 5 % ) citedtheirmost regrettable action (binomial z = 3.74, p < .001).

Thus, in addition to making more material changes to over-come the negative ramifications of a regrettable action (see re-sults above), people also engage in more psychological work di-rected toward the same end. What cannot be accomplished ma-terially can be dealt with psychologically. The net effect ofbothof these processes is that the initial sting of a regrettable actiontends to diminish with time, more so than the pain associatedwith a regrettable failure to act.

Psychological repair work II: Dissonancereduction. Iden-tifying a silver lining in a dark cloud can be one way to reducethe cognitive dissonance associated with a negative outcomebrought on by a freely chosen action (Aronson, 1969; Cooper&Fazio, 1984; Festinger, 1957; Wicklund & Brehm, 1976). In-deed, the statement that "I learned so much" can be thought ofas a generic form of dissonance reduction that can apply tonearly any unfortunate course of action. As we have shown,people are more likely to use such a generic dissonance reduc-tion technique fortheir regrettable actions than fortheir regret-table failures to act. Are people also more likely to use othermodes of dissonance reduction fortheir regrettable actions thanfor their regrettable inactionsmodes more precisely tailoredto the specific behavior in question?

There are reasons to believe they might. First, as the literatureon counterfactual thinking indicates, people initially findnega-tive outcomes brought on by their actions more troublesomethan identical outcomes brought on by inaction. The more aver-sive the outcome, in turn, the more motivation there is to reducedissonance (Cooper & Fazio, 1984; Cooper & Worchel, 1970;Nel, Helmreich, & Aronson, 1969). Second, actions are typi-cally more salient than inactions and therefore constitute amore noticeable target for psychological repair work than fail-ures to act. As the extensive literature on the "feature positiveeffect" attests, what does happen has more psychological im-pact than what does nothappen (Fazio, Sherman, &Herr, 1982;Newman, Wolff, & Hearst, 1980). Finally, because inaction isoften viewed as the status quo and action as a departure fromthe norm (Kahneman & Miller, 1986), people generally feelmore personally responsible for their actions than for their in-actions. With a greater sense of personal responsibility for theoutcome comes a greater amount of dissonance to be reduced(Cooper, 1971; Cooper & Fazio, 1984; Goethals, Cooper, &Nancy, 1979; Hoyt, Henley, & Collins, 1972). Thus, becausepeople's regrets of action are more salient, more painful ini-tially, and engender more of a sense of personal responsibility,they should elicit more cognitive dissonance (and dissonancereduction) than corresponding failures to act. If people engagein more dissonance reduction for their regrettable actions thanfor their regrettable inactions, they might regret their actionsmore initially but end up regretting their inactions more in thelong run.

A recent experiment verified this tendency for people to re-duce dissonance more for commissions than for omissions(Gilovich, Medvec, &Chen, 1995). The experiment was a lab-oratory analog of the "Monty Hall" or "three doors" problem(Engel & Venetoulias, 1991; Tierney, 1991; vos Savant, 1990a,1990b, 1991). Subjects were recruited for a study of group de-cision making and were told they would receive, depending onthe decisions they made, either a grand prize or a modest prizefor their participation. The prizes were hidden in three boxesand, as on Monty Hall's Let's Make a Deal television program,


