Sie sind auf Seite 1von 11

Journal of Personality and Social Psychology 1994, Vol. 67, No.

4, 585-595

Copyright 1994 by the American Psychological Association, Inc.

'If Only I Weren't" Versus "If Only I Hadn't": Distinguishing Shame and Guilt in Counterfactual Thinking
Paula M. Niedenthal, June Price Tangney, and Igor Gavanski
The role of counterfactual thinking in 2 emotionsshame and guiltwas examined. In 1 series of studies, Ss read about situations evocative of shame and guilt or described personal experiences of guilt or shame. They then generated counterfactual alternatives to "undo" the distressing outcomes. Consistent with predictions derived from Tangney (1991), Ss tended to undo shame situations by altering qualities of the self and to undo guilt situations by altering actions. In a 2nd series of studies, Ss imagined themselves in a situation that could evoke either guilt or shame. Ss were then led to mutate the self or behavior to undo the situation. Mutation manipulations amplified shame and guilt such that the former Ss anticipated feeling greater shame, whereas the latter anticipated feeling greater guilt. The role of counterfactual thinking in specific emotions and in differentiating shameand guilt-prone personalities is discussed.

Individuals are naturally drawn to search for the causes of events, particularly unexpected events, that occur in the course of day-to-day life (Kelley, 1972; Weiner, 1985). In the process of this causal search, individuals often engage in counterfactual thinking; they reflect on how past events might have otherwise unfolded had some aspect of the situation or their behavior been different. The mental simulation of alternative outcomes appears to guide individuals' attributions about the causes of actual outcomes (Gavanski & Wells, 1989; Lipe, 1991; Roese & Olson, in press; Wells & Gavanski, 1989). Counterfactual thinking is also a determinant of individuals' affective reactions to events. Some past research has provided evidence to suggest that the consideration of near outcomes events that almost happened, but did notcan influence satisfaction with an actual outcome (Johnson, 1986; Kahneman & Miller, 1986; Markman, Gavanski, Sherman, & McMullen, 1993; Roese, 1994). In this view, counterfactuals are alternatives against which reality is compared. Thus, a person might be quite satisfied with having won a small prize, but unsatisfied if she perceives herself as having just missed winning a much larger prize. Other work has linked counterfactual thinking to
Paula M. Niedenthal and Igor Gavanski, Department of Psychology, Indiana University; June Price Tangney, Department of Psychology, George Mason University. This research was supported by Grant MH44811 -01A1 from the National Institute of Mental Health and Grants BNS-8919755 and DBS291019 from the National Science Foundation to Paula M. Niedenthal and Grant RR07031-27 from the Biomedical Research Support Grant Program to Igor Gavanski. We are grateful to Denise Beike, Russell Fazio, Tom Gilovich, Jamin Halberstadt, Jim Sherman, and Carolin Showers for their helpful comments on a draft of this article and to Progna Choudhury, Chukwudube Egbuniwe, Darcie Dimitroff, Kelly Mann, Marc Setterlund, and Priya Singh for their competent technical assistance. Correspondence concerning this article should be addressed to Paula M. Niedenthal, Department of Psychology, Indiana University, Bloomington, Indiana 47405, or to June Price Tangney, Department of Psychology, George Mason University, Fairfax, Virginia 22030. 585

the intensity of an affective reaction to an outcome such that the ease of imagining an alternative to reality is positively related to the intensity of the reaction to an actual outcome (Gleicher et al., 1990; Kahneman &Tversky, 1982;Landman, 1987; Roese, 1994). In this research we extend the idea that counterfactual thinking mediates affect and propose that, through its role in assessments of causation, counterfactual thinking helps to shape the specific emotions an individual experiences in reaction to a situation. Of interest are two emotions that have the same (negative) valence, but different phenomenologies: shame and guilt. Although rarely mentioned by students of counterfactual thinking, shame and guilt are, in fact, noted by clinicians to be associated with obsessive mental undoing of a past emotional situation. Research by Tangney (1989a, 1992) suggested to us that the two emotions involve different types of counterfactual thinking and that the counterfactual alternatives that people consider in distressing situations may serve to produce or at least amplify those specific feelings. Next, we discuss the processes by which such effects may occur. Counterfactual Thinking and Causal Attribution Consistent with the ideas of Kahneman and Miller (1986) and Einhorn and Hogarth (1986), Wells and Gavanski (1989) reported empirical evidence demonstrating that judgments of causation are guided by a comparison between reality and knowledge of, or belief about, what might have been (counterfactual alternatives). The research was based on the assumptions that an event will be judged as causal of an outcome to the extent that it is mutable (i.e., the event can be psychologically altered), and the mental mutation of the event to a different value can logically change or undo the outcome. In two experiments, Wells and Gavanski (1989) manipulated the salient ("default") counterfactual alternatives to events resulting in negative outcomes. The counterfactual alternatives either did or did not alter the actual outcomes. For instance, in Experiment 1, subjects read a story in which a woman named

586

P. NIEDENTHAL, J. TANGNEY, AND I. GAVANSKI

Karen is taken out to dinner by her boss. The boss considers ordering two different entrees for Karen, ultimately choosing in favor of a dish containing wine. Unbeknownst to the boss, Karen has an allergy to wine, and she dies from an allergic reaction to the meal. In one version of the story, one of the two dishes that the boss considers ordering for Karen does not contain wine (the one-wine version). In another version, both of the entrees under consideration contain wine (the two-wine version). Thus, in the one-wine version the most available alternative event would have resulted in a different, and better, outcome (Karen living), whereas in the two-wine version, the available counterfactual alternative would have resulted in the same outcome (Karen dying). After reading one of the two versions of the scenario, subjects wrote counterfactual statements to "undo" Karen's death and listed the causes of Karen's death. Compared with subjects who read the two-wine scenario, subjects who read the one-wine scenario were more likely to mutate the boss's choice of entrees in mentally undoing Karen's death and to blame the boss for the death. Subjects who read the two-wine scenario mutated other factors such as Karen's choosing to eat the dish and blamed other agents such as Karen's behavior or her allergies. This result nicely illustrates the role of counterfactual alternatives in causal attribution: When the most salient counterfactual (default) alternative resulted in Karen living (in the one-wine scenario), subjects found the boss responsible for Karen's death. When the most salient alternative still produced Karen's death, the boss's behavior seemed less causal. As Wells and Gavanski (1989) pointed out, this result cannot be easily explained by traditional attribution theories in which only the features of the actual event are evaluated in judgments of causality (Lipe, 1991). Causal Attribution and Emotion Cognitive appraisal theories of emotion hold that specific emotions are shaped by individuals' interpretations of evocative situations (e.g., Arnold, 1960; Fridja, Kuipers, & ter Schure, 1989;Roseman, 1984; Smith & Ellsworth, 1985;Scherer, 1988; Weiner, 1985). Empirical research in this area has focused primarily on identifying the appraisal dimensions that account for the greatest variance in emotional experience. Specific emotions are thought to correspond to distinct profiles of values on the appraisal dimensions. Although the critical appraisal dimensions vary slightly in name and number from theory to theory, nearly all of the major empirical undertakings have revealed that causal attribution dimensions play a prominent role in shaping affective reactions. For instance, dimensions such as responsibility (i.e., the extent to which self, other, and environment are responsible for instigating the situation) and control (i.e., the extent to which the self, other, and environment influence the course and outcome of the situation) characterize thefindingsof several studies. In Weiner's (1985; Weiner & Graham, 1984) view, the extent to which an outcome is attributed to internal-external, stable-unstable, and controllable-uncontrollable factors allfigureimportantly into the causal attribution and, consequently, the emotional reaction to the outcome. Cognitive appraisal theories of emotion thus predict that the

