Guilt: An Interpersonal Approach

Roy F Baumeister, Arlene M Stillwell, Todd F Heatherton. Psychological Bulletin. Volume 115, Issue 2. March 1994.

Guilt is a common form of emotional distress and a common factor in behavioral decisions. People invoke guilt feelings to apologize for misdeeds, to express sympathy, to manipulate others, to refuse sex, to discipline children, to bolster self-control, and more, and they perform or avoid a stunning variety of actions because of the anticipation of guilt. Yet the fields of personality and social psychology have tended to ignore this important aspect of everyday life. Not including the few articles on specifically sexual guilt, the entire decade of the 1980s witnessed only three titles in the Journal of Personality and Social Psychology  that mentioned guilt (Cunningham, Steinberg, & Grev, 1980; McGraw, 1987; Wertheim & Schwartz, 1983). During the same decade, the Annual Review of Psychology contained no articles on guilt, and the volume indexes listed only three pages that mentioned the word. Ironically, even undergraduate textbooks on motivation devote little or no space to guilt. Apart from sex guilt and legal decision making, both of which involve technical definitions of guilt that may differ in important ways from moral and everyday guilt, the topic has been neglected.

Social psychology’s neglect of guilt is especially ironic in that guilt may be a pervasively and essentially social emotion. This article examines the social and interpersonal nature of guilt. In contrast to some theoretical traditions that regard emotions as primarily intrapsychic responses and treat guilt in particular as a matter of self-evaluation against abstract standards, our purpose is to examine guilt as a result and a mechanism of human relatedness. In brief, we focus on an understanding of the fundamentally social nature of guilt.

We propose that guilt is something that happens between people rather than just inside them. That is, guilt is an interpersonal phenomenon that is functionally and causally linked to communal relationships between people. The origins, functions, and processes of guilt all have important interpersonal aspects. Guilt can be understood in relationship contexts as a factor that strengthens social bonds by eliciting symbolic affirmation of caring and commitment; it is also a mechanism for alleviating imbalances or inequities in emotional distress within the relationship and for exerting influence over others. Furthermore, the social nature of guilt goes far beyond the familiar recognition that moral standards are instilled in the young by significant others and by society in general. Indeed, guilt continues to arise mainly out of interpersonal transactions throughout life. Often people make their relationship partners feel guilty because of current transgressions and in the hope of producing particular changes.

This is not to deny that some experiences of guilt can take place in the privacy of one’s individual psyche, in social isolation. Still, many of those instances may be derivative of interpersonal processes and may reflect highly socialized individuals with internalized reference groups, and such instances may often turn out to revolve around interpersonal issues as well. Our argument follows the social science tradition reflected in symbolic interactionism, social learning theory, socialization theory, Sullivanian personality theory, Heideggerian phenomenology, and others that regard many intrapsychic phenomena as having interpersonal origins. Moreover, we argue that most instances of guilt continue to be clearly and essentially linked to interpersonal processes.

Interpersonal Approach in Prior Theory

The social nature and interpersonal origins of guilt have been affirmed by some past theorists, denied by others, and downplayed by most. Some theorists paradoxically combined analyses that seemingly implied important social aspects of guilt with explicit statements that denied any such aspects. Thus, Freud (1930/1961, 1933/1964) treated guilt as the product of intrapsychic conflicts; more precisely, guilt was the weapon used by the superego to influence the ego’s decisions. In his view, the “moral sense of guilt is the expression of the tension between the ego and the super-ego” (1933/1964, p. 76). Indeed, he later insisted that the operation of the superego involved generating feelings of guilt without regard to the external world (1933/1964, p. 97). Yet lurking in the background of Freud’s theory was some role of interpersonal factors, insofar as he considered the superego to be an adaptation of the human organism to civilized life (Freud, 1930/1961), which almost by definition includes living together with other people.

Likewise, Helen Block Lewis (e.g., 1971) asserted baldly that interpersonal factors and processes were irrelevant to guilt: “Guilt is evoked only from within the self” (p. 85). In fact, she insisted, guilt does not even derive from an imaginary contact with another person, such as a generalized other or internalized reference group (cf. Baldwin & Holmes, 1987): “The imagery of the self vis-a-vis the ‘other’ is absent in guilt” (Lewis, 1971, p. 251). Thus, her analysis explicitly denied any significant role of interpersonal processes, even though many of her arguments seemed to suggest interpersonal factors.

Other theorists have explicitly and consistently rejected an interpersonal aspect of guilt. Piers and Singer (1953/1971) derived guilt from castration anxiety and treated it as a response to id impulses of aggressiveness and destructiveness and possibly sexuality (especially incest). They insisted that “genuine guilt feelings” are “experienced in solitude and contain no conscious or realistic reference to an audience” (p. 68), thus explicitly denying any interpersonal dimension. Likewise, Gilligan (1976) asserted that guilt arises in response to aggressive drives connected with the early stages in the Freudian scheme, namely “oral-biting-cannibalistic-sadistic, anal-sadistic, and phallic-competitive” (p. 149), and he emphasized the wish for punishment as an important aspect of guilt. From a different perspective, Buss (1980) linked guilt to private self-consciousness, which by definition does not necessarily involve other people or their perspectives: “Guilt is essentially private. The best test of guilt is whether anyone else knows of the transgression” (p. 159). He added that the fact that no one else need know about it confirms its intrapsychic nature. Thus, again, guilt is a solitary affair and a product of mainly intrapsychic processes.

Jones, Kugler, and Adams (in press) concluded recently that the “traditional view” of guilt has limited the role of interpersonal factors to that of teaching and transmitting standards, such as in socializing children, and has generally assumed that once moral standards and prohibitions have been learned, guilt becomes a private matter of self-evaluation. After that point, other people are involved only peripherally or not at all. Thus, for example, Kohlberg (1981, 1984) described guilt in psychoanalytic terms of self-punishment and self-judgment, although his notion of guilt as based on the formation of internal moral standards presumably involves acquiring standards from others. Undoubtedly, Freud and Lewis would also concur that guilt is learned socially.

A similar assessment of the literature was reached independently by Vangelisti, Daly, and Rudnick (1991), who concluded that the “predominant theories of emotion” merely “pay lip service to the potential social nature of emotion” and rarely do more (p. 4). Thus, such theories may state that a social dimension exists, but the interpersonal processes and dynamics in emotion are rarely discussed and even more rarely studied in systematic empirical work.

Likewise, the behaviorist theory of guilt proposed by Mosher (1965) asserted that “guilt may be defined as a generalized expectancy for self-mediated punishment (i.e., negative reinforcement [sic]) for violating, anticipating the violation of, or failure to attain internalized standards of proper behavior” (p. 162). The reference to self-mediated punishment makes clear that the focus of guilt is not punishment by others but the administration of aversive outcomes to oneself. Thus, in this behaviorist view, guilt is an expectation of self-harm, not an interpersonal phenomenon. This definition does not assign any role to interpersonal factors, except again for the possibility that “internalized standards” can be learned from others.

On the other hand, some theorists have suggested social roots of guilt. Rank (e.g., 1929) started out with standard psychoanalytic views about guilt, but his thinking gradually evolved into a theoretical position that one recent commentator characterized as “unique in psychoanalytic psychology” (Menaker, 1982, p. 51). Although Rank retained such standard features as the belief that guilt causes a wish for punishment, he came to understand guilt as an inevitable by-product of the individuation process. Guilt thus originates in the infantile attachment to mother and in the fear and anxiety over breaking that attachment, according to Rank, and it operates as a force that perpetuates that relationship. De Rivera (1984) proposed that all emotional states are based on interpersonal relationships and, indeed, that all emotions are fundamentally concerned with adjusting these relationships. Ausubel (1955) asserted that shame is a reaction to actual or presumed judgments of others and that guilt always involves a “special type of moral shame” (p. 389), which seemingly implies a strong link to interpersonal contexts. A similar implication can be found in Ausubel’s (1955) claim that guilt “always implies an offense against the group” (p. 388). Horney (1937) inverted traditional psychoanalytic theory by asserting that guilt feelings derive from a basic fear of other people’s disapproval, and she speculated that efforts to make others feel guilty and insincere and exaggerated confessions of one’s own guilt can arise from neurotic interpersonal motives. Micelli (1992) has offered thoughtful speculations about the processes of inducing guilt in others. Evolutionary theorists (e.g., Trivers, 1985) have suggested that human guilt emerged from natural selection because it prevented human beings from performing exploitative actions that might damage their relationships with others (because such relationships are vital to survival and reproduction).

These inconsistencies within and among prior theories about guilt are understandable when one considers the empirical vacuum in which they were generated. Many of the theorists we have cited had to rely mainly on intuition, unsystematic observation, and clinical impressions in formulating their veiws about guilt. In contrast, there are many empirical studies scattered through the literature of the past several decades, and these studies can presumably offer a firmer basis for understanding guilt. Indeed, the accumulation of findings may be outstripping the development of theory. Zahn-Waxler and Kochanska (1990), among others, have called for new conceptual work on guilt to replace outmoded theories and to integrate recent empirical findings. The purpose of this article is to elaborate one aspect of guilt that has been relatively neglected and undervalued in previous theory but that has emerged as significant in many empirical investigations: the interpersonal dimension of guilt.