10/17

388 THOMAS GILOVICH AND VICTORIA HUSTED ME D V E Cthe subject's first task w as to choose one of the boxes. Tw o of theboxes were said to contain modest prizes, and one was said tocontain the grand prize. After the initial box was selected, thesubject confronted a choice of whether to keep his or her un-opened box or trade it in for one of the remaining boxes. How-ever, before the choice was made, the experim enter (like Monty)opened one of the two remaining boxes and revealed a modestprize. The subjec t's choice, th en, was whether to keep his or herbo x or exchange it for the rem aining, unopened box.Subjects made this latter decision in consultation with an-other "subject" who was their partner in the group decisiontask. In reality, the partner was a confederate whose job wasto ensure that the subject chose in accordance w ith his or herrandomly assigned conditionto Stay with the original choiceor Switch to the other box. Th e confederate thus influenced thesubject's choice when necessary (i.e., when the subject ap-peared inclined toward the "wrong" box) , but did so in asufficiently delicate way such that the subject was unaware ofany influence.6The ex perim ent was arranged so that regardless of the sub-ject's "choice," he or she received a modest prize (a Cornellbumper sticker) an d missed out on the grand prize (an expen-sive Cornell T-shirt). The experiment was designed to ascertainwhether subjects who lost out on the grand prize because of anaction (i.e., those who switched boxes) would reduce their dis-sonance over the outcome more than those who experienced thesame fate after deciding not to act (i.e., those who kept theirinitial boxes). In particular, the experiment exam ined whethersubjects in the Switch condition came to view the modest prizethey received as more attractive than did subjects in the Staycondition. After learning about w hich prize they won , subjectswere asked to specify how much money we would have to offerthem to get them to sell back the bumper sticker to us.The results are reproduced in Figure 1. Subjects wh oswitched boxes and ended up with the bumper sticker de-manded a significantly higher price to sell it back than did thosewho received the same prize by sticking with their originalchoice. Regrettable actions seemed to induce more dissonancereduction than did regrettable failures to act. The responses ofboth groups of subjects were also comp ared to those of subjectsin a control condition whose "decisions" were determined ran-domly. As dissonance theory would predict (Cooper, 1971; Coo-per & Fazio, 1984; Goethals et al., 1979; Hoyt et al., 1972),these subjects, wh o were not personally responsible for theiroutcomes, v alued the bum per sticker least of all.Summary. We have demonstrated three ways in which peo-ple are able to reduce their regret over unfortu nate actions withthe passage of time. People m ake more life changes to cope withtheir regrettable actions than their regrettable inactions; theyare more likely to find a silver linin g that justifies such actions;and they engage in more dissonance reduction to reduce thepain of their errors of commission. Over time, these three pro-cesses collu de to lessen the impact of mistaken actions. By com-parison, regrettable failures to act seem more troublesome astime goes by. This latter tendency is accentuated by several psy-chological processes that serve to bolster the intensity of regretsof omission w ith the passage of time. It is to these processes thatwe no w turn .Factors That Bolster th e Pain of Regrettable Inactions

Inexplicable inaction I : Confidence and temporal perspective.Many regrets of inaction arise from an inability to conquer our

fears or overcome our doubts when the "moment of truth" is athand. We fail to m ake a career change because we are un sure ofwhat the outcome will be. We do not ask someone for a datebecause we are afraid of rejection. We do not "let loose" or evenventure out on the dance floor because we think we might lookungainly or out of fashion.These concerns, however, which seem so pressing when thetime to act is at hand, may diminish with the passage of time.The fur ther removed we are from the occasion, the more con-vinced we become that w e could have or would have done justfine. It is easy to be confident when the task is not imminent ; itis hard to be so assured wh en the challenge is at hand.Of course, any such tendency for confidence to soar with in-creasing distance from an eve nt would tend to magnify the feel-ing of regret over a failure to act. If the fears that kept us fromperforming some action tend to dim inish with time, the reasonswe had for not acting will no longer seem compelling. With nocompelling reason for failing to do something we wish we haddone, we regret our failure to act even more (Lecci, Okun , &Karoly, 1994). We curse ourselves by asking "Why didn't I atleast try?" and torm ent ourselves with accusations that "I'mjust too timid" or "I'm too indecisive."A recent series of studies demonstrates that people are indeedmore c onfident that they will do well at a task the further theyare, either prospectively or retrospectively, from the time thetask is to be performed (Gilovich, Kerr, & Medvec, 1993; seealso Nisan, 1972, 1973). People feel much better about theirchances of success well beforehand than right beforehand. Like-wise, people are more convinced that they would have done well