same outcomes can evoke very different emotional reactions depending on an individual's beliefs about the causes of the outcome. There exists good support for this idea (e.g., Smith & Ellsworth, 1987; Weiner, 1980). For instance, in a study by Weiner (1980), individuals ("lenders") responded with anger to a request by another person ("borrower") to borrow their class notes when the need for the notes arose because the borrower went to the beach rather than to class on a particular day. On the other hand, lenders responded with pity to the same request when it arose because the borrower suffered from an eye problem. The present work shares with the appraisal theories the assumption that attributions of causes of positive and negative outcomes can influence individuals' specific emotional reactions to those outcomes. However, it shares somewhat more with Abelson's (1983) view that an analysis of the counterfactual ideas that individuals entertain following such outcomes may go a long way in predicting specific emotional reactions: Counterfactual thoughts provide both hints to causes of outcomes and provide anchors against which individuals perceive discrepancies between reality and possibility. Such discrepancies (or "inconsistencies") importantly determine the instigation of emotion itself (Carver & Scheier, 1981; Higgins, 1989). Thus, we believe that specific emotions can arise "whenever alternative construals of events with sharply different hedonic imports for an individual are concurrently mentally exercised" (Abelson, 1983, p. 46).

Shame and Guilt


Of particular interest in the present work were the emotions of shame and guilt. Shame and guilt are both dysphoric feelings that involve negative self-evaluations. In a recent study of the appraisals and action tendencies associated with different emotions, shame and guilt, along with regret, were uniquely associated with endorsement of an appraisal item that read "you wished you had not done something or that you could undo it," suggesting to us that spontaneous counterfactual thinking plays an important role in shaping these feelings in particular (Fridja et al., 1989). However, cognitive appraisal theorists sometimes describe shame and guilt as very similar emotions that both involve perceiving the self as the causal agent (e.g., Fridja et al., 1989; Roseman, 1984; Smith & Ellsworth, 1985). This description is probably due to the lack of precision in defining appraisal dimensions (cf. Brown & Weiner, 1984). Investigations that have specifically compared shame and guilt find that the emotions focus on quite different concerns about the self that result in distinct phenomenological experiences (Lindsay-Hartz, 1984; Tangney, 1989a, 1993; Wicker, Payne, & Morgan, 1983). Shame involves an evaluation of the self. Although a specific failure or transgression may trigger a shame reaction, the implications of the event are attributed to the self. When ashamed, people feel as if they are a "bad person" and that the self has been humiliated or disgraced. Shame is an overwhelming feeling characterized by a sense of being "small" and worthless in the eyes of both the self and others. With this feeling of shame comes a desire to hide or escape from the situ-

SHAME AND GUILT

587

ation (Lewis, 1971; Lindsay-Hartz, 1984; Tangney, 1989a, 1991, 1993; Wicker, Payne, & Morgan, 1983). In contrast, guilt involves a concern with a specific behavior or transgression. When guilty, people are consumed with the idea that they did a "bad thing" (or failed to do a good thing). Guilt is characterized by feelings of tension, remorse, and regret over the bad thing that was done. But the cognitive-affective processes in guilt stop short of the self-condemnation involved in shame. Not surprisingly, with this guilt over specific behaviors comes a concern with reparative action (Lewis, 1971; Lindsay-Hartz, 1984; Tangney, 1989a, 1993). Interestingly, although shame and guilt involve distinct phenomenologies, the situations that elicit shame and guilt are not necessarily distinctive. Studies of narrative accounts of real-life shame and guilt experiences reported by children and adults (Tangney, 1992; Tangney, Marschall, Rosenberg, Barlow, & Wagner, 1993) reveal that there is substantial overlap in the types of events and behaviors that give rise to the two emotions.' In fact, there appear to be few, if any, "classic" shame-inducing or guilt-inducing situations; most situations seem to be met with the two emotions equally often across individuals. For example, whereas 21% of all guilt-eliciting situations nominated by a sample of young adults referred to an experience in which they had lied, 11% of all shame situations nominated also referred to lying (Tangney, 1992). Similarly, 17% of the participants in the same study nominated hurting another person emotionally as guilt eliciting, and 9% nominated the same type of behavior as shame eliciting. Counterfactual Thinking, Shame, and Guilt What is it, then, that determines whether people feel shame or guilt in a distressing situation if not the nature of the situation itself? Clearly, some people habitually respond to ambiguous distressing situations by feeling ashamed, whereas others respond by feeling guilty (Tangney, 1992). But what are the cognitive processes that mediate these reactions? We believe that the tendency to spontaneously undo distressing situations such as those in which one has lied or cheated (Fridja et al., 1989), importantly mediates the quality of the emotional reaction. Specifically, the consideration of alternative scenarios may guide the assessment of the cause of the actual situation (self vs. behavior), which in turn leads to the specific emotional experience (shame vs. guilt). For a variety of reasons, an individual might perceive the possession of a different personal quality (i.e., something inherent to the self) as the default counterfactual alternative for a particular situation. In the same situation, a different individual might perceive an alternative course of action to be the default value. These counterfactual alternatives may amplify the resulting emotions by emphasizing a discrepancy between reality and a more desirable state of affairs, and by locating the cause of the distressing situation in the attributes of the self and in behavior, respectively. To illustrate by way of example, recall that having lied to someone is variously associated with shame and guilt reactions across different individuals (Tangney, 1992). Suppose two people, Bruce and Frank, both tell a lie, and the lie results in a negative outcome. Bruce, for reasons of focus of attention, or developmental history, might think "if only I were stronger, I