Definition of Guilt

Efforts to construct simple definitions of guilt are plagued by the fact that people use the term in multiple and conflicting ways. In particular, people use shame and guilt interchangeably, even though the terms refer to distinct and distinguishable experiences (Ferguson, Stegge, & Damhuis, 1991; Lewis, 1971; Tangney, 1989, 1990, 1991; Wicker, Payne, & Morgan, 1983). The main interest of psychologists is in the subjective feeling of guilt, along with its causes and behavioral consequences. This article is concerned with what makes people feel guilty and what that feeling—or the motivation to avoid that feeling—causes them to do.

Our approach to guilt as a subjective state renders irrelevant some other important and influential usages of guilt. Legal guilt, for example, has technical definitions that are quite independent of subjective feelings or even of responsibility for past actions. Legal guilt is based on the notion of violating legislated rules, although the technical meaning of legal guilt has evolved beyond that to refer to the condition and extent of evidence. (Thus, a not-guilty verdict refers to the inadequacy of evidence for conviction rather than a definite conclusion that the defendant did not perform the act in question.) According to Lewis (1971), people may be guilty without any particular feeling (see also Ortony, 1987); such a definition fits the legal uses of guilt but is not our focus. Similarly, the term sex guilt is often used in misleading ways, particularly in connection with certain questionnaire measures, when, in fact, what is meant is sexual inhibitions and sexual shame (see Tangney, 1990). Ortony (1987) has argued that guilt has two meanings, a sociolegal one and an emotional one. In that perspective, this article focuses on the emotional sense of guilt.

By guilt we refer to an individual’s unpleasant emotional state associated with possible objections to his or her actions, inaction, circumstances, or intentions. Guilt is an aroused form of emotional distress that is distinct from fear and anger and based on the possibility that one may be in the wrong or that others may have such a perception. Thus, we focus here on feelings of guilt rather than guilt in a legal, technical, ontological, or theological sense.

Guilt can be distinguished from shame on the basis of specificity. Guilt concerns one particular action, in contrast to shame, which pertains to the entire self (e.g., Lewis, 1971; Tangney, 1990, 1991). Guilt can likewise be distinguished from fear of punishment on the basis that the distress pertains to the action itself rather than to the expectation of hedonically aversive consequences of the action. One can clearly feel guilt in cases in which there is no likelihood of punishment and hence no fear; the knowledge that one has harmed another person may be enough to cause guilt, even if the victim is in no position to retaliate. (On the other hand, it should be relatively difficult to fear punishment by others without feeling any guilt oneself, unless perhaps the other person is regarded as motivated by hostility rather than hurt or as outside the realm of desirable or potential relationship partners.)

By defining guilt in terms of the subjective feeling state, we are clearly acknowledging the role of intrapsychic processes and events. It would be a caricature of our view to assert that there is nothing intrapsychic about guilt. Our analysis suggests, however, that the causes, consequences, and functions of this intrapsychic response have substantial interpersonal aspects.

Interpersonal Analysis of Guilt

From an interpersonal perspective, the prototypical cause of guilt would be the infliction of harm, loss, or distress on a relationship partner. Although guilt may begin with close relationships, it is not confined to them; guilt proneness may become generalized to other relationships, including even minimal intergroup phenomena (Brewer, 1979) and “fellow feeling” based on community spirit or collective membership or commonality. In particular, a well-socialized individual would presumably have learned to feel guilty over inflicting harm to even a stranger. An interpersonal approach would predict, however, that guilt reactions would be stronger, more common, and more influential in close relationships than in weak or distant ones. As the commonality (i.e., community, relationship, or fellow feeling) between two people approaches zero, the possibility for guilt should also approach zero.

Communal Relatedness

A useful framework for understanding the relevant type of relatedness is the distinction between communal and exchange relationships, as elucidated by Clark and her colleagues (e.g., Clark, 1984; Clark & Mills, 1979). Communal relationships are defined by the existence of implicit rules that the individuals must be concerned about each other’s welfare (e.g., Clark, Mills, & Powell, 1986). As a result, communal relationship partners do things simply to benefit each other without expecting equal or immediate benefits in return, in contrast to partners in exchange relationships, who seek to maintain equity and who benefit each other only when anticipating equal or greater benefits in return. Research has established that people seem ready to interact with even a stranger on a communal basis simply because they anticipate the possibility of a communal relationship (e.g., Clark, 1986; Clark et al., 1986; Clark, Ouellette, Powell, & Milberg, 1987). In communal relationships, people spontaneously keep track of each other’s needs, even if nothing can immediately be done to satisfy those needs (Clark et al., 1986; Clark, Mills, & Corcoran, 1989); they provide more help to their relationship partners (Clark et al., 1987); they feel better after helping, both in terms of mood and in terms of self-evaluation (Williamson & Clark, 1989); and they are more responsive to their partner’s emotional states (Clark et al., 1987). In our understanding, Clark’s characterization of communal and exchange relationships is intended to present them as ideal types, with the understanding that, in reality, many relationships (e.g., with an employer or neighbor) may contain some mixture of both. Nonetheless, it is plausible that guilt in all relationships arises primarily from the communal component and is proportional to it.

To describe guilt as an interpersonal or social phenomenon might mean quite different things depending on whether one is focusing on communal or exchange relationships. For example, Freud’s (1930/1961) comments on the social nature of guilt can be entirely understood in terms of exchange relationships. He portrayed guilt as a by-product of the human adaptation to life in civilized society. The essence of such an adaptation to is that all members must renounce certain inclinations and satisfactions so that all may be protected from victimization by others. Guilt, to the extent that Freud reflected on its social nature, was seen as a result of the internal mechanism that causes each individual to obey the group’s rules, thereby making bargains and trade-offs viable.

If (unlike Freud) one analyzes guilt in terms of communal relationships, however, cost–benefit analyses lose their central importance. Instead, one can see guilt as designed to enforce the communal norms of mutual concern and nurturance and to protect the interpersonal bond between close individuals. The functions of guilt should, therefore, be relationship enhancing.

Furthermore, it appears that people desire and perhaps need communal relationships, so people may sometimes respond on the basis of communal relationship norms simply because the other person is a potentially available relationship partner (i.e., even if no communal relationship yet exists). Studies have shown that many people will adopt a communal approach (e.g., attending to another person’s needs but not keeping track of his or her inputs and contributions) simply because they believe that the person they have just met could conceivably become a communal partner (e.g., being unmarried and interested in meeting people, as opposed to being married; see Clark, 1984, 1986; Clark et al., 1987). These findings are of importance later in this review in understanding why some people respond to apparent strangers in ways that would seemingly be reserved for intimate partners.

Affective Sources of Guilt

What makes people capable of feeling guilty? As Zahn-Waxler and Kochanska (1990) wrote, “The capacity for guilt is innate and universal; [only] its modes of expression are learned” (p. 232). Thus, some consideration of this innate capacity is a necessary point of departure for guilt theory. Clearly, guilt means feeling bad, and so the capacity for guilt presumably begins with a natural basis for feeling bad. Guilt must be understood as an unpleasant arousal akin to anxiety (Tennen & Herzberger, 1987). We propose two sources: empathic arousal and anxiety over social exclusion. Both of these are important, powerful sources of affect and motivation in close, communal relationships.

People are innately prepared to feel empathic distress in response to the suffering of others, and guilt combines empathic distress with a self-attribution of causal responsibility for the other’s suffering, according to Hoffman (1982). When seeing others suffer, a person will feel bad, and this bad feeling is the basis for guilt. Although empathic distress can occur with anyone’s suffering, empathic responses are generally recognized to be strongest in close relationships. Communal concern for another’s welfare would presumably have a strong link to empathic responses.

Along with empathy, belongingness and attachment are powerful sources of affective response. Human beings experience anxiety at the threat of separation or exclusion from their mothers (Bowlby, 1969, 1973) or, indeed, from significant others in general (Baumeister & Tice, 1990). Events that raise the threat of social exclusion should therefore cause anxiety, and one form this anxiety may take is guilt. Jones and Kugler (in press) found strong and significant correlations between anxiety and guilt. In particular, if one has done something (e.g., transgressed) that might cause rejection by one’s partner, then the resultant anxiety over the possible rejection could be experienced as guilt.

Combining empathic distress and exclusion anxiety furnishes a potentially powerful basis for analyzing guilt and predicting its patterns of occurrence. Guilt should arise in the context of communal social attachments and be linked to disturbances of the sense of belonging, particularly those that arise from suffering that oneself has caused. Developmentally primitive guilt should focus mainly on the suffering of those closest (in similarity or relationship) to oneself. More mature guilt may reflect the increasingly broad sense of community, but there should still be tendencies for guilt to be strongest with intimate partners and in-group members.

Viewed in this way, the emotional basis of guilt has a strong interpersonal component. This view differs substantially from treatments that base guilt on such factors as castration anxiety (e.g., Piers & Singer, 1953/1971), internally directed aggression (e.g., Freud, 1930/1961; Lewis, 1971), or the conditioned expectation of punishment (e.g., Mosher, 1965), as well as treatments that acknowledge a social role of guilt but cover this role in terms of exchange relationships. The affective roots of guilt lie in human relatedness, that is, in the human capacity to feel the suffering and distress of others and in the basic fear of alienating actual or potential relationship partners.