6 The three doors problem was chosen as the backdrop of ourexperi-ment because the choice of either decisionto stay or to switchcanbe made to seem compelling, and therefore subjects could be readilyinfluenced in the direction of their randomly assigned conditions. Th ecorrect response to the problem is to switch (see next paragraph).Bydoing so one has a 67%chance of receiving the grand prize (and thusonly a 33% chance of receiving it by sticking with one's original choice).Subjects could thus be induced to switch by getting them th inking alongth e lines of the correct solution to the underlying probability problem.At th e same time, however, it is easy to convince a subject that, withonly two remaining unopened boxes, there is a 50-50 chance that thegrand prize is in either one. Thus, with no "objective" reason to favorone over the other, most people can be induced to stick with their origi-na l box.To see why the correct solution is to switch boxes, consider thefollowing: A person's chance of initially picking the box with the grandprize is, obviously, 33%. Because th e person already knows that th e"host" can only open a box containing a modest prize (uncovering thegrand prize would eliminate the d ilemm a) , nothing relevant is learnedby revealing th e contents of one of the boxes. Since nothing ha schanged, the chance that the person's original box contains the grandprize is still 33%, and therefore the person should switch to the otherbo x that has a 67%chance of having the grand prize.A more compelling explication of the correct solution can be ob-tained by simply listing all combinations of (a) the person's choice ofinitial box, (b) the person's choice to stay orswitch, and ( c ) the "host's"choice of which of the two remaining boxes to open. Such a list readilyreveals that on e wins 67% of the time when on e switches an d only 33 %of the time when one stays.The reader may wish to consult Engel and Venetoulias (1991), Tier-ne y ( 1 9 9 1 ) , and vos Savant (1990a, 1990b, 1991) fo r accounts of therather entertaining and controversial history of this problem.


11/17

THE EXPERIENCE OF REGRET 389$1.58

Switch Stay ControlFigure 1. Mean amount of money subjects demanded to exchange bumper sticker, by condition. From"Omission, Commission, and Dissonance Reduction: Overcoming Regret in the Monty Hall Problem," byT. Gilovich, V. H. Medvec, S.Chen, 1995, Personality and Social Psychology Bulletin, 21, p. 186. Copyrightby Sage Publications, Inc. Adapted with permission.

at a task long after, rather than right after, the critical momentwas at hand.It is this latter effect of retrospective temporal distance that ismost relevant to the experience of regret. In the critical test ofthiseffect, weasked Cornell students andalumni to consider theimpact of adding a challenging course to their work load duringa typical semester. How much would it affect their grade-pointaverage for the semester? The amount of sleep they got? Theirextracurricular andsocial lives?One group ofrespondents con-sisted of Cornell alumni who had been out of college for anaverage of 3.5 years. Another was a group of current studentswho were asked ho w much the extra course would affect theircurrent semester. Finally, a third group consisted of studentswho were asked ho w much it would have affected themthe pre-vious semester.The relevant data are presented in Table 4. As anticipated,participants' assessments of how well they could cope with anincreased work load were directly related to their distance fromthe time the extra burden was o be faced. The alumni indicatedthat the extra work load would take less of a toll on their aca-demic and social lives than did either group of current students.Furthermore, current students whowere asked to consider theimpact on a recently completed semester expressed less concernabout disruption than didthose making assessments fo r thecur-rent semester. It seems that the further away the challenge, theless threatening it appears.These data have obvious implications for the experience ofregret. As our confidence that we could have done well at sometask increases, the rationale for our original inaction fades. Notonly might we now wish we had gone ahead and acted, but wefindit difficult to sympathize with our previous failure to do so.Our regret intensifies because our failure to act now seems soinexplicable. This retrospective increase in confidence some-times increases a person's regret over failing to act when thechances of success were in fact rather high. On other occasions,however, such revisionist tendencies may be sufficiently strongto convince a person of the feasibility ofrather implausible, pie-in-the-sky wishes.

In either case, it should be clear how this process gives rise tothe observed temporal pattern to the experience of regret. Asmany have noted (Kahneman & Miller, 1986; Kahneman &Tversky, 1982a, 1982b; Miller etal., 1990) regret is fundamen-tally a counterfactual emotion: It arises from the contrast be-tween what has actually happened and an easily imagined al-ternative outcome that might have happened. As those workingin the area of counterfactual thinkinghave shown, action tendsto generate more regret than inaction (in the short term) be-cause it is typically easier to "undo" an action than an inaction.However, the present results suggest that this may not be true inthe long run. Because our confidence that we could have donesomething increases with the passage of time, it becomes pro-gressively easier to psychologically "undo" a regrettable failureto act. Thus, the relative ease of undoing actions and inactions

Table 4Mean Estimated Impact of Increased Work Load as aFunction of Temporal Perspective

Dependent measureTemporal perspective

Current Recent DistantGPA (grade points lost)Sleep (minutes lost)Extracurricular activity (9-pointrating scale)Social life (9-point rating scale)

0.3361.44.24.9

0.2445.23.34.9

0.1824.52.93.7

Note. The Current condition consisted of Cornell University studentsmaking assessments of the impact on their current semester. The Recentcondition consisted of Cornell students making assessments of the im-pact on their most recently completed semester. The Distant conditionconsisted ofCornell alumni making assessments of the impact on a typ-ical semester during their undergraduate years. From "The Effect ofTemporal Perspective on Subjective Confidence," by T. Gilovich, M.Kerr, and V. H. Medvec, 1993, Journal of Personality and Social Psy-chology, 64, p. 555. Copyright 1993 by the American Psychological As-sociation. Adapted with permission.