would have told the truth." This counterfactual alternative would emphasize the self as bad (i.e., weak) and a change in the self as necessary to undo the outcome. If qualities of the self are spontaneously mutated in order to undo the situation, then something about the self has caused the bad situation. Bruce is therefore likely to experience shame. On the other hand, Frank might reflect on the fact that he could have told the truth, and the bad outcome would have been avoided. The counterfactual alternative "if only I had told the truth" emphasizes a single behavior (i.e., the one lie) as the cause of the bad outcome. Frank, then, is more likely to experience guilt. Overview of the Current Studies In thefirstseries of studies reported, we investigated the proposed differences in the focus of counterfactual alternatives that people consider in situations in which they report feeling shame and guilt. College students imagined themselves in (Studies la and lb) or recalled (Study lc) experiences that elicited shame and guilt and were instructed to mentally undo the experiences. The resulting counterfactual statements were coded according to whether the self, behavior, or other aspects of the situations were mutated. We expected situations in which individuals report feeling shame to be accompanied by counterfactual thinking involving mutations to the self; we expected situations in which individuals report feeling guilty to be accompanied by counterfactual thinking about alternative courses of action. In a second series of studies we tested the implied causal (or amplifying) relationship between counterfactual thinking and feelings of shame and guilt. Subjects in Study 2a imagined themselves in a distressing situation that college students find very plausible and familiar. They then completed counterfactual statements that directed them to mutate the self or to mutate their actions in order to undo the situation. Finally, subjects rated how much shame or guilt they would experience in the situation. We predicted that subjects who generated counterfactual alternatives about the self would report greater shame and that those who mutated their actions would report greater guilt about the situation. In Study 2b subjects imagined themselves in the same situation used in Study 2a. They then mutated the self or behavior or were not asked to engage in any counterfactual thinking at all. We expected that, compared with those sub' One question that arises is whether participants could actually follow instructions to identify and report on shame and guilt experiences. As noted earlier, the terms shame and guilt are often used interchangeably by laymen and psychologists alike. In an earlier study, we asked college students to provide definitions of shame and guilt (Tangney, 1989b). Results indicated that even these verbal, well-educated young adults had a great deal of difficulty defining and distinguishing between shame and guilt in the abstract. However, when people are asked to describe personal shame and guilt experiences (without any further definition of these emotion terms), they clearly describe different experiences. In several studies involving children, college students, and noncollege adults (Tangney, 1989a, 1992; Tangney, Marschall, etal., 1993), such subject-generated narrative accounts of personal shame and guilt experiences varied systematically in a theoretically consistent manner along a broad range of phenomenological dimensions. Thus, this seems to be a case where participants "know more than they can say," at least in the abstract context.

588

P. NIEDENTHAL, J. TANGNEY, AND I. GAVANSKI actually. All you have to do is eat their food, collect the mail, and feed their bird. Everything is going just fine until one morning you discover that the bird died during the night. You turned the airconditioning on "high" during the day and forgot to turn it down at night as you had been instructed. The bird died from the excessive cold, [bird scenario] Pilot subjects rated the bird scenario 5.70 on the guilt scale and 4.62 on the shame scale (p < .01). Procedure. Participants in Study la completed a questionnaire in which they were asked to vividly imagine each of the two situations (professor scenario and bird scenario) actually happening to them. The order of the scenarios was counterbalanced. Subjects were informed that the experimenter was not interested in whether they thought the experiences could actually happen but just in their thoughts about the experiences if they were to happen. After reading each scenario, subjects were posed with the following question: "What could have been different about you, your behavior, or any aspect of the situation so that the situation would have ended differently?" The phrase "If only . . ." then appeared, with spaces to write four responses beneath. The instructions for counterfactual thinking used in Study la were quite directive, and subjects may have thought that they were required to change all aspects (self, behavior, and situation) in their counterfactual responses. Although all subjects did receive the same instructions, the strong focus on distinct counterfactual categories potentially limited our discovery of what people would think without such direct instruction. We therefore conducted Study 1 b using the same scenarios. There was a single change made to the questionnaire: Following each of the scenarios, subjects were asked to list four ways in which the scenario could have unfolded differently so that the negative outcome would not have occurred. There was no reference to self, behavior, or situation. Again, the phrase "If only . . ." appeared with spaces to write four responses beneath. Coding. Counterfactual statements (for both studies) were coded by two independent judges, one of whom was unaware of the experimental hypotheses, according to the following categories (examples in italics come from the bird scenario; the rest come from the professor scenario): 1. Self-chronic. This category included mutations of enduring aspects of the self. Included were mentions of (a) personality traits (weren't impulsive, weren't so self-centered), (b) attitudes/preferences (didn't always care what others think), and (c) intelligence (weren't so dumb). 2. Self-acute. These were mentions of transient qualities of self including the following: (d) personal states (hadn't been so impulsive, hadn 't been absent-minded), (b) self-presentation (hadn't tried to impress), (c) knowledge/insight (considered the needs of the bird), and (d) effort/care (e.g., thought more carefully, paid more attention). 3. Behavior. The second category consisted of mutations of specific behaviors or failures to act. Included were (a) thoughts (had thought of the right answer, had thought to turn down the AC), and (b) actions (hadn't raised my hand, had read the material, hadn't come to class, covered the bird cage, turned down the AC, had moved the bird cage). 4. Situation. The last category consisted of statements that mutated the situation, with a focus on (a) the self (I wasn't in class that day; / wasn 't house-sitting), (b) other people (the professor had called on someone else, the owners came home early), and (c) other agents (it had happened in a different class, the bird hadn't been so sensitive to cold). The coding procedure proved to be very reliable. The two judges agreed on 83% of the specific categorizations for Study la. Discrepancies were resolved by discussion.

jects who engaged in no counterfactual thinking, directed mutations of the self would yield greater shame and directed mutations of the behavior would yield greater guilt. In this research we explicitly solicited (Series 1) and guided (Series 2) counterfactual thinking because our interest was in how the content of counterfactual thinking mediates specific emotions. Other work has shown that people spontaneously generate counterfactual thinking when they experience negative outcomes (Gleicher et al., 1990) and even when they experience positive outcomes (Markman et al., 1993). And, as mentioned, we know shame and guilt to be associated with a concern with undoing the outcome of the precipitating experience (Fridja et al., 1989). Series 1: Counterfactual Alternatives in Shame and Guilt Situations

Studies la and lb Method


Participants. One hundred undergraduates in introductory psychology courses at Johns Hopkins University volunteered their efforts for Study lain return for partial credit toward a course requirement. Approximately equal numbers of female and male students participated. Forty-five undergraduates from the same population participated in Study lb. Scenarios. As noted earlier, the types of situations that elicit shame and guilt are generally very similar (Tangney, 1992; Tangney, Marschall, et al., 1993). For this study, however, we attempted to identify a pair of situations thatin content and emphasiswould be more likely to elicit feelings of shame and guilt, respectively. Eight scenarios were initially constructed by examining the classes of situations that young adults nominate as being shame and guilt eliciting (Tangney, 1992) and by discussing specific situations with a small group of college undergraduates. On the basis of these sources of information, we wrote four scenarios that were thought, a priori, to elicit greater shame than guilt and four that were thought to elicit greater guilt than shame. The situations were also thought by the undergraduates who helped generate them to be easy to imagine and likely to happen to almost any college student. In a pilot test, approximately 150 Johns Hopkins undergraduates read each of the scenarios and rated them on two 7-point scales according to how much shame and guilt they would feel in the situation were they to actually experience it. A 1 meant that the situation would elicit no shame (guilt) and a 7 indicated that it would elicit a lot of shame (guilt). Two scenarios, one that subjects rated on average as likely to elicit more shame than guilt and one that was rated as likely to elicit more guilt than shame, were selected for use in Studies la and lb. The shame scenario involved failure in a school setting and read as follows: You are sitting in a class of about 15 students that is taught by a professor you admire more than any other on campus. You really believe that he/she is brilliant. The professor asks a question about one of the readings/assignments and you immediately volunteer the answer with some enthusiasm. But it's the wrong answer, the professor informs you somewhat dryly, and he/she turns to address another student in the class, [professor scenario] Pilot subjects rated the professor scenario 4.82 on the shame scale and 2.51 on the guilt scale (p < .01). The guilt scenario involved unintentional damaging of another's possession and read: You are house-sitting for a friend's parents. It's a pretty easy task,

Results ofStudy la
Four participants did not provide at least one codable counterfactual for each scenario. The data from these questionnaires