It is also clear that human beings do generalize and internalize, and so in some cases a person may feel guilty over a violated standard even when no one is harmed or disappointed and no one knows about the incident. According to an interpersonal analysis, however, these would be the most derivative instances of guilt; therefore, they would require the most training or socialization to create and would be the easiest forms of guilt to escape and the patterns most likely to extinguish quickly. In contrast, the pattern of feeling guilty over hurting a relationship partner should be easy to create and maintain because it is linked to the fundamentally interpersonal (and prototypical) nature of guilt.

Three Functions of Guilt

The interpersonal perspective suggests that guilt serves three broad functions for relationships; that is, guilt strengthens social bonds and attachments in three ways. (However, even though guilt should be regarded as a tool designed for constructive ends, it also contains the potential for destructive use; although guilt may often enhance relationships, it may harm them too.)

First, guilt motivates relationship-enhancing patterns of behavior. In other words, guilt helps enforce the communal norms that prescribe mutual concern, respect, and positive treatment in the absence of self-interested return. Guilt may punish and hence reduce the frequency of interpersonal transgressions so that it makes people less likely to hurt, disappoint, or alienate their partners. Guilt may also motivate people to pay attention and express positive feelings to their partners. This could serve the relationship-enhancing function of guilt insofar as interpersonal neglect, including the careless failure to pay attention to a partner, may be harmful (Derber, 1979; Fiore & Swensen, 1977; Rice, 1990; Rusbult, Johnson, & Morrow, 1986). In general, if people feel guilty for hurting their partners, for neglecting them, and for failing to live up to their expectations, they will alter their behavior (to avoid guilt) in ways that seem likely to maintain and strengthen the relationship.

Second, guilt may operate as an interpersonal influence technique that allows even a relatively powerless person to get his or her way. If Person A wants Person B to do something, A may induce guilt in B by conveying how A suffers over B’s failure to act in the desired fashion. Person B finds the guilt aversive and, to escape from guilt, complies with A’s wishes. This is clearly an influence pattern suited to communal relationships, because A is not offering reciprocal benefits to get his or her way but is using the relationship bond to create an affective state that will motivate B to do what A wants.

Several aspects of guilt as an influence strategy deserve brief mention. Guilt does not depend on formal power or influence and may even work best in the absence of such power, because one induces guilt by depicting oneself as the helpless victim of another’s actions. Hence, guilt may restore equity within a relationship by enabling the less powerful person to get his or her way. Although formal power is unnecessary, the social bond is often important, because guilt succeeds as influence only to the extent that the transgressor cares about the victim. (Thus, again, guilt may operate most strongly and commonly in close, communal relationships.) Furthermore, guilt may depend on the partner’s (i.e., victim’s) appraisal as well as on the person’s self-appraisal, and so some people could conceivably feel guilty despite believing that they have done nothing wrong. People may, however, resent the manipulation of guilt feelings if their self-evaluation exonerates them, and the result may be resentful compliance. Guilt would thus be an effective but potentially costly technique for getting one’s way.

The third function of guilt is to redistribute emotional distress within the dyad. After a transgression, the victim is presumably suffering while the transgressor has benefited. If the transgressor feels guilty, however, his or her enjoyment is diminished, and the transgressor’s guilt may make the victim feel better. In this way, emotional equity is restored (because bad feelings are, in effect, restored to the person who caused them). Moreover, if guilt brings the two partners’ emotional states into closer harmony, this very equality may be beneficial for the relationship. Locke and Horowitz (1990) found that people who were in similar affective states were able to have a mutualty satisfactory interaction with a positive outcome, whereas people in mismatched affective states entered into a vicious cycle of poor communication, mutual dissatisfaction, and increasing withdrawal. To generalize broadly from their findings, wide affective discrepancies may be harmful to a communal relationship, and so a mechanism for redistributing negative affect may be relationship enhancing. Likewise, it seems intuitively plausible that an episode involving guilt could improve communication between two relationship partners when one has wronged the other.

Why should victims feel better when a transgressor feels guilty? First, the transgressor’s guilt affirms a commitment to the relationship, which is thus a potentially powerful indication of affection, caring, and intimacy that may be pleasing and reassuring to an intimate partner: Feeling guilty is a way of showing that one cares. Second, if the transgressor acknowledges guilt, the victim may see this as an implicit commitment not to repeat the offense, as a promise to rectify that transgression by making amends, or as an acknowledgment of a nonspecific debt toward the victim. Thus, despite the victim’s misfortune or distress, he or she may be pleased to regard the partner’s acknowledgment of guilt as an implicit promise of better treatment in the future.

Positive Inequity and Guilt

Although we have emphasized interpersonal harm as the prototypical cause of guilt, it is apparent that mere positive inequity may sometimes be sufficient to cause guilt. Equity theory (e.g., Walster, Berscheid, & Walster, 1976) holds that people prefer outcomes to be fair and proportional to inputs (or other causes of deservingness). Deviations in either direction produce emotional distress. Receiving less than one deserves may cause resentment, anger, envy, and feelings of having been cheated. Receiving more than one deserves may cause guilt, especially in relation to other people who failed to be similarly overrewarded (Hassebrauck, 1986). Guilt is thus one subcategory of distress over inequity: Specifically, it is the distress suffered by people who are overrewarded. Positive inequity may resemble the prototype causation of guilt in the sense that when one harms a relationship partner, one presumably benefits at the partner’s expense; thus, benefiting inequitably—and thus seemingly at the partner’s expense—may be sufficiently similar as to cause guilt feelings.

In principle, an equity theory of guilt could be formulated in terms of calculations by solitary individuals. That is, people might feel guilty simply because they regard their outcomes as being better than they deserve. Research, however, has generally shown inequity guilt in interpersonal contexts, in the sense that the person is overrewarded vis-à-vis some salient other person. Solitary overrewards (e.g., receiving extra change from a pay telephone) seem to result in positive moods rather than guilt (see Isen, 1984). By focusing on interpersonal inequities, then, we are consistent with the main thrust of the research evidence. The equity view is linked to the interpersonal view in its emphasis on the interpersonal context, insofar as the comparison between one’s own outcomes and others’ outcomes is the criterion of fairness.

Exchange relationships require some mutual understanding of norms of fairness and equitable exchange, so equity guilt might be expected to be found primarily in exchange rather than communal relationships. On the other hand, we propose that even equity guilt is based on a fundamental concern for the other person and hence implies communality. In our view, it is necessary to assume that there is some need for approximate equity over the long term even in close, communal relationships, and persistent inequities will eventually weaken and destabilize the relationship. Partners in exchange relationships can presumably settle issues of fairness and equity on a regular, ongoing, or immediate basis, because both parties understand equitable exchange as the basis for their interaction and so each can appeal to it readily. In contrast, such discourse may be more difficult in communal relationships because mutual concern, rather than equitable exchange, is understood as the basis for the relationship. To protect a communal relationship by enhancing equity over the long term would therefore require some mechanism based on mutual concern, and guilt may be an effective mechanism of this sort. If the interpersonal view is correct, inequities should be more distressing in close relationships than in casual or distant ones. Thus, as a mechanism for promoting interpersonal equity, guilt may again serve a relationship-enhancing function.

Reviews of Empirical Findings

The literature reviewed here consists of empirical studies, primarily in social and personality psychology. Some developmental research has also been included, but a fine recent review is available for that literature (Zahn-Waxler & Kochanska, 1990), so we have not sought to duplicate that effort. There is a substantial psychoanalytic literature on guilt, including case studies, but we have not included that material in this review, consistent with the widespread view that single case studies are suitable for developing and illustrating theory but not for testing it. In addition, we did not cover the literatures on sex guilt and legal guilt, as already noted. Our purpose was to examine from an interpersonal perspective the bulk of accumulated empirical knowledge about guilt.

Autobiographical Accounts and Subjective Experience

Perhaps the most direct method of studying the sources and subjective experience of guilt in everyday life is to ask people to describe recent or significant events from their lives that have made them feel guilty. This approach relies on autobiographical narratives, in which subjects are asked to describe true incidents from their lives (e.g., Gergen & Gergen, 1988; J. H. Harvey, Weber, & Orbuch, 1990). It is especially useful for studying topics like guilt in which ethical and pragmatic concerns limit what can be done in the laboratory. One can at least be certain that one is dealing with actual experiences of guilt, as understood by people who have had them, although these studies have generally left it up to the subject to decide what precisely was meant by guilt. For present purposes, the key question is whether people’s accounts about guilt emphasize interpersonal contexts.

Tangney (1992) found that reports of incidents that led to guilt were overwhelmingly interpersonal. Even when people reported feeling guilty alone, the incident about which they felt guilty tended to be interpersonal. Surprisingly few accounts dealt with solitary or victimless transgressions. Likewise, Brooke’s (1985) sample of first-person accounts led him to observe that guilt typically invokes other people, real or imagined, who adopt an accusatory role, and he concluded that “guilt always occurs in an interpersonal context of shared values” (p. 37; see also Millar & Tesser, 1988).

Vangelisti et al. (1991) and Baumeister, Stillwell, and Heatherton (in press) asked subjects to describe incidents in which they induced guilt in other people or in which other people induced guilt in them, and subjects had no difficulty recalling such incidents. The fact that people make other people feel guilty is a powerful sign of interpersonal causation. It is readily apparent that people induce and manipulate guilt feelings in others.