12/17

390 THOMAS GILOVICH AND VICTORIA HUSTED M EDVECappears to reverse with the passage of time, with predictableconsequences for the experience of regret.Inexplicable inaction I I : The differential impact of compel-ling and restrainingforces. Th e increase in confidence that oc -curs with the passage of time represents one way in which theelements that originally restrained a person's actions can losetheir psychological impact over time and thereby intensify re-gret over inaction. There is reason to believe, moreover, that thismay be but one aspect of a very general pattern: Forces thatrestrain h um an action may be inheren tly less salient than forcesthat com pel action. As a result, it may b e easy for individuals toge t in touch with why they acted (thereby justifying regrettableactions), but more difficult for them to understand why theyfailed to act (thereb y rendering regrettable failures to act rathermystifying).Ku rt L ewin (19 51) w as one of the first to point ou t that com-pelling forces are typically more salient than restraining forces.Lewin noted that whe n trying to induce another person to dosomething, people are generally tempted to increase the per-son's motivation for doing sothat is, they try to "push" theperson in the direction they want him or her to go. However,Lewin also pointed out that it is often more effective to elimi-nate w hatever has kept the person from behav ing in the desiredfashion in the first place. H e argued that often one should devotemore effort toward removing the barriers to change and lesseffort toward supplying additional reasons to change.Lewin would not be credited with this idea if it were not tosome degree co unterintuitive. Thus, the fact that Lewin's advicewas novel suggests that those factors that compel action loomlarger in intuitive assessments of behavior than those factorsthat restrain action. People have known all along about how toinduce behavior by increasing the pressures to change; their at-tention had to be drawn to the possibilities inheren t in removingth e impediments to change. This implies that reasons for re-fraining from doing something may typically be less accessiblethan reasons for doing something, and, as we argue below, thisasymmetry may increase over time .A number of recent experiments support this thesis. Read(1985 [described in Kahneman & Miller, 1986]) taught stu-dents a simple card game and then showed them various handsof tw o hypothetical players. They were then asked to reverse theapparent outcome by changing the han d of one of the players.The studen ts typically chose to change the outco me b y strength-ening the losing hand rather than by weakening the winninghand. In other words, they apparently found it more natural toadd elemen ts that w ould compel a victory than to delete ele-ments that hadprevented it.

Similar results have been obtained in the domain of causalattribution (Hansen & Hall, 1985). When people were toldabout the outcom e of various contests (e.g., hockey game, a tug-of-war), they were more inclined to attribute the outcome tothe strength of the winning side than to the weakness of theloser. That which causes victory looms larger than that whichprevents it. Dunning and Parpal (1989) obtained analogous re-sults in their work on "mental addition" an d "mental subtrac-tion." Participants in the mental addition condition of their ex -perim ents were asked questions such as "How m any more ques-tions will you get right if you study for an upcoming exam?"Those in the m ental subtraction condition were asked questionssuch as "How many ewer questions will you get right if you do