SHAME AND GUILT Table 1 First and Total Counter/actuals Reported by Category: Study la Scenario First counterfactual Professor scenario Bird scenario Total counterfactuals Professor scenario Bird scenario Self-chronic 3(3%) 2(2%) 31(9%) 14(4%) Self-acute 44 (46%) 29 (30%) 128(37%) 66 (20%) Behavior 31(32%) 53 (55%) 96 (28%) 147 (43%) Situation 18(19%) 12(13%) 93 (27%) 112(33%) Total 96 96 348 339

589

were not analyzed. Thus, the results are based on the remaining 96 subjects. Table 1 presents the frequency and percentages of selfchronic, self-acute, behavior, and situation counterfactuals for each scenario and includes both the first counterfactual generated by subjects and the total number of counterfactuals nominated by subjects. The pattern of results, however, was the same in each case: Compared with imagining themselves in the guilt situation, after imagining themselves in the shame situation, subjects were more likely to mutate aspects of the self, particularly transient aspects. Subjects mutated their own behavior in order to undo the guilt experience far more than they mutated their behavior to undo the shame experience. For statistical analysis, mutations of chronic aspects of self were coded - 1 , mutations of acute aspects of self were coded 0, and mutations of behavior were coded 1 to reflect a continuum from mutations of the self to mutations of actions. Mutations of the situation were eliminated from the analysis (and this resulted in variability of degrees of freedom for the analyses, below). We first examined the nature of the first counterfactuals generated; these are thought to be the most accessible beliefs about alternatives to reality. A t test showed that after imagining themselves in the bird scenario, subjects tended to mutate aspects of their behavior (M = .607) more than they did after imagining themselves in the professor scenario (M = .359), We also analyzed the data by averaging the codes for all four counterfactuals. These analyses also revealed that on average subjects mutated behavior more often when they tried to undo the bird scenario (M = .597) than when they tried to undo the professor scenario {M = .270), t{91) = 6.03, p < .001. Finally, the analyses were repeated using only the codes originally (i.e., prior to discussion) generated by the coder who was uninformed about the experimental hypotheses. This was done to address the possibility that the results were due to the knowledge of the informed coder. The pattern of results was unchanged: Codes from the naive coder showed that subjects were more likely to mutate behavior in undoing the bird situation (first counterfactual M = .595; all counterfactuals M = .571) than they were in undoing the professor situation (first counterfactual M = .410; all counterfactuals M = .302), ts = 2.20 and 5.24, ps < .05 and < .001, respectively.

Results of Study lb
Three subjects did not follow instructions, and their questionnaires were discarded. The same two judges coded the

counterfactuals generated by the remaining 42 subjects using the previously described coding scheme. This time they categorized the items with 94% agreement. Disagreements were resolved by discussion. Table 2 shows that the change in instructions had virtually no effect on the pattern of results. Even when participants were not directed to mutate aspects of the self, behavior, or situation, they did so, and in a manner that replicated the results of our previous study. Mutations were coded as in Study la. Tests of the first counterfactual showed that on average subjects mutated their behavior to undo the bird scenario (M = .722) more than they did to undo the professor scenario {M = .432), t(30) = 2.90, p < .05. Examination of the average codes for all four counterfactuals showed the same result (M = .770 for the bird scenario .250 for the professor scenario), <(40) = 6.84, p < .001. Once again, it appeared as if subjects were more likely to mutate specific actions rather than aspects of themselves. Studies la and lb lend consistent support for the notion that feelings of shame and guilt are associated with quite different types of counterfactual thinking. In contrasting the counterfactual statements that were freely generated by participants in response to two scenariosone scenario that pilot testing indicated was more likely to engender shame, and the other more likely to engender guilta reliable pattern of results was observed. The scenario likely to induce feelings of shame was followed by a higher proportion of counterfactual mutations of the self, compared with the mutations of the guilt-inducing scenario. In contrast, the scenario likely to induce substantial feelings of guilt was followed by a higher proportion of mutations of specific behaviors, compared with the mutations of the shameinducing scenario. It is possible, however, that something about the scenarios used for the first two studies, other than the fact that pilot subjects indicated that the situations described would be met with greater shame or guilt, accounted for the results we observed. For instance, although Tables 1 and 2 show that subjects generated all three types of counterfactual alternatives (self, behavior, and situation) for each scenario, features of the self may have been more salient and more mutable in the professor scenario. Likewise, specific behaviors done or left undone may have been more salient in the bird scenario. Therefore, in a third study in this series we examined young adults' counterfactual mutations to personal, self-generated shame and guilt experiences recalled from the recent past. This procedure allowed us to consider a broad range of situations and content areas associated with

590
Table 2

P. NIEDENTHAL, J. TANGNEY, AND I. GAVANSKI

First and Total Counterfactuals Reported by Category: Study lb


Scenario First counterfactual Professor scenario Bird scenario Total counterfactuals Professor scenario Bird scenario Self-chronic 2(5%) 1 (2%) 19(13%) 3 (2%) Self-acute 17(40%) 8(19%) 52 (37%) 18(12%) Behavior 18(43%) 27 (65%) 30(21%) 79(51%) Situation 5(12%) 6(14%) 41(29%) 55 (35%) Total 42 42 142 155

shame and guilt rather than one example of each. In addition, such autobiographical accounts move our consideration beyond lab-generated hypothetical events to college students' naturally occurring self-relevant emotional experiences.

Study 1c Method
Participants. Subjects were 120 undergraduates at Johns Hopkins University and Indiana University who participated in return for credit toward an undergraduate psychology course. Procedure. Participants completed questionnaires that instructed them to describe a situation that had really happened to them. Half of the subjects described an event that caused them to feel shame. The remaining subjects described an event that caused them to feel guilt. Subjects also rated the situations according to how much shame or guilt they had felt in the situation on a 7-point scale. Finally, subjects listed four things that might have been different so that the situation would not have ended the way it did. Again, this portion of the task was introduced with the sentence stem "If only. . ." Sixteen of the subjects either did not describe an actual event or did not provide at least one codable counterfactual statement, so their questionnaires were discarded. Two new judges, unaware of the experimental hypothesis, read the remaining 104 scenarios and coded the counterfactual thoughts according to the same coding scheme used in Studies la and lb. The judges agreed 83% of the time, and disagreements were resolved through discussion. Example scenarios, counterfactuals, and codes are presented in the Appendix.