Furthermore, some subjects in the Baumeister et al. (in press) study reported feeling guilty (and complying with the wishes of the guilt inducer) even when they disagreed with the other person’s assessment of the situation and resented his or her manipulations (see also Rubin & Shaffer, 1987). When people described how others made them feel guilty, they frequently referred to differing expectations and to the other person’s standards. Such discrepancies between one’s own standards or expectations and those of the other person may be especially salient to the target of a guilt induction, because the inducer is presumably emphasizing his or her own expectations and standards. Appealing to such discrepancies may allow people to feel justified and decent while still acknowledging that someone made them feel guilty, as if they were to say that their behavior was objectionable in another’s view but nonetheless correct in their own estimation. The fact that one can be innocent in one’s own eyes but still feel guilty as a result of another person’s evaluation underscores the need for an interpersonal understanding of guilt.

In general, these data on the interpersonal genesis of guilt support the hypothesized relationship-enhancing functions of guilt. We observed ample evidence of the hypothesized function of guilt as an interpersonal influence technique: People induced guilt to get another person to comply with their wishes, and they recognized that others made them feel guilty for similar reasons. Furthermore, the single largest category of guilt inductions was failure to pay sufficient attention to relationship partners. People would inform a relationship partner that they felt neglected by him or her, and the partner would then feel guilty and subsequently spend more time and energy on maintaining the relationship. Finally, inequity was repeatedly implicated, such as when one person would make someone feel guilty by pointing out that one had done or invested more than the other.

One additional finding of this study was a very relevant insight into the experience of inducing guilt in someone else. We found that guilt inducers often felt better when their partner began to feel and show guilt. Meanwhile, of course, people began to feel worse when they themselves felt guilty. This pair of emotional shifts supports the view that guilt redistributes emotional distress within a dyad. During the transgression, perhaps, the transgressor enjoys whatever benefits there are while the victim suffers; however, if the victim can induce guilt in the transgressor, the transgressor’s state becomes less pleasant while the victim’s state improves. Moreover, several cases in our sample made explicit reference to the relationship-enhancing consequences of increased understanding and communication that resulted from guilt induction.

Some theories of guilt have emphasized intention and voluntary choice, but these concepts do not appear to be crucial for guilt. People feel guilty for accidental transgressions as well as for voluntary ones. McGraw (1987) found that people reported more frequent and more extensive guilt for accidental transgressions than for deliberate transgressions. This finding makes it necessary to take the emphasis off the intrapsychic act of choice in formulating a theory of guilt (see also Frijda, 1989). Clearly, intention can be one factor contributing to the link between the self and the victim’s suffering; therefore, intention can contribute to guilt (e.g., Graham, Doubleday, & Guarino, 1984) but should probably be considered a secondary factor rather than an essential one.

Thus, subjective reports of guilt experiences repeatedly attest to interpersonal contexts, causes, and consequences. When people describe events that made them feel guilty, these are almost always interpersonal events. Moreover, they report that other people can make them feel guilty and that they deliberately try to induce guilt in other people. Guilt often arises because different people hold differing expectations and one person’s well-meant actions violate another’s standards. Guilt also arises even over unintended transgressions. Interpersonal transgressions, neglecting to pay attention to relationship partners, and positive inequities are repeatedly implicated as causes of guilt.

Transgression, Compliance, and Altruism

Because guilt is most commonly associated with specific misdeeds, researchers have sought to learn about guilt by studying the consequences of transgressions for the transgressors themselves. In the 1960s, guilt was invoked to explain the finding that transgressions increase subsequent helping and compliance by the transgressor (Carlsmith & Gross, 1969; Freedman, Wallington, & Bless, 1967; see also Brock, 1969; D. T. Regan, Williams, & Sparling, 1972). From the start, it was apparent that this phenomenon was a pattern of interpersonal behavior; the transgression was typically an interpersonal act, as is helping others or complying with others’ wishes. Whether guilt was involved is more difficult to say, however. After the initial studies, subsequent findings questioned the guilt explanation by indicating that transgressions can stimulate altruistic behavior even among uninvolved observers (Cialdini, Darby, & Vincent, 1973; Konecni, 1972; Rawlings, 1968) and even toward recipients other than the victim (Darlington & Macker, 1966; Katzev, Edelsack, Steinmetz, Walker, & Wright, 1978; Rawlings, 1968). Some studies failed to find the effect (Noel, 1973; Silverman, 1967). Brock (1969) challenged the interpretations using guilt: “Guilt explanations can be abandoned without producing an abhorrent conceptual vacuum” (p. 145). The resultant ambiguity may have created a general reluctance among social psychologists to use the construct of guilt.

Making sense of this welter of findings is not easy, especially because they vary in many details of procedure, including the relationships among transgressor, witness, and victim; whether the transgression was an infliction of harm required by the experimental procedures or was the result of apparent negligence on the subject’s part; whether helping undid the damage; and, indeed, whether the putative transgressors perceived themselves as having done anything wrong. A further problem is that the direct evidence of guilt feelings, such as might be reflected in manipulation checks, was generally missing from most of these studies (see Brock, 1969). Guilt was inferred rather than measured directly, which clearly increased the potential ambiguity as to whether the studies were dealing with guilt at all. Finally, there was an odd theoretical vacuousness about the entire phenomenon; as McMillen (1976) observed, “there is no obvious reason why compliance should alleviate guilt” (p. 179).

One area of ambiguity that was regarded as theoretically decisive but that, unfortunately, turned out to be empirically inconsistent was whether transgressions increased help among bystanders who had no involvement and hence no reason to feel guilty. Several studies showed increased bystander helping (Cialdini et al., 1973; Konecni, 1972; Rawlings, 1968), whereas others found that bystanders did not help more after witnessing a transgression (Carlsmith & Gross, 1969; Freedman et al., 1967). One appealing resolution was offered by J. W. Regan (1971), whose findings suggested that guilt mediated the helping of transgressors, whereas bystanders and witnesses helped for other reasons (e.g., feeling sorry for the victim). Regan’s findings thus entail that guilt can be one mediator of helping after transgression even if, as other authors indicated, a guilt hypothesis is clearly inadequate to account fo all of the findings.

Once one accepts Regan’s conclusion, it becomes fair to ask why guilt should stimulate helping. An interpersonal view of guilt can easily answer this question: Helping is an act of communion, and guilt is based on a threat to communion; thus, helping is a subjective means of overcoming this threat. Helping can restore equity, repair possible damage to the relationship, and, in general, promote social attachment. Evidence confirms that, in ordinary life, helping is mainly associated with close, communal relationships (see Clark, 1983, for a review). Consistent with the interpersonal inequity view, there is some evidence that people want to help in proportion to their guilt. Berscheid and Walster (1967) demonstrated that harm doers prefer to do things for their former victims when the benefit would exactly match, and hence undo, the harm they did (as opposed to overcompensating or undercompensating). A precise compensation would remove the inequity and put the relationship back on an even footing.

The fact that transgressors may become helpful to someone other than their victim could have two relevant explanations. First, this may simply reflect some imprecise, biologically based proclivity. There would, after all, be survival value in having an impulse to help others after transgressing, and in many cases, of course, that impulse would be directed toward the victim, even if substitutes were occasionally accepted. Survival and reproduction might, in any case, be facilitated by an increase in helpfulness after transgression, if only because such an increase would keep the transgressor from losing his or her social ties. Although helping intimates would be most important and valuable, even helping a stranger might be beneficial because it might increase the chances of forming a positive bond with the stranger that could be useful if one’s victims return to seek revenge.

Second, if guilt is linked to separation or exclusion anxiety, then it may stimulate prosocial behavior as a means of promoting any sort of communal attachment. As noted earlier, laboratory subjects seem to adopt a communal approach to interactions even with newly met strangers, unless these strangers signify that they are unavailable for such a relationship (Clark, 1984). From the point of view of the individual, a threat to social ties would produce an affective state (guilt) that would motivate behavior (altruism) that will generally tend to strengthen social ties. Again, the benefiting relationship would often be the same one that is threatened, but substitutes would be acceptable. This hypothesis could be tested by evaluating whether guilt stimulates attachment behavior such as in-group loyalty, affiliative behavior (even involving people unrelated to the guilt), and so on. Some support for this hypothesis was offered by Meindl and Lerner (1984), but a more direct test is desirable.

The importance of the interpersonal context was attested in a different way by Wallace and Sadalla (1966), who showed that only publicly recognized transgressions led to a significant increase in helping. Subjects who committed similar transgression but were not caught were not especially helpful. Ironically, these authors noted that several subjects confessed spontaneously but then did not comply with the request for help. This suggests that not only the public recognition of the transgression but the detection by others is decisive. If the transgressor confesses, there is no perceived need to help; if others catch and reproach the transgressor, however, helping does increase, presumably because the self-presentational implications of being caught are radically different (and much more damaging to a communal attachment) from those of confessing spontaneously. This suggestion also fits the results of Katzev et al. (1978), who found that transgressions alone failed to increase helping, whereas public reprimands for identical transgressions did lead to subsequent helping. Likewise, Silverman (1967) noted that his findings with children and Freedman et al.’s (1967) with college students confirmed the view that transgression alone fails to produce helpfulness. The interpersonal context of the transgression appears to be a potent moderator of the transgression–compliance pattern. Inequity alone is not enough.