not study for an upcoming exam?" Participants indicated thatadding a bit m ore studying time would help their performa ncemore than subtracting an equal amount ofstudying time wouldhurt it. That which compels (e.g., studying) has a greater im-pact than th at which impedes (e.g., no t studying).These various findings suggest tha t, after the fact, it may beeasier to get in touch with why one did something that oneshould not have done than it is to fully comprehend why onefailed to do something one should have done. This differenceha s obvious implications for the experience of regret. K nowingwhy one acted tends to lessen the self-recrimination and re-morse that accompany regrettable actions. In contrast, the lackof a satisfying explanation for inaction intensifies the amountof self-recrimination and regret that is experienced over failuresto act. F urtherm ore, this asymm etry in the accessibility of rea-sons for action and inaction may become more pronouncedwith the passage of time. In the short term, one can rely onrelatively accurate "bottomup" processes that call up specificmemories of the thought processes that led to action or inac-tion. Over time, however, one must employ less reliable "topdown" processes that generate inferences about the kinds of rea-sons on e must have had for one's actions or failures to act. It iswhen these mo re abstract in feren tial processes are engaged th atthe differential impact of compelling an d restraining forces islikely to be most pronounced.What was versus what might have been: Differences in theimagined consequences of regrettable actions and inactions.There are systematic differences in the feedback people receivefrom acting an d from no t acting, and these differences tend tomake regrettable omissions loom larger than regrettable com-missions. What is troubling about a regrettable action is the setof bad things that actually happened as a consequence of theaction taken. Thus, the consequences of regrettable actions areoften finite: They are bounded by wha t actually happened. Incontrast, w hat is troublesom e about a regrettable inaction is theset of good thing s that would have happened had one acted. Theconsequences of inactions are therefore potentially infinite:They are bounded only by one's imagination.To documen t this asym me try in the perceived consequencesof regrettable actions and inactions, we asked a group of 20adults to think of their greatest regret ofaction andtheir great-es t regret of inaction. W ithout describing their regrets, they thenrated each one on a 10-point scale indicating ho w different ". . .your life would be or would have been had you done what yo uregret not doing (not done what you regret doing)" It is inter-esting that 2 subjects stated that they could no t th ink of anyregrettable actions in their lives, and 2 other subjects indicatedthat their regrettable ac tions have, in the long run , actually im-proved their lives.Among the remaining 16 respondents, the av-erage rating of how mu ch their lives would change by undoingtheir regrettable inactions was significantly higher (M = 6.8, SD= 2. 1) than their average rating of how muc h their lives wou ldchange by undoing their regrettable actions (M = 4.8, SD =

To pursue this further, we asked an additional group of 38adults to likewise th in k of their biggest regret of action and the irbiggest regret of inaction. We then asked each of them to discusswhy their action and inaction were regrettable hat is, to de-scribe the unfo rtuna te consequences, immediate or delayed, ofeach. Their comments were tape-recorded and later scored by


13/17

TH E EXPERIENCE OF REGRET 391judges unaware of our hypothesis. The judges determined howmuch time was spent discussing the consequences of each re-gret. Consistent with our earlier findings, the participants hadmore to say about the consequences of their regrettable inac-tions than their regrettable actions. On average, they spent 42seconds talking about the consequences of their inactions andonly 29 seconds talking about their regrettable actions, f (37) =3.33, p < .005. When combined with the data from the studyjust described (in which participants rated how different theirlives wouldhave been had they acted differently), these resultsindicate that people tend to believe that more harm has comefrom their failures to act than from their actions.

Note that this difference in the perceived consequences of re-grettable actions and inactions is likely to become increasinglypronounced over time. After all, one can always add elementsto the list of good things that would have happened if one hadacted. As this list grows, so too does the regret over a failure toact. People tend to idealize many aspects of their distant pasts,and lost opportunities are no exception. The cost of such distor-tion can be rather high, as people can end up tormenting them-selves farbeyond what is called for by the actual state of affairs.There is a second way in which the perceived consequences ofactions and inactions diverge,and it has similar effects. Becauseregret over action is due to the bad things that actually hap-pened as a result of the action, the consequences are tied to aparticular event or decision. Although the consequences on thewhole may be bad, often there is some good that stems fromthe action as well. Thus, when people think of their regrettableactions, they often think of compensatory silver linings, a phe-nomenon that wedocumented earlier. In contrast, many regret-table inactions are not tied as closely to a particular event ordecision (see footnote 1). Often it is not the failure to act in aparticular moment that is regrettable, but one's accumulatedfailure to, say, get closer to one's parents, spend more time withone's children, or more diligently pursue one's career aspira-tions. As a result, inactions are less likely to prompt thoughtsabout compensatory gains. Regrettable omissions, in otherwords, tend to be remembered in the long term only forbeingregrettable, whereas regrettable commissions are more likely tobe remembered for a number of qualitiesqualities that onlyon the wholeare unfortunate.Factors That Promote the Co gnitive Availability ofRegrettable Inactions

W e suspect that for some readers our analysis thus far maybe less than completely satisfying. Some may be experiencingarecurrent and disquieting sense of incompleteness. Good.There is an additional mechanism we have yet to discuss, and,as it happens, it is one that derives from a similarly disquietingstate of psychological incompleteness.