Results ofStudy 1c
As directed, subjects tended to report very emotional episodes. Of those subjects who described situations in which they had felt shame, 81% rated their feelings in the situation as a 5, 6, or 7 on the 7-point scale of shame. Similarly, of the subjects who described situations in which they had felt guilt, 73% rated their feelings in the situation as a 5,6, or 7 on the 7-point scale of guilt. Table 3 presents the frequencies and percentages of self-

chronic, self-acute, behavior, and situation counterfactuals for the subject-generated shame and guilt scenarios. Again, we report the frequencies and percentages of classifications of the first counterfactual nominated by each subject as well as the total number of counterfactuals nominated by subjects. As in thefirsttwo studies, for statistical analysis mutations of chronic aspects of self were coded - 1 , mutations of acute aspects of self were coded 0, and mutations of behavior were coded 1. In examination of the first counterfactuals it appeared that regardless of the type of situation they were mentally undoing, subjects were somewhat more inclined to mutate their actions. However, a t test revealed that subjects who described guilt experiences were marginally more likely to mutate their behavior (M = .738) than were subjects who described shame experiences (M = .500), /(86) = 1.78, p < .08. Analysis of all four counterfactuals generated yielded a stronger result (Ms were .640 and .385, respectively), t(l00) = 2.44,p < .05. Consistent with thefindingsfrom Studies la and lb, participants describing personal guilt experiences were more likely to mutate their own behaviors compared with those generating counterfactual statements in response to personal shame experiences. Thus, the relationship between counterfactual thinking and emotion does not appear to be due simply to idiosyncratic features of the bird and professor scenarios used in the previous studies. One other descriptive result that deserves some note is the difference in the raw number of self and behavior mutations. The data from all three studies reported thus far (Tables 1, 2, and 3) suggest that people are more likely to mutate behavior than aspects of the self. This makes sense because specific actions should be, on average, more psychologically mutable than aspects of the self. If our present argument about the role of counterfactual thinking in shame and guilt is correct, then guilt should be a more commonly experienced emotion than shame. We are not aware of data on the relative frequency of guilt and shame. However, the prevalence of means to admit guilt and be

Table 3 First and Total Counterfactuals Reported by Category: Study lc


Experience First counterfactual Shame experience Guilt experience Total counterfactuals Shame experience Guilt experience Self-chronic 4 (8%) 3 (6%) 20(10%) 10(5%) Self-acute 15(28%) 5 (9%) 50 (26%) 27(14%) Behavior 27(51%) 34 (67%) 77 (40%) 95 (50%) Situation 7(13%) 9(18%) 45 (24%) 59(31%) Total 53 51 192 191

SHAME AND GUILT

591

forgiven of sins in organized religion might be some indication that guilt is a very common experience. Series 2: Causal Influences of Counterfactual Thought in Shame and Guilt Together, thefirstthree studies provide consistent support for the idea that when people engage in counterfactual thinking about situations that could cause, or did cause, them shame and guilt, they focus on different events to mentally undo the outcomes. Shame experiences were associated with the tendency to undo aspects of the self. Guilt experiences were associated with the tendency to undo specific behaviors. So far, therefore, we have provided some evidence for the idea that counterfactual thinking about situations evocative of shame and guilt involve a different focus on the self. Of course, it is unclear from thefindingsof thefirstthree studies just how counterfactual thinking is related to these emotions. If shame and guilt involve a focus on the self and on behavior, respectively, then it should not be surprising that the content of counterfactual thinking should be different. However, we argued initially that counterfactual thinking may also serve to produce or amplify feelings of shame and guilt by highlighting the self or behavior, respectively, as the cause of the distressing situation. Studies 2a and 2b were designed to test the direction of the effect. Subjects in Study 2a imagined themselves in a situation that could elicit either shame or guilt, as determined by pilot testing. Half of the subjects were then directed to undo the outcome of the situation by mutating aspects of the self, whereas the remaining subjects were led to undo the outcome of the situation by mutating their behavior. All subjects then reported how much shame or guilt they would feel in the situation. We predicted that subjects who mutated the self would feel stronger shame than guilt and that subjects who mutated their behavior would feel stronger guilt than shame.

tion happening to them. After reading and imagining the scenario, participants were instructed to engage in one of two types of counterfactual thinking. Half of the participants received instructions designed to induce them to focus on counterfactual alternatives involving the self. The instructions read as follows: Now complete the sentence stem "If only I were (not). . "Complete the stem twice. Both times, describe how if you were or were not a particular TYPE OF PERSON (e.g., with particular personality traits, emotional style, intelligence) you wouldn't have gotten into this situation in thefirstplace. Thus your sentence should take the form: "If only I were (were not) [a particular type of person]." The remaining participants were instructed to imagine alternative behaviors and to complete the stem "If only I had (not). . ."They were told to describe specific behaviors that they could have taken or not taken so that they would not have gotten into the situation. Thus, their sentences were to take the form "If only I had (had not) [done something]." Half of the participants in each mutation condition then rated the amount of shame they would anticipate feeling in the flirting situation, using a 7-point scale where 1 indicated they would feel no shame and a 7 indicated they would feel extreme shame. The remaining participants in each condition rated the amount of guilt they would feel,'using a 7point scale where 1 indicated no guilt and 7 indicated extreme guilt. The decision to conduct a between-subjects collection of emotion ratings was guided by pilot work indicating that thefirstrating of an emotion (shame or guilt) tended to determine the second rating such that there was very little within-subjects variance on ratings of shame and guilt. This experiment thus involved a 2 (mutate: self vs. behavior) X 2 (feelings rated: shame vs. guilt) fully between-subjects design.

Results ofStudy 2a
Participants had no trouble imagining themselves in the flirting situation or following the mutation instructions. Comments written on questionnaires as well as spontaneous comments to the experimenters indicated that the scenario was both plausible and troubling. Example responses from participants who completed the stem "If only I were (not) . . ." included "smart enough to avoid a situation in which no matter what I do someone might be hurt by my actions," "a disciplined, trustworthy friend," "not such a backstabber," "not the kind of person who loved attention," "not so selfish," "stronger," and "not such aflirt."In completing this stem subjects sometimes listed acute aspects of the self, which reinforces the validity of the coding procedure for the first series of studies. Such mutations included "were not attracted to Chris," "were more attentive to the situation," "were not so desperate for someone to go out with," "were able to tell my friend that I was interested in Chris too," and "more sensitive to the feelings of my friend." Typical completions of the stem "If only I had (not) . . ." included "not given Chris my phone number," "notflirtedwith Chris," "refused to go on this damn date in the first place," "given them some time alone to get to know each other," and "brought a date for myself to distract me." The shame and guilt ratings were subjected to a 2 (mutate) X 2 (feelings rated) analysis of variance (ANOYA). As can be seen by inspecting the means in Table 4, there was a significant main effect of mutate condition, F(\, 104) = 4.89, p < .05: Subjects who mutated the self reported higher levels of bad feelings (both guilt and shame) than did subjects who mutated their behavior.

Study 2a Method
Participants. Participants were 108 undergraduates at Indiana University, who were participating individually in an unrelated experiment. For their efforts they received partial credit toward an undergraduate psychology course. Approximately equal numbers of male and female students participated. Scenario. On the basis of the pilot testing of scenarios for Studies la and lb, we chose to use a scenario that possessed the following characteristics: (a) The situation elicited equal amounts of shame and guilt (Ms = approximately 46), (b) the situation could happen naturally to both male and female participants, and (c) the situation was a relatively high-probability event in a college setting. The selected scenario read as follows: Your good friend, who rarely dates, invites you to attend a party with him/her and his/her date, Chris. It is your friend's first date with Chris. \bu go along and discover that Chris is not only very attractive, but is also flirting with you. Youflirtback. Although you are not seriously interested in him/her, at the end of the night you give Chris your phone number. The next day your good friend raves to you about how much he/she liked Chris, [flirting scenario] Procedure. Participants completed a questionnaire that included the flirting scenario. They were prompted to vividly imagine the situa-

592

P. NIEDENTHAL, J. TANGNEY, AND I. GAVANSKI

This effect was, however, qualified by the predicted interaction between mutate condition and type of feelings rated, F(l, 104) = 4.44, p < .05. The interaction indicates that shame ratings were higher than guilt ratings in the mutate self condition and guilt ratings were higher than shame ratings in the mutate behavior condition. Thus, our predictions were supported: Selfmutation led to greater anticipated shame, and behavior mutation led to greater anticipated guilt.