Thus, if guilt does indeed mediate between transgression and helping, these findings about social context seem to require an interpersonal view of guilt. Only when guilt has interpersonal reality does it stimulate helping. Furthermore, these results seem to underscore the hypothesized function of guilt as motivating prosocial, relationship-enhancing behavior.

Moreover, the argument that transgression can stimulate helping even among observers and even toward nonvictims is consistent with Hoffman’s (1982) suggestion that guilt is based on empathic distress in response to another’s suffering. Guiltbased helping could thus be seen as one pattern deriving from the more general response of empathic concern for someone who suffers. J. W. Regan’s (1971) conclusion that transgressors and bystanders both become altruistic but for different reasons is consistent with the view that there is a common base in empathic distress but the mediating mechanisms differ as a function of personal causal responsibility. The hypothesis of common base is also supported by M. D. Harvey and Enzle’s (1981) finding that norms regarding helping are closely related to norms regarding transgression.

Cunningham et al. (1980) showed that guilt increased helping only in response to requests that played on a sense of obligation. Positive moods made people want to help in response to requests that emphasized the desirability of helping (specifically, by claiming that a donation to the World Children’s Fund would “help keep the children smiling”). In contrast, an experimental guilt manipulation increased donations only when the requestor said “You owe it to the children.” Thus, the capacity of guilt to promote helping was dependent on generating a sense of interpersonal debt.

In conclusion, it is apparent that transgressors tend to become more helpful and compliant after a transgression, and guilt is one of several factors implicated in that response. The phenomenon was itself inherently interpersonal; in the typical research paradigm, the transgression was interpersonal, as was the result (helping someone). Furthermore, interpersonal empathy, a sense of interpersonal obligation, self-presentational concerns with how one is viewed by others, and recognition of the transgression by others appear to be potent mediators of the link between transgression and compliance. Transgressions apparently increase helping by motivating people to strengthen and reaffirm social attachments, to undo relationship damage, and to rectify interpersonal inequities.

Transgressions: Other Findings

Helpful compliance is not the only result of a transgression, and several studies have examined further consequences. We have already touched briefly on the study of Meindl and Lerner (1984), who showed that transgressors subsequently sought to boost the relative status of their in-group. More specifically, transgressors adopted attitudes toward an out-group that ranged from hostile and aggressive to benevolent; the common thread was that the interaction occurred on the basis of a status differential in favor of the in-group’s superiority (Meindl & Lerner, 1984, p. 80). If these results were indeed mediated by guilt, then guilt appears to promote a wish to regard one’s group as superior.

Katz, Glass, and Cohen (1973) showed that subjects derogated out-group (different-race) confederates to whom they had delivered severe shocks, and this derogation was greater than the derogation of either mild shock victims or same-race severe shock victims. Ratings of guilt were equally high in the severe shock condition regardless of the victim’s race, however. Katz et al. explained this seeming anomaly by suggesting that they might have had a ceiling effect in guilt ratings. Apparently, these authors assumed that the maximum derogation must have arisen from maximum guilt, contrary to their data. It may be, however, that their data meant exactly what they indicated, namely that guilt only sometimes leads to derogation. Apparently, derogating the victim was a preferred means of handling the guilt when the victim was from an out-group (and hence no communal relationship would be expected), whereas people may be less willing to resort to that method for an in-group member. Derogating the out-group may be partly a means of affirming one’s ties to the in-group, which helps deal with guilt by strengthening one’s sense of belongingness. Derogating a member of one’s own group would not, of course, strengthen any sense of belongingness, which would make it a less appealing strategy for dealing with guilt.

Taken together, these findings suggest that guilt motivates people to affirm their social bonds, not only by increasing attachment to their own social group but by derogating members of other groups relative to their own group. These results fit the view of guilt as an interpersonal phenomenon associated with threats to relationships.

Kugler and Jones (1992) developed scales to measure trait guilt, state guilt, and moral standards separately, on the assumption that having moral standards had been confounded with guilt in some prior measures. In their view, morality and guilt are not the same and therefore require separate measures. Jones et al. (in press) examined how these disparate measures correlated with the perceived wrongfulness and self-reported frequency of various transgressions, which they sorted into relational and nonrelational categories. Relational transgressions involved direct harm to a significant other, such as in lying to a friend, betraying a confidence, or hating a family member. Nonrelational transgressions, in contrast, primarily involved mere violations of standards without apparently harming anyone (e.g., masturbation, marijuana use, and looking at nude pictures), although some might involve interpersonal aspects (e.g., in plagiarism). Clearly, the relational transgressions were much more relevant to communal relationships than were the nonrelational ones, and so they should be associated with higher levels of guilt.

State and trait guilt were heavily and pervasively correlated with both frequency and perceived wrongfulness of these relational transgressions. Nonrelational transgressions, in contrast, showed only weak and occasional correlations. (The measure of moral standards correlated with both types of transgressions, consistent with Jones et al.’s [in press] view that having moral standards is not the same as experiencing guilt.) Thus, the findings of Jones et al. indicate that guilt is mainly associated with transgressions that involve harm to a relationship partner.

Graham et al. (1984) showed that young children feel guilty over accidental outcomes but older children increasingly link guilt to contollability. Indeed, among adults, controllability appears to be a significant factor in producing guilt (Weiner, Graham, & Chandler, 1982). Controllability presumably implies responsibility, and so these findings suggest that issues of perceived responsibility should be retained in guilt theory, particularly for understanding guilt phenomena in maturity. They suggest that guilt originates developmentally from something quite different from principled self-evaluation but that people, as they grow up, gradually assimilate their guilt reactions to such principles. These findings fit Kohlberg’s (1981, 1984) suggestion that moral judgment evolves through stages, and so people gradually come to use abstract moral principles in their judgments more and more. Even if the roots of guilt are not in violation of abstract standards, people may well gradually learn to discuss, justify, defend, and judge their actions in those terms.

Guilt Without Transgression

Research has examined a number of phenomena in which people report feeling guilty despite the absence of wrongdoing. These phenomena are of particular importance for understanding the boundaries of guilt and the link between guilt and behavior. Many of them point to a deeply ingrained preference for fairness and equity in interpersonal relations, and guilt may be one potent mechanism that motivates that preference.

The phenomenon of survivor guilt (e.g., Lifton, 1967; Niederland, 1961) makes clear that people can feel guilty without having done anything wrong or, indeed, having done much of anything at all. One simply feels guilty for surviving when others died. It is ironic that this guilt often surfaces in victims of oppression and misfortune, such as survivors of Hiroshima and the Holocaust. Despite their own suffering, they feel guilty vis-à-vis others who suffered even more extremely. More recent research has found evidence of survivor guilt among homosexual men who have tested negative for the human immunodeficiency virus that is, among gay men who have been “spared at random” (Wayment, Silver, & Kemeny, 1993) from the lethal acquired immunodeficiency syndrome (AIDS). They appear to feel guilty vis-à-vis their fellow gay peers who do suffer and die, and, indeed, Wayment et al. found that the more sexual partners one has had, the more one tends to feel guilty over not contracting AIDS. These feelings of guilt were significantly reduced, however, among gay men who maintained a strong social support network and were active members of gay organizations, suggesting that strong integration into the community reduces one’s vulnerability to guilt. These correlational data do not permit causal inferences, but they are consistent with the hypothesis linking guilt with a perceived threat to one’s communal relatedness.

Other recent work has extended the concept of survivor guilt to people who keep their jobs when co-workers are fired (see Brockner, Davy, & Carter, 1985). Self-serving biases, positive illusions, just world beliefs, and other patterns predict that the lucky few would quickly conclude that their less fortunate co-workers deserved their fate, but evidence suggests that such mechanisms are, at best, only partly effective. In particular, many organizational shake-ups entail reducing the number of employees, and the fate of particular individuals is felt to be largely determined by external factors, including luck. The losers suffer, often acutely (see Newman, 1988), but those who remain employed feel guilty toward them. Thus, friendships and other social relationships between co-workers are often strained or terminated when one is fired (see Newman, 1988).

Experiments conducted by Brockner et al. (1985, 1986) have demonstrated survivor guilt. Subjects were recruited for these studies with the promise of course credit; on arrival, however, they were told that space limitations meant that not all could complete the studies and some would be sent away without credit. On the basis of either a bogus random draw or an ostensible assessment of performance, a confederate was dismissed and departed after protesting the unfairness of the layoff. Subjects reported feeling guilty about being allowed to stay while the confederate was dismissed, and, in some conditions, subjects derogated the confederate or increased their own subsequent work output.

Similar patterns have been shown with inequitable misfortune. Tesser and Rosen (1972; replicated by Johnson, Conlee, & Tesser, 1974) showed that people feel guilty when they are not sharing the misfortune of another. Johnson et al. (1974) showed that such guilt feelings are correlated with a reluctance to transmit bad news.

Survivor guilt appears to attach not to any particular action but merely to the inequity of one’s outcome in relation to the outcomes experienced by family members, friends, and co-workers. This guilt is based on a social relationship, insofar as one feels guilty about inequities in one’s favor in comparison with significant others. This form of guilt therefore seems well suited to support a preference for equitable relations between people. To be sure, if the intimate partner has actually died, then there is no point in trying to enhance the relationship with that person; however, it seems appropriate to consider survivor guilt as an extreme symptom of a more general pattern in which people feel guilty over positive inequities, and this general pattern would presumably be extremely beneficial for promoting fair, equitable, and hence strong and durable relationships.