There have been suggestions since at least the 1920s that peo-ple tend to remember incomplete tasks and unrealized goalsbetter than those that have been finished, accomplished, or re-solved. This phenomenon has come to be known as the Zeigar-nik effect, after the person who first tested it empirically(Zeigarnik, 1935; see also Seifert & Patalano, 1991). The orig-inal interpretation of this effect was that the intention to carryout a task generates a state of psychological tension that keepsthe issue alive until the task is complete and the tension is re-

leased (Lewin, 1935,1951).The central idea, however, does notrequire that one accept the Lewinian metaphor of the mind asa tension system. All that is necessary is that one grant thatpeople typically are more preoccupied by and perseverate longerover incomplete tasks than completed ones (Martin &Tesser,1989; Martin, Tesser, & Mclntosh, 1993). The extra mentalwork that is devoted to unfulfilled intentions constitutes addi-tional rehearsal time that makes them more memorable.

There are striking parallels between incomplete tasks and re-grettable omissions. Many regrets over previous failures to actinvolve things we could still do at any time. We might regretnever having learned to speak French, play the violin, or talkwith real intimacy to those whose love and friendship we value,but these are all things we can still do wheneverthe sting ofregret is felt. Our accent may neverbe as good, our ear neverasdeveloped, and our intimate moments not as numerous as theywould have been if we had acted earlier, but much of the grati-fication we imagine such activities to bring could still be ob-tained whenever we choose to start. To be sure, many regretta-ble inactions involve failing to seize a moment that is long past.For these, there is no second chance. The case is closed. Formany other regrettable omissions, however, the opportunityandtemptation to act still exists. Regrettable omissions often belongas much to the present as they do to the past. This sense ofincompleteness and possibility that surrounds our failures toact keeps them alive longer.

This is in marked contrast to many regrettable actions, whichbelong almost entirely to the past. We messed up. We mightworry about it now and vow to do better in the future, but theevent itself lies entirely in the past. The story of regrettable ac-tions tends to be closed; the story of failures to act, open. Be-cause regrettable inactions are more alive, current, and incom-plete than are regrettable actions, we are reminded of themmore often. And a regret that we are reminded of more often isa regret that we experience more often. Thus, this Zeigarnik-like aspect of regrettable omissions may not increase the inten-sity of the emotional pain of failures to act, but it does increasethefrequency with which it is felt.

This idea was tested in a two-stage experiment in whichadults were recruited from an outdoor shopping mall in Ithaca,New \brk, and asked to describe their three biggest regrets ofaction and their three biggest regrets of inaction. Three weekslater, they were contacted by telephone and asked to recall thesix regrets they had mentioned during the first session. As an-ticipated, the participants recalled more of their regrettable in-actions (M = 6 4 % ) than their regrettable actions (M = 39%),t(24) = 3.76,p < .001. This effect held up, furthermore, whenthe severity of each regret was controlled statistically. Holdingthe amount of regret constant, regrets of inaction seem to stayin the mind longer than regrets of action (Savitsky, Medvec, &Gilovich, 1994).

GeneralDiscussionW e reviewed evidence indicating that when people look back

on their lives, it is mainly those things they did not do but wishthey had done that generate the most regret. When combinedwith earlier results from the literature on counterfactual think-ing, these data imply a temporal pattern to the experience ofregret. Commissions generate greater regret initially, but over


14/17

392 THOMAS GILOVICH AND VICTORIA HUSTED MEDVECtime it is omissions that dominate the experience of regret. Wethen reviewed additional evidence supporting the existence ofsuch a temporal pattern. Finally, we introduced a number ofpsychological mechanisms that appear to give rise to this tem-poral pattern and described the results of experiments designedto examine each of them. Some of the mechanisms we identifiedreduce the sting of regrettable action over time, others bolsterthe pain of regrettable inaction, and an additional mechanismdifferentially affects how often one is reminded of these twodifferent sources of regret.Multiple Mechanisms and How They Interrelate

To this point, wehave largely discussed each mechanism sepa-rately, but they are, of course, interrelated in various ways. Oneconnection we havealready noted is that the tendency for peopleto offset the pain of regrettable action by identifying compensa-tory silver linings s but one specific version of a broader tendencyfor regrets of action to induce more dissonance reduction thanregrets of inaction. We have also discussed how the relationshipbetween confidence and temporal distance is connected to therelative invisibility and mild psychological impact of restrainingforces. Whatever inhibited a course of action that now looks at-tractive often seems less and less compelling, and therefore lessand less legitimizing, over time.An important categoryof such inhibitory causes that we haveye t to discuss explicitly is an action's "opportunity costs." Oftenone refrains from doing something because its pursuit wouldforeclose other attractive options. It wouldbe nice to become anaccomplished pianist, but at what cost? Such trade-offs mayseem sufficient to stymie action at the time, but perhaps notwhen reconsidered from a much later vantage point. Over time,one may lose track of what the cost of a particular course ofaction may have been, or the costs maysimply not seem as great.The extent to which people retrospectively downgrade opportu-nity costs remainsan interesting topic for future research.