Table 5 Mean Shame and Guilt Ratings From Subjects in Mutate Self, Mutate Behavior, and No Mutate Conditions: Study 2b
Emotion rated Shame Guilt Mutate self 5.52 (n = 48) 5.13 (n = 38) No mutate 4.77( = 47) 5.56 (n = 52) Mutate behavior 4.87( = 54) 5.60 (n = 50)

Study 2b Method
As described in the introduction, Wells and Gavanski (1989) found that individuals tend to rate an event as more causal of an outcome if the most salient counterfactual event would have yielded a different outcome. Importantly, in their study this effect was enhanced when subjects engaged in counterfactual thinking before (versus after) specifying the causes of the outcome in question. The present Study 2b was a replication of Study 2a except that one third of the subjects did not explicitly engage in counterfactual thinking at all before rating the amount of shame or guilt they would feel were they to experience the flirting situation (the remaining subjects were directed, as before, to either mutate the self or mutate their behavior). We expected to replicate the result of Study 2a such that subjects directed to mutate the self would anticipate feeling greater shame than guilt, and subjects directed to mutate their behavior would anticipate feeling greater guilt than shame. Using Wells and Gavanski's logic, we also expected that the former subjects would anticipate more shame than subjects in the no-mutate condition, and that the latter subjects would anticipate feeling more guilt than subjects in the no-mutate condition. The experiment entailed a 2 (feelings rated: shame vs. guilt) X 3 (mutate: self vs. behavior vs. no mutation) fully between-subjects design. Participants were 305 undergraduates at Indiana University who participated individually in a longer, unrelated experiment for course credit. Sixteen subjects failed to complete the questionnaire or follow instructions and their data were discarded.

Note. Shame and guilt were rated on scales from 1 to 7 where 1 meant that subjects would feel that emotion not at all and 7 meant that subjects would feel the emotion very much.

subjects thought they would feel more guilt than shame in the flirting situation. In addition, the two-way interaction was significant, F(2, 283) = 4.35, p = .01, indicating that ratings of guilt and shame differed across mutation conditions. Separate ANOVAs were therefore conducted on shame and guilt ratings. The analysis of shame revealed a main effect of mutation condition, F(2, 146) = 3.13, p < .05. Subjects who mutated the self reported more shame than did subjects in the mutate behavior condition, t(93) = 2A6,p< .05, and than did subjects who were not asked to engage in counterfactual thinking at all, /(100) = 2.39, p < .05. The analysis of guilt ratings did not reveal a main effect. Subjects who mutated the self reported marginally less guilt than the subjects who mutated their behavior, ?(88) = 1.58, p < . 12, and than subjects who did not engage in counterfactual thinking, ;(86) = 1.34, p < . 19. In summary, the presentfindingsreplicate those of Study 2a. Subjects directed to mutate the self felt more shame than guilt, and subjects directed to mutate behavior felt more guilt than shame after imagining themselves in theflirtingscenario. In addition, as predicted, subjects who were directed to mutate aspects of the self to undo the situation reported much higher levels of anticipated shame than did subjects who did not explicitly engage in counterfactual thinking before rating their feelings. Results ofStudy 2b Thus, mutations of the self had a clear amplifying effect on feelings of shame. Subjects who mutated the self reported slightly Means are reported in Table 5. Analysis of the data from subless guilt than did subjects in the other two conditions, but those jects in the two mutate conditions revealed the same Mutate X who mutated behavior did not anticipate higher guilt than those Feelings Rated interaction observed in Study 2a, F(l, 181) = in the no-mutate condition. The failure to observe an amplifi6.99, p < .01. This indicates that subjects in the mutate self cation of guilt through directed counterfactual thinking about condition anticipated greater shame than guilt and that subjects behavior may be due to two factors. First, as evidenced by the in the mutate behavior condition anticipated greater guilt than inclusion of a no-mutate condition, undergraduates at Indiana shame. Ratings of shame and guilt from all subjects were then University felt that theflirtingsituation would cause them both submitted to a 2 (feelings rated: shame vs. guilt) X 3 (mutate: more guilt than the Johns Hopkins undergraduates did and also self vs. behavior vs. no mutate) ANOVA. There was a main that they would feel more guilt than shame. This high baseline effect of feelings, F(l, 283) = 4.33, p < .05, such that, overall, level of anticipated guilt may have imposed a ceiling effect on guilt such that directing subjects to mutate their actions could not raise guilt beyond the base level. Furthermore, it could be Table 4 the case that for the flirting scenario subjects tended to spontaMean Shame and Guilt Ratings From Subjects in Mutate neously mutate their actions in the absence of any directions. Behavior Conditions: Study 2a This explanation is also consistent with the pattern of reported shame. In both the mutate behavior and also the no-mutate conMutate self Mutate behavior Emotion rated ditions, subjects anticipated the same moderate amounts of 5.70 ( = 29) 4.38 (n = 26) shame. Subjects directed to mutate the self anticipated much Shame 5.15 ( = 26) Guilt 5.18 ( = 27) higher levels of shame. Nevertheless, the data do clearly show that focus of counterfactual thinking mediates quality of anticNote. Shame and guilt were rated on scales from 1 to 7, where 1 meant ipated feelings following an outcome that is ambiguous with rethat subjects would feel that emotion not at all and 7 meant that subjects spect to its cause. would feel the emotion very much.

SHAME AND GUILT

593

General Discussion
Previous research on counterfactual thinking has indicated that the discrepancy between factual and counterfactual outcomes can affect the valence and intensity of emotional reactions to events. The results of the current series of studies extend this literature by demonstrating that different patterns of counterfactual thinking are associated with qualitatively different affective experiences, above and beyond the dimensions of evaluation (valence) and intensity. Shame and guilt are both negatively valenced, self-relevant emotions that people experience in response to failures and transgressions. Although several empirical studies of the phenomenology of shame and guilt (Lewis, 1971;Tangney, 1989a; Tangney, Miller, & Flicker, 1992; Wicker et al., 1983) suggest some differences in intensity of these emotions (shame is generally a more powerful and painful experience than guilt), the difference between shame and guilt is far more than a matter of intensity. Shame is not merely a more intense form of guilt. Rather, shame and guilt appear to vary along a range of cognitive, affective, and motivational dimensions. One of the most fundamental distinctions has to do with the functional role of the self in these affective experiences. Whereas shame involves feelings of fairly global self-condemnation in which the self is seen as the root of some focal failure or transgression, guilt involves a much more articulated condemnation of an unacceptable behavior (or failure to act), somewhat apart from the self. Thefindingsfrom the studies in Series 1 clearly demonstrate that the counterfactual thoughts that accompany shame and guilt differ systematically, paralleling these differences in orientation to self versus behavior. When induced to feel (or asked to recall episodes of) shame compared with guilt, people were more likely to mentally undo aspects of self. When induced to feel (or asked to recall episodes of) guilt compared with shame, people were more likely to mentally undo aspects of their behavior. Our findings indicate not only that induced or remembered feelings of shame and guilt were associated with distinct patterns of counterfactual thinking, but thefindingsfrom studies in Series 2 also show that the nature of counterfactual thinking, itself, mediates these distinct affective experiences. Participants who were directed to produce counterfactuals about the self in connection with a negative interpersonal event subsequently reported higher levels of shame than guilt. Conversely, participants who were directed to produce counterfactuals about their behavior subsequently reported higher levels of guilt than shame. We interpret the results of the five studies together as indicating that counterfactual alternatives that come to mind when individuals face negative outcomes or experiences can influence the nature of their emotional reactions. We cannot conclude from these studies that counterfactual thinking is a requirement for the experiences of shame and guilt, but the results of Studies 2a and 2b certainly show that the focus of counterfactual thinking can contribute to the extent to which shame or guilt is felt. It is likely that there exists a rather complex, bidirectional relationship between the content of counterfactual thinking and emotional experience. In the spirit of M. Lewis's (1990) "cognitive-emotional fugue," perhaps certain types of emotional experiences foster characteristic ways of thinking, which in turn can highlight and intensify the experienced affect. Thus, cognition and affect can work hand in