One prediction of the interpersonal view is that guilt should be strongest in close, communal relationships. If guilt is based on attachments, then it should occur with greater power and frequency in the context of long-term, emotionally intense relationships marked by fundamental concern for the other’s welfare. Studies of survivor guilt support this view (e.g., Niederland, 1961). Observations of Holocaust victims and Hiroshima victims suggest that survivor guilt is felt most strongly and commonly vis-à-vis family members, relatives, and intimates. Thus, for example, Lifton (1967) observed that this form of guilt was “most focused and intense in relationship to family members” (p. 38; Lifton did not, however, offer a quantified or statistical basis for this conclusion). After all, in principle anyone could feel survivor guilt when considering unfortunate victims from all over the world, but in practice few people have powerful guilt feelings about the sufferings of distant, unrelated others.

The Brockner studies showed that people could feel an increase in guilt over the layoff of someone with whom their relationship was minimal, namely a fellow student and fellow experimental participant with no prior acquaintance. Although clearly no established communal relationship existed between the subject and confederate in those studies, there is no way of knowing whether there was any sense of communion at all (e.g., from being fellow students at the same university and taking the same courses). As noted earlier, laboratory subjects are inclined to adopt a communal approach toward fellow subjects unless circumstances or the other’s actions overtly discourage that approach (e.g., Clark, 1984, 1986), and so some sense of the fellow subject as a potential communal partner may have existed (enough to create some level of guilt). On the other hand, Schmitt et al. (1989) showed that people feel some guilt over the misfortunes of people in other countries. In our view, these findings reflect the generality of effective socialization that has taught adults to strive for equity and mutual concern in all relationships. The fact that survivor guilt continues to be strongest vis-à-vis family members and other intimate partners is indicative of an interpersonal, communal basis of the phenomenon.

If inequity can cause guilt, then one should be able to generate guilt without causing any suffering. Research on inequity has shown that people can feel guilty simply because the rewards they receive are larger than what someone else receives. Austin, McGinn, and Susmilch (1980) found that people felt guilty when they received a larger reward than someone else who had performed equally well. Their findings are important because there was no transgression at all, either voluntary or involuntary; guilt arose simply from the sense of having benefited inequitably. Likewise, Hassebrauck (1986) found that subjects associated guilt with inequitable advantages and benefits rather than with inequitable disadvantages and costs. These findings, in particular, show the need to extend guilt theory beyond notions of intentional, voluntary transgression to encompass a central role of interpersonal inequities in causing guilt.

Additional evidence of guilt without a clear transgression was provided in a recent study of unrequited love (Baumeister & Wotman, 1992). People who found themselves to be the objects of someone else’s romantic attentions and did not reciprocate these feelings became the reluctant agents of the would-be lover’s heartbreak. According to their accounts, they had done little or nothing to encourage the amorous attentions of the other person and were morally innocent, but they felt guilty afterward anyway. The relatively high rate of self-justifying statements in the rejectors’ accounts likewise suggests that they struggled with guilt feelings that were not dispelled by their own arguments of technical, moral innocence. To be sure, people who had encouraged the would-be lover felt especially guilty, and so self-perceived responsibility may increase guilt; however, guilt feelings arose even in people who regarded themselves as not having done anything wrong. These findings suggest that the infliction of harm on another person, rather than any intentional or self-perceived wrongdoing, creates guilt. Moreover, they seem consistent with the hypothesized link between guilt and empathy, because people probably experience an empathic understanding of someone who loves them.

Finally, there are some suggestions that even victims of misfortune feel guilt. Empirical support of victim guilt has been limited to clinical observations and impressions, so its very existence could be disputed if the phenomena can be reinterpreted. Still, if victim guilt could be established, the implications for guilt theory would be substantial. For example, Lamb (1986) reported that childhood victims of incest and sexual abuse often feel guilty (although guilt and shame may be confounded in some observational studies), partly because parents may overtly blame them. Children may also feel guilty about the family crisis that results when they reveal the abuse to others. According to Lamb, therapy for incest victims often revolves around convincing these victims that they are not to blame for the event and hence should not feel guilty.

Victim guilt cannot readily be analyzed as a result of positive inequity, because the victim suffers a negative inequity rather than a positive one. Still, interpersonal causation is supported by suggestions that transgressors often blame their victims, thereby causing the victims to feel guilty. There is little reason to expect that solitary victims would feel guilt, although more systematic research is needed to verify the importance of an interpersonal context for victim guilt.

Thus, it is apparent that transgressions are not always involved in guilt, and, in fact, people may be able to feel guilty without believing that they have done anything wrong. The perception of oneself as responsible for a transgression is not necessary to produce guilt feelings. Many people feel guilty over positive inequities in their favor and over hurting someone unintentionally. Moreover, it appears that social attachment and a sense of communal relatedness are important determinants of the magnitude and power of guilt.

Situational and Contextual Predictors

The relationship context is an important determinant of guilt. Millar and Tesser (1988) examined guilt over lying and deception as a function of one’s own expectations about proper behavior, one’s parent’s expectations, and one’s employer’s expectations. Guilt appeared to be specific to the relationship involved. Thus, the employer’s expectations about proper behavior had little impact on whether the person felt guilty about lying to his or her parents, and the parents’ expectations had little impact on whether the person felt guilty about lying to the employer.

Furthermore, guilt appeared to arise mainly when there was a match between one’s own expectations and those of the victim (i.e., the person to whom one lied). Millar and Tesser (1988) noted that guilt depended on a concurrence of self-judgment and the other person’s judgment. This finding suggests one direction for integrating interpersonal and self-evaluation aspects of guilt, namely that guilt is an interactive product of self-judgment and judgment by one’s victim or partner.

The importance of relationship context for guilt was further corroborated by Tangney, Marschall, Rosenberg, Hill-Barlow, and Wagner (1992). These authors found that guilt was typically experienced in the presence of other people (but fewer other people than shame or pride). One apparent implication is that the role of others in pride may be to serve as a kind of “generalized other” (i.e., as an undifferentiated representative of other people and of the social community), whereas guilt tends to be tied to the presence of specific other people. The relevance of particular other people for generating guilt suggests again that guilt tends to be relationship specific.

Zahn-Waxler and Kochanska (1990) summarized several studies indicating that parental warmth and affection (rather than rejection and punishment) are positively linked to guilt development in the child. These findings raise serious difficulties for the view of guilt as the expectation of punishment (cf. Mosher, 1965). Zahn-Waxler and Kochanska also reported that love withdrawal is linked with high guilt. When parents instead respond by asserting their power, as in direct punishment, guilt is less extensive. This combination of findings attests to the importance of strong, positive interpersonal attachments in the genesis of guilt and, in particular, supports the hypothesis that anxiety over exclusion from important social relationships is a prominent cause of the emotional distress out of which guilt evolves.

Intriguing findings by Okel and Mosher (1968) suggest another relevant situational cause of guilt. These authors induced undergraduate male students to derogate a fellow student (actually a confederate), who then became quite distressed. Subjects who were prone to guilt, on a trait measure, reported feeling guilty about the event. They also reported that they would not have performed those actions had they known how upset the victim would be. Thus, the interpersonal consequences produced the guilt. Subjects were saying, in effect, that the same action would produce different levels of guilt depending on its interpersonal impact (i.e., on how someone else would react to it).

Two studies that we conducted suggest additional links between guilt and interpersonal factors (Baumeister et al., in press). In the first, subjects were asked to write accounts of events in which some other person judged their behavior negatively and responded with anger; half the subjects were instructed to describe incidents about which they did feel guilty, and the other half were instructed to describe incidents about which they did not feel guilty. One significant difference between the guilty and the not-guilty stories was the relationship to the other person: People felt more guilty about transgressions involving an esteemed other than about transgressions involving someone they held in low regard. This too fits the interpersonal view that the relationship bond to the other person (i.e., the victim) is a crucial determinant of guilt. For purposes of this article, we recoded those data using the criteria for communal versus exchange relationships defined by Clark (1984). Nearly all (93%) of the accounts of transgressions about which subjects felt guilt involved communal relationships, whereas only 69% of the accounts of transgressions that did not lead to guilt involved such relationships; the difference was significant, χ(1, N  = 86) = 8.05, p  < .001. These findings support the hypothesis that guilt is linked to communal relationships.

In our second study, subjects were asked to write accounts of making someone feel guilty or of being made to feel guilty by someone else. This study was thus explicitly concerned with the interpersonal genesis of guilt and was useful for testing one tenet of the interpersonal theory, namely that interpersonal manipulations of guilt will primarily involve people in close relationships. The other person’s relationship to the subject was coded into four categories: close relationship, casual acquaintanceship or ongoing role-oriented interaction, stranger with some passing role relationship, and stranger with no relationship. The fourth category was completely empty, and only 1 story fit the third. Six incidents fell in the second category, and the remaining 95 were in the close relationship category. Thus, it appears that interpersonal manipulations of guilt mainly occur in close relationships, consistent with the interpersonal view.