Another set of important links between the mechanismswehave described involves the factors that affect the magnitude ofregret that is experienced and those that influence the extent towhich a particular regret is thought about and obsessed over.The three mechanisms that diminish the pain of regrettable ac-tionswill, in turn,decrease theircognitiveavailability. Similarly,the three factors that increase the angst of regrettable inactionswill tend to increase their availability.

Perhaps the best wayto examine the similaritiesand differencesamong the various mechanisms we have proposed is to take acloser look at their temporal profile. As a first approximation, itappears that the mechanisms that diminish the regret of actiontend to come into play relatively quickly. Mistaken actions are of-ten corrected immediately, before one is trapped in a new web ofinertia! forces that take extra effort to overcome. Also, the ten-dency for errors of commission to induce greater dissonance re-duction than errors of omission no doubt stems in part from thegreater immediate discomfort that arises from regrettable actions.Indeed, the results of the one dissonance experiment that directlyexamineddissonance reduction after mistakes ofcommission andomission showed how quickly the relevant psychological repairefforts can be mobilized (Gilovich et al., 1995 ).In contrast, the factors that reinforce the pain of regrettableinaction often take longer to develop, as does the tendency for

regrets of inaction to remain more cognitively available thanregrets of action. Differences in cognitive availability obviouslycannot occur until sufficient time has elapsed for memory todecay and "ceiling effects" to disappear. Similarly, the increasein retrospective confidence with temporal distance has beenshown to vary directly with the amount of time that has elapsedbetween the initial moment of opportunity and the subsequentperiod of assessment (Gilovich et al., 1993).

Some readers may be troubled by our invoking so many dis-tinct mechanisms to account for our primary results. All elsebeing equal, it is surely more satisfying when a single mecha-nism explains a given phenomenon. However, for many com-plex social psychological phenomena, the search for a singlecause is likely to be in vain. Indeed, we must confess that wewere initially drawn to this research topic in part because somany mechanisms seemed to be involved. Each of the psycho-logical processes that we identified seemed to us to be intriguingnot just as an explanation of the regret results, but also as apsychological phenomenon in its own right. The relationshipbetween confidence and temporal distance, the amount of dis-sonance reduction devoted to errors of omission and commis-sion, and the relative impact of compelling versus restrainingforces all tell us something potentially significant about the hu-man condition. These phenomena are worthy subjects of inves-tigation quite apart from their role in the experience of regret.Cultural Influences

But we did not undertake this research because of the mecha-nismsalone, ofcourse. Everyone hasregrets, andyet,as wealludedto earlier, little isknown about the underlying psychologyof regret.This is unfortunate because our current cultural and historicalcontext is one that is likely to maximize the experience of regret.In earlier times and in cultures with more rigid behavioral pre-scriptions, there were fewer opportunities for regret. Marriageswere arranged; stations in life were inherited; choices among vari-ous material goods were much narrower. Things could hardly bemore different today.One of the difficulties ofmodern life iscopingwith all of the choices that are available. With a greater range ofchoices comes increased opportunities for regret, and so a relatedproblem with modern life is how to cope with the inevitable andincreasingly common experience ofregret.

This raises the question of whether the observed temporal pat-tern to the experience of regret is largely universalacross cultures.Because it is ourbelief that many of the mechanisms that accountfor this pattern are universal,we would expect to observe somereasonable facsimile ofthese results everywhere. This is not to sug-gest that the findings presented hereboth the greater regret ofaction in the short term and the greater regret of inaction in thelong runare immune to cultural influence. They almost cer-tainly are not. Forexample, Westernsociety seemsto revere actionand disdain inaction (Farley, 1986). The abundance oflong-termregrets of inaction in our sample ofrespondents may be attenuatedin peo

the experience of regret

Documents