hand, mutually sustaining and reinforcing quite distinct phenomenological experiences.

Differentiating Shame- and Guilt-Prone Individuals


Our findings do more than extend the literature on counterfactual thinking as a mediator of specific emotions. The present results are also relevant to our understanding of the nature and implications of individual differences in proneness to shame and proneness to guilt. Previous research has demonstrated that children, adolescents, and adults alike vary considerably in the degree to which they are likely to experience shame, guilt, or both, when faced with negative self-relevant situations (Tangney, 1990, 1991; Tangney, Wagner, Gavlas, Hill, & Marschall, 1992). Results from the current studies suggest that such individual differences in proneness to shame versus proneness to guilt may hinge, to some degree, on people's characteristic patterns of counterfactual thinking. That is, when faced with negative events, shame-prone individuals may be more likely to spontaneously think about their personal qualities and engage in counterfactual thinking about the self (e.g., "If only I were (or were not) such and such kind of person"). In addition, guilt-prone individuals may be more likely to focus on specific aspects of their behavior and engage in counterfactual thinking about their specific actions (e.g., "If only I had (or had not) done such a thing"). Perhaps more important, such individual differences in patterns of counterfactual thought may, in part, explain the consistent observation that proneness to shame and proneness to guilt are differentially related to a broad range of psychological and social characteristics (Tangney, 1991, 1993; Tangney, Wagner, Fletcher, & Gramzow, 1992; Tangney, Wagner, Gavlas, Hill, & Marschall, 1992; Tangney, Wagner, & Gramzow, 1992). In a nutshell, shame-proneness appears to be the more maladaptive affective style. Across numerous studies, proneness to shame has been linked to a range of psychological symptoms, as well as to an impaired capacity for interpersonal empathy, high levels of anger and hostility, and unconstructive anger management strategies. In contrast, guilt-proneness appears to be a fairly adaptive affective style, especially in the interpersonal realm. Proneness to "shame-free" guilt has been linked to empathic responsiveness and constructive anger management strategies, and this style appears generally unrelated to psychological symptomatology. Why is it that a tendency to experience shame has such negative implications for psychological and social adjustment? One possibility lies in the nature of the associated counterfactual thoughts. Counterfactual thoughts involve a mental undoing of events from the recent past. But such counterfactual thoughts also point us in alternative directions for the future (Markman et al., 1993). That is, when we contemplate how a negative situation might have otherwise unfolded, leading to a more agreeable outcome, we are also contemplating how things might be different in future similar situations. Counterfactual thoughts associated with guilt (about specific behaviors) typically represent quite straightforward possibilities for change in future events (Roese, 1994). We can readily imagine turning off the air conditioner, studying harder, resisting temptation, telling a truth, or helping another in some similar future circumstance.

594

P. NIEDENTHAL, J. TANGNEY, AND I. GAVANSKJ Affective responses to positive and negative outcomes. Personality and Social Psychology Bulletin, 13, 524-536. Lewis, H. B. (1971). Shame and guilt in neurosis. New York: International Universities Press. Lewis, M. (1990). Thinking and feelingthe elephant's tail. In C. A. Maher, M. Schwebel, & N. S. Fagley (Eds.), Thinking and problemsolving in the developmental process: International perspectives (pp. 89-110). Hillsdale, NJ: Erlbaum. Lindsay-Hartz, J. (1984). Contrasting experiences of shame and guilt. American Behavioral Scientist, 27, 689-704. Lipe, M. G. (1991). Counterfactual reasoning as a framework for attribution theories. Psychological Bulletin, 109, 456-471. Markman, K. D., Gavanski, I., Sherman, S. J., & McMullen, M. N. (1993). The mental simulation of better and worse possible worlds. Journal of Experimental Social Psychology, 29, 87-109. Roese, N. J. (1994). The functional basis of counterfactual thinking. Journal of Personality and Social Psychology, 66, 805-818. Roese, N. J., & Olson, J. M. (in press). Counterfactual thinking: A critical overview. In N. J. Roese & J. M. Olson (Eds.), What might have been: The social psychology of counterfactual thinking. Hillsdale, NJ: Erlbaum. Roseman, I. (1984). Cognitive determinants of emotion: A structural theory. In P. Shaver (Ed.), Review ofpersonality and social psychology: Emotions, relationships and health (Vol. 5, pp. 11-36). Beverly Hills, CA: Sage. Scherer, K. R. (1988). Cognitive antecedents of emotion. In V. Hamilton, G. H. Bower, & N. H. Fridja (Eds.), Cognitive perspectives on emotion and motivation (pp. 89-126). Dordrecht, The Netherlands: Kluwer. Smith, C. A., & Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. Journal of Personality and Social Psychology, 48, 8 1 3 838. Smith, C. A., & Ellsworth, P. C. (1987). Patterns of appraisal and emotion related to taking an exam. Journal of Personality and Social Psychology, 52, 475-488. Tangney, J. P. (1989a, August). A quantitative assessment ofphenomenological differences between shame and guilt. Poster presented at the 97th Annual Convention of the American Psychological Association, New Orleans, LA. Tangney, J. P. (1989b). Shame and guilt in young adulthood: A qualitative analysis. Poster presented at the meetings of the American Psychological Society, Arlington, VA. Tangney, J. P. (1990). Assessing individual differences in proneness to shame and guilt: Development of the Self-Conscious Affect and Attribution Inventory. Journal of Personality and Social Psychology, 59, 102-111. Tangney, J. P. (1991). Moral affect: The good, the bad, and the ugly. Journal of Personality and Social Psychology, 61, 598-607. Tangney, J. P. (1992). Situational determinants of shame and guilt in young adulthood. Personality and Social Psychology Bulletin, 18, 199-206. Tangney, J. P. (1993). Shame and guilt. In C. G. Costello (Ed.), Symptoms of depression (pp. 161-180). New York: Wiley. Tangney, J. P., Marschall, D. E., Rosenberg, K., Barlow, D. H., & Wagner, P. E. (1993). Children's and adults' autobiographical accounts of shame, guilt, and pride experiences: A qualitative analysis of situational determinants and interpersonal concerns. Manuscript submitted for publication. Tangney, J. P., Miller, R. S., & Flicker, L. (1992, August). A quantitative analysis of shame and embarrassment. Poster presented at the 100th Annual Convention of the American Psychological Association, Washington, DC. Tangney, J. P., Wagner, P. E., Fletcher, C , & Gramzow, R. (1992). Shamed into anger? The relation of shame and guilt to anger and self-

The counterfactual thoughts associated with shame (about the self), however, are more likely to present a serious dilemma. As clinicians well know, changes to the self pose a much greater challenge. Particularly when shame-related counterfactual thoughts center on fairly global characteristics of the self, the opportunities for change in the future may be quite limited. In this light, the anxiety, depression, anger, and unempathic selfinvolvement of the shame-prone individual are easily understandable.