A series of studies conducted by Vangelisti et al. (1991) has also confirmed the overwhelming tendency for guilt induction to occur mainly in the context of close relationships. Their subjects were asked to recall a recent conversation in which someone made them, or they made someone, feel guilty. None of the conversations reported by their subjects were with strangers; 80% took place in the context of very close relationships, and the remaining 20% were in “somewhat close” relationships (p. 16). In a second study, subjects were asked how typical it was that a particular person induced guilt in them, and typically correlated .76 with the intimacy of the relationship. They also noted that two of the main reasons for inducing guilt were to influence the behavior of the other person and to maintain relationships (e.g., by motivating the other person to devote more time and attention to the relationship).

Thus, research on situational causes of guilt has established several interpersonal aspects. Guilt appears to depend on particular relationship contexts and to be experienced most commonly in the context of close, communal relationships. Both the nature of the relationship and the specific interpersonal consequences of the transgression appear to be decisive determinants of guilt.

Personality Correlates of Guilt

Empathy

Proneness to guilt is empirically correlated with empathy (Tangney, 1989, 1991), such that more empathic people are more likely to experience guilt than are less empathic people. Tangney’s (1991) data suggested that the correlation between the two tends to be around .4, fluctuating somewhat as a function of measure and sample.

Empathy is, by definition, a form of interpersonal sensitivity (e.g., Eisenberg & Miller, 1987; Hoffman, 1981, 1982), and the fact that guilt is linked with such sensitivity strongly suggests an interpersonal aspect for guilt as well. Likewise, empathy could be relevant to guilt arising from positive inequities, because empathy could help people appreciate the other person’s disappointment. These interpersonal implications are underscored by the fact that empathy is strongly linked to success in close relationships. Highly empathic children are much more popular than others (Walden & Field, 1990), and empathic adults have more satisfactory intimate relationships than others (Davis & Outhut, 1987). Zahn-Waxler, Kochanska, Krupnick, and McKnew (1990) found that young children’s reports of guilt were associated with empathy and concern about good interpersonal relationships.

Indeed, Hoffman (1982, p. 304) suggested that guilt develops out of empathy and that the two are “the quintessential prosocial motives.” As Hoffman explained, guilt has its roots in a distress response to the suffering of others. Yet all others are not the same. People respond empathically more to similar others than to dissimilar others (e.g., Batson, Duncan, Ackerman, Buckley, & Birch, 1981). Even the responsive crying of neonates, which may be the developmentally earliest form of empathic response (because it happens in the first days of life; Sagi & Hoffman, 1976; Simner, 1971), is selective; neonates cry in response to the crying of other human neonates but cry less in response to the crying of chimpanzees or older human children or to synthesized crying (Martin & Clark, 1982; Sagi & Hoffman, 1976; Simner, 1971). Empathy is thus selective in that people are most likely to respond empathically to those who are like themselves in some way. Zahn-Waxler and Kochanska (1990) also provided evidence suggesting that the development of empathy is linked to the development of guilt.

Interpersonal traits

Recent work has linked guilt to interpersonal traits. Jones and Kugler (in press) found significant positive correlations between guilt (both trait and state) and loneliness, shyness, resentment, suspicion, and anger, as well as negative correlations between guilt and satisfaction with social support. They concluded that guilt is strongly linked to dissatisfaction with relationships and possibly linked to low relational competence. They also confirmed that merely having high moral standards is unrelated to trait and state guilt and unrelated to these interpersonal variables. Thus, the role of interpersonal factors in guilt apparently has much more to do with relationship problems and transgressions than with merely learning high standards from others.

The trait correlates of a tendency to induce guilt in other people were studied by Vangelisti et al. (1991). They found that use of guilt was inversely related to several measures of social assertiveness and positively related to shyness. Meanwhile, it was positively related to several measures of interpersonal aggressiveness, such as irritability, resentment, and suspicion. The authors concluded that people who lack social assertiveness may seek to get their way by inducing guilt rather than using more openly coercive or confrontational strategies. These results fit well with our hypothesis that guilt functions as an influence technique that may allow less powerful people to get their way in relationships.

Sex differences

Gender differences in guilt have also been found. Contrary to Freudian hypotheses suggesting higher guilt among males, Hoffman (1975) found that moral transgressions led to higher levels of guilt in females; males were more likely to respond with fear of being caught. Likewise, Tangney (1990) found higher levels of guilt among women and girls than among men and boys. Thompson and Hoffman (1980) found that boys (in Grades 1, 3, and 5) reported more intense guilt than girls over imaginary transgressions, and Silverman (1967) found that girls were more likely than boys to engage in mild (but not substantial) cheating; however, if there is any tendency for boys to be more affected with guilt than girls, it is apparently reversed by the time children reach adolescence. Zahn-Waxler and Kochanska (1990) likewise noted that the Freudian view would predict more guilt among males (see also Lewis, 1971), but almost all of the evidence is in the opposite direction (e.g., Frodi, Macaulay, & Thome, 1977). With children, the effects are less consistent, but still there is no strong indication of boys feeling more guilt than girls.

The sex difference of higher guilt in women is not large but is significant. Hoffman (1975) found evidence that women’s greater concern and consideration for others was linked to their greater vulnerability to guilt over interpersonal transgressions. He pointed out that in many laboratory paradigms, females have been found to be more compliant than males; however, when compliance involves harming others, females comply less than males. Thus, the sex differences in guilt may reflect sex differences in interpersonal concerns and patterns. These are particularly relevant to the present analysis, because the higher levels of communal concern (e.g., nurturance and interest in others’ welfare) among females (e.g., Block, 1973; Eagly, 1987; C. Gilligan, 1982) should be related to higher guilt proneness if guilt is indeed linked to communal relatedness.

Self-esteem

Tennen and Herzberger (1987) showed a negative correlation between self-esteem and guilt proneness; that is, low self-esteem is associated with a high tendency to feel guilty (see also Vangelisti et al., 1991). The meaning of this finding is far from clear, of course; perhaps low self-esteem causes guilt, or perhaps frequent guilt causes low self-esteem. Although self-esteem involves patterns of self-knowledge and self-appraisal, it also encompasses interpersonal patterns (see Baumeister, 1993; Baumeister, Tice, & Hutton, 1989; Baumgardner, 1990; Campbell, 1990), which could help explain differences in guilt. Moreover, people with low self-esteem are socially pliable and malleable (Brockner, 1983), and this receptivity to influence may make them vulnerable to others’ standards and expectations so that they feel guilty more often when violating the implicit demands of others. To regard oneself badly is to regard oneself as an undesirable partner for others, and such a self-view could make people especially sensitive to the danger of alienating potential partners by committing transgressions (i.e., could make them especially guilt prone). Interpersonal inequities could also be relevant: Low self-esteem would presumably entail a reduced sense of deservingness, so identical outcomes would strike the person with low self-esteem as positive inequity more often than the person with high self-esteem.

Feelings about the self

Guilt also seems linked to a desire to enhance or recover self-esteem, suggesting that one aspect of guilt is a temporary loss of esteem. McMillen (1976) found that receiving a flattering message removed the tendency for recent transgressors to become altruistic, which suggests that boosting self-esteem and helping others have similar functions for reducing guilt. Meindl and Lerner (1984) found that transgressors derogated out-group members in ways that would enhance the esteem of their in-group. Wertheim and Schwartz (1983) found that guilt made people increase their effort on a subsequent performance, which could well be a strategy of compensatory self-enhancement (Baumeister & Jones, 1978; Greenberg & Pyszczynski, 1985).

All of these effects could be due to some emotional process rather than reflecting an emotionally neutral process of cognitive self-appraisal. The reason for McMillen’s (1976) effect on helping could be that getting a good evaluation puts one in a good mood, and, as Cialdini et al. (1973) showed, being put into a good mood after a transgression removes the tendency to help. Cunningham et al. (1980) showed that a positive mood induction prevented a subsequent guilt induction from increasing helpfulness. Likewise, a success experience, or sense of belonging to a desirable, esteemed group, may bring an emotional boost that could erase the bad feeling associated with guilt.

Conclusion

Various personality factors are correlated with guilt, and these correlations are consistent with an interpersonal approach to guilt. Guilt is strongly linked to empathy, which is a form of interpersonal sensitivity. Interpersonal traits such as relational competence, shyness, lack of social assertiveness, and loneliness predict guilt. Links to self-esteem may or may not have substantial interpersonal implications. Evidence about sex differences in guilt is consistent with our emphasis on communal relatedness.

Consequences of Guilt

The effects of guilt may shed valuable light on its causes and essential nature. We next examine evidence on the desire for punishment, prosocial effects, and antisocial effects.

Desire for punishment

Freud suggested that guilt is derived from “fear of the super-ego” that produces a desire for punishment (1930/1961, p. 74); indeed, he equated the terms “unconscious need for punishment” and “unconscious sense of guilt” (1933/1964, p. 135) and proposed that such needs often caused people to become ill and to resist therapy (1933/1964, p. 136). The notion that guilt stimulates a desire for punishment has been maintained in subsequent Freudian theorizing about guilt (e.g., Bulka, 1987; Piers & Singer, 1953/1971; see also Menaker, 1982, chap. 5), and it has been widely cited in connection with masochism and self-defeating behavior (e.g., Menninger, 1934/1966; Panken, 1983).