Conclusion
In sum, the present findings are consistent in showing that counterfactual thinking can mediate specific emotions. Fridja et al. (1989) showed that both shame and guilt (as well as regret) are associated with a concern with mental undoing, something that clinicians have noted for decades. The present work shows that this mental undoing takes different forms in shame and guilt and that these different forms of undoing contribute to the phenomenologies of the two emotions.

References
Abelson, R. P. (1983). Whatever became of consistency theory? Personality and Social Psychology Bulletin, 9, 37-54. Arnold, M. B. (1960). Emotion and personality (Vols. 1 & 2). New York: Columbia University Press. Brown, J., & Weiner, B. (1984). Affective consequences of ability versus effort ascriptions: Controversies, resolutions, and quandaries. Journal of Educational Psychology, 76, 146-158. Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A control theory approach to human behavior. New "York: Springer Verlag. Einhorn, H. J., & Hogarth, R. M. (1986). Judging probable cause. Psychological Bulletin, 99, 3-19. Fridja, N. H., Kuipers, P., & ter Schure, E. (1989). Relations among emotion, appraisal, and emotional action readiness. Journal of Personality and Social Psychology, 57, 212-228. Gavanski, I., & Wells, G. L. (1989). Counterfactual processing of normal and exceptional events. Journal of Experimental Social Psychology, 25, 314-325. Gleicher, F. H., Kost, K. A., Baker, S. M., Strathman, A., Richman, S. A., & Sherman, S. J. (1990). The role of counterfactual thinking in judgments of affect. Personality and Social Psychology Bulletin, 16, 284-295. Higgins, E. T. (1989). Self-discrepancy theory: What patterns of selfbeliefs cause people to suffer? In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 22, pp. 93-136). San Diego, CA: Academic Press. Johnson, J. T. (1986). The knowledge of what might have been: Affective and attributional consequences of near outcomes. Personality and Social Psychology Bulletin, 12, 51-62. Kahneman, D., & Miller, D. (1986). Norm theory: Comparing reality to its alternatives. Psychological Review, 93, 136-153. Kahneman, D., & Tversky, A. (1982). The simulation heuristic. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases. Cambridge, England: Cambridge University Press. Kelley, H. H. (1972). Attribution in social interaction. In E. E. Jones (Ed.), Attribution: Perceiving the causes of behavior (pp. 1-26). Morristown, NJ: General Learning Press. Landman, J. (1987). Regret and elation following action and inaction:

SHAME AND GUILT

595

reported aggression. Journal of Personality and Social Psychology, 62, Weiner, B. (1980). May I borrow your class notes? An attributional analysis of judgments of help giving in an achievement-related context. 669-675. Journal ofEducational Psychology, 72, 676-681. Tangney, J. P., Wagner, P. E., Gavlas, J., Hill, D. A., & Marschall, D. E. (1992, August). Shame, guilt, and constructive v. destructive anger. Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological Review, 92, 548-573. Poster presented at the 100th Annual Convention of the American Weiner, B., & Graham, S. (1984). An attributional approach to emoPsychological Association, Washington, DC. tional development. In C. E. Izard, J. Kagan, & R. B. Zajonc (Eds.), Tangney, J. P., Wagner, P., & Gramzow, R. (1992). Proneness to shame, Emotions, cognition, and behavior. Cambridge, England: Cambridge proneness to guilt, and psychopathology. Journal of Abnormal PsyUniversity Press. chology, 103,469-478. Wells, G. L., & Gavanski, I. (1989). Mental simulation of causality. Tangney, J. P., Wagner, P. E., Barlow, D. H., Marschall, D. E., Sanftner, Journal ofPersonality and Social Psychology, 56, 161-169. J., Mohr, T., & Gramzow, R. (1993). The relation ofshame and guilt to constructive vs. destructive responses to anger across the lifespan.Wicker, F. W., Payne, G. G, & Morgan, R. D. (1983). Participant descriptions of guilt and shame. Motivation and Emotion, 7, 25-39. Manuscript submitted for publication.

Appendix Personal Experiences of Shame and Guilt Generated by Subjects in Study lc Shame Scenarios
"A most shameful event of my life occurred when I attempted to cheat on a quiz in myfirstyear of high school. The class consisted of about 30 other kids and most cheated on every test. I had never cheated (never did again) before, but did it by peer pressure. A person I disliked mocked my behavior and at the end of the day I confessed to the teacher. I felt shame because I compromised my own values, drawing criticism from those around me and myself." If only... 1. I had not cheated (obvious) [Behavior] 2. I had more self-confidence to avoid such situations [Self] 3. I had better study skills [Self] 4. I could think holistically, see past immediate gratification [Self] "When my parents found out about a class I Fed. They thought I was doing so well in school, but really I was partying too much, not studying for exams, not going to class. My own ignorance caused me to behave this way. Just the look on my parent's face, you know that disappointed look." If only... 1. I could be the daughter they expected me to be [Self] 2. I was a better student [Self] 3. My friends weren't such bad influences on me to party all of the time [Situation] 4. I could just say "no" [Self] Guilt Scenarios "I got into afightwith my boyfriend. I love him and have been dating him for over a year. I picked thefightbecause I was in a bad mood and taking it out on him. I said things that hurt him and he says he forgets them, but I know he won't. I told him he bored me, I was sick of him and I wasn't having fun. I told him he was ignorant, lazy, self-centered, irresponsible. I feel awful, I cried, I could kill myself for doing it." If only... 1. Stopped saying the things instead of keep going [Behavior] 2. Not said a word to begin with [Behavior] 3. Told him I was in a bad mood and just not to talk [Behavior] 4. Not even gotten into the car [Behavior] "I was out with some friends (at the bars) and saw a girl I had recently met. We talked, drank, etc., and I eventually ended up driving her home. I spend the evening and although nothing sexual happened, except for a few kisses, I felt guilty because I have a girlfriend who I'd been dating for 11 months at the time. Alcohol was possibly a slight cause because everything with my girlfriend was goingfine(no rocky road) and I was happy with her and still am." If only . . . 1. I could have taken my girlfriend out that night, instead of going out with the boys [Behavior] 2. I could have drank less, so I could have gone home safely after dropping her off [Behavior] 3. I would have stayed home because summer session started the next day [Behavior] 4. I wouldn't have taken her home [Behavior] Note. Terms in brackets indicate how judges categorized each counterfactual statement.

Received June 10,1993 Revision received January 20, 1994 Accepted February 10, 1994

Das könnte Ihnen auch gefallen