Empirical work has largely failed to demonstrate that guilty people wish to suffer or be punished. There is some evidence that guilt feelings are associated with the expectation or anticipation of punishment (e.g., Wicker et al., 1983); however, expectation is far from desire, and, insofar as all transgressions violate norms (and hence risk sanctions), it is hardly surprising that guilt would often be correlated with anticipation of punishment. Freedman (1970) and Walster et al. (1976) noted that empirical studies had failed to demonstrate that guilt stimulates a wish for punishment, and that appears to be still true today, probably despite multiple efforts (see also Silverman, 1967). Meanwhile, masochism and self-defeating behavior have been reinterpreted in light of new findings to reflect escapism, judgment errors, and trade-offs of positive and negative outcomes rather than self-destructive wishes (see Baumeister, 1988, 1989, 1991; Baumeister & Scher, 1988; Scott, 1983; Weinberg & Kamel, 1983).

Our literature search yielded two findings that come closest to demonstrating a desire to suffer arising from guilt. Wallington (1973) induced subjects to respond falsely to a question (which she regarded as lying) and then used a mislabeled shock machine to have them deliver moderate shocks to themselves. The finding that induced liars self-administered higher levels of shock than nonliars contradicted an earlier study using a similar procedure (Freedman et al., 1967; see Brock, 1969) that showed that transgression failed to increase preference for an unpleasant task as opposed to a pleasant one. There was no direct evidence that Wallington’s subjects felt guilty or recognized any transgression, and, from reading the procedure, one doubts that subjects did. That, along with further ambiguities in the procedures (e.g., the question was constructed such that the modal answer was at least partly truthful), makes it difficult to accept these findings as evidence that guilt creates a desire for punishment.

The other relevant finding was provided by Wertheim and Schwartz (1983), who showed that guilt made people prefer immediate over delayed punishment. As Wertheim and Schwartz noted, however, this was not a desire for punishment per se but a desire to get it over with; as such, it may be interpreted as a choice of the lesser of two evils. Indeed, this preference for immediate punishment may be regarded as an adaptive response because it minimizes suffering through anticipation, fear, and other forms of intervening distress. More generally, Wertheim and Schwartz’s finding supports the view that guilty people want to put the incident behind them and escape from the unpleasant guilty state (see also Baumeister, Stillwell, & Wotman, 1990, on temporal bracketing). There was no evidence of an increased desire to suffer or to be punished.

Further work is clearly needed to resolve these ambiguities and ascertain whether, indeed, there is any overt desire to suffer among liars and other transgressors. If there is, it would presumably not be difficult to document, and one could use serious transgressions and genuine suffering rather than the ambiguous behaviors and nonpainful discomfort used by Wallington (1973). We suspect that if guilt did indeed promote a desire to suffer, this would have been of sufficient interest and complexity that the research literature would contain a substantial number of studies on it, and the absence of such evidence may be due to many failures to find such effects.

Some might suggest that findings from other areas can be interpreted to indicate that guilt creates a wish to suffer. Murder–suicide combinations have, in particular, been interpreted psychodynamically in this way: Specifically, the notion is that someone first murders someone else, then feels guilty over the murder, and so kills himself or herself as punishment for the misdeed. In this view, the suicide is a secondary act, deriving from guilt over the murder. Recent studies have concluded, however, that the reverse is generally true; that is, the suicide is the primary act and the murder is secondary or derivative (e.g., Hendin, 1982; Rhine & Mayerson, 1973; see also Allen, 1983; Berman, 1979; Palmer & Humphrey, 1980). Typically, the suicidal person feels victimized by another person, often a partner in a close relationship, and decides to commit suicide to escape from the intolerable situation; killing the other person functions as a kind of revenge for one’s own death. Thus, murder–suicide combinations do not generally conform to a pattern of guilt creating a wish for punishment.

In retrospect, perhaps it was never really plausible that guilty people want to suffer. People such as criminals who are guilty in an objective or legal sense (and some of whom would presumably also feel guilty) rarely seem to desire punishment. Indeed, criminals often hire lawyers and do whatever they can to avoid or minimize punishment, even if they have, in fact, committed the transgression of which they are accused. Our own accounts of episodes about which people feel guilty have not contained any references to wishing or choosing to suffer. Meanwhile, people who do seem to seek out pain or suffering (such as masochists) do not appear driven by guilt (see, e.g., Baumeister, 1989; Scott, 1983). Research on masochism and self-defeating behavior has failed to reveal any guilt-driven wish for punishment.

Yet another seeming contradiction of the desire-for-punishment hypothesis is a pattern of findings indicating that guilt seems to promote the opposite motivation, namely the desire for a self-enhancing experience of belonging to a superior, esteemed group (Meindl & Lerner, 1984) or achieving success (Wertheim & Schwartz, 1983). In fact, receiving flattering feedback seems to eliminate some of the behavioral effects of guilt (McMillen, 1976).

Prosocial effects

Guilt does appear to promote a variety of prosocial effects. In particular, guilt often motivates people to make reparations or amends or at least to apologize (Friedman, 1985; Hoffman, 1982; Lewis, 1971; Zahn-Waxler & Kochanska, 1990). These interpersonal actions presumably reduce the feeling of guilt. Several of the transgression–altruism findings (e.g., Carlsmith & Gross, 1969; Freedman et al., 1967; J. W. Regan, 1971) suggest a similar conclusion: Guilt makes a person want to do something for someone, but only if the transgression has a strongly interpersonal aspect (Wallace & Sadalla, 1966). Compensating the victim is, of course, especially appealing to the guilty transgressor (Berscheid & Walster, 1967). One could conceivably interpret apologizing and making amends as indicative of a desire to suffer, but it seems more plausible to interpret them as efforts to remedy interpersonal damage and to restore equity.

O’Malley and Greenberg (1983) showed that the expression of guilt and remorse is itself a means of helping the relationship recover from the transgression. Their subjects indicated that transgressors who felt guilty did not need to make as many material sacrifices or restitutions to complete the reparations for their misdeed as did transgressors who did not feel guilty. O’Malley and Greenberg used the term down payment effect to describe these results, because the guilt feelings appeared to function as a down payment (i.e., an initial, partial payment) of the interpersonal debt. This finding is particularly relevant to the interpersonal approach, which proposes that guilt feelings themselves can function to repair damage to a relationship arising from a transgression.

Guilt also appears to stimulate another interpersonal behavior pattern, namely confession (e.g., Karniol, 1982). Lawson (1988) found that some people feel moved by guilt to confess marital infidelity to their spouses, presumably in the (often misplaced) hope of being forgiven. Confession has a variety of positive effects, including increased calm and improved health in the confessor (see Pennebaker, 1989, 1990). Interpersonally, confession is a means of acknowledging one’s debt to another person and, one hopes, accepting the necessity of repairing any damage to the relationship. Thus, from an interpersonal perspective, it is the remorse and the positive commitment to the relationship contained in the confession that may be crucial. If people were to describe their transgressions to their victims in a boastful, sarcastic, or casual fashion, presumably the relationship would not benefit. Guilt should therefore motivate people to confess their misdeeds in a way that expresses a contrite commitment to improve or preserve the relationship (especially by not repeating the transgression).

It is also noteworthy that the appeal of confessing seems limited to communal relationships. In an exchange relationship, admitting a misdeed merely puts one at a disadvantage and decreases one’s future expected outcomes relative to inputs. Confession does not, after all, result in material or tangible benefits. In a communal relationship, however, one may confess one’s misdeed out of concern for the integrity of the relationship (e.g., to preserve assumptions of honesty) and to apprise the partner that he or she does deserve added benefits or specially desirable treatment in the future. One removes the uncertainty about the relationship’s future (e.g., if the partner finds out, he or she may terminate the relationship) at the cost of precipitating the crisis; again, this would make no sense in regard to an exchange relationship, but in a communal relationship any such uncertainty may be a pervasive cause of anxiety, and so confessing might be a way of ending that distress.

The common root of the altruistic and confessional motives is attested by their substitutability; that is, confession apparently reduces the urge to help. Harris, Benson, and Hall (1975) showed that people made larger donations to a requestor before going to church than when returning home after confession.

Our own data comparing guilty and not-guilty accounts suggest additional prosocial effects of guilt (Baumeister et al., in press). When people felt guilty, they were more likely to report learning lessons and subsequently changing their behavior. These results suggest that guilt is a valuable mechanism by which people alter their behavior to conform to others’ expectations and possibly to abstract norms and standards, consistent with the hypothesized relationship-enhancing functions of guilt. It is worth adding, however, that the interpersonal view portrays guilt as an effective but potentially costly technique for influencing others. We found that guilt did make people comply with others’ wishes, but they also exhibited resentment of the guilt induction. Rubin and Shaffer (1987) showed that observer subjects likewise reported that they would probably comply with a guilt-inducing request but would also resent the request and be likely to think less favorably of the requestor.

Thus, for the most part, it is apparent that guilt stimulates prosocial actions and motives. Guilt may therefore benefit communal relationships in many ways, and indeed an absence of guilt may be associated with patterns that could be detrimental to relationships, such as hostility and inconsiderateness (Tangney, 1991, 1992; Zahn-Waxler, Radke-Yarrow, & King, 1983).

Antisocial effects: Avoidance

If prosocial effects of guilt represent important support for an interpersonal approach to guilt, antisocial effects represent a potentially significant challenge. The interpersonal view is based on the belief that guilt is generally beneficial for relationships, and, indeed, some empirical findings support that general conclusion (see Tangney, 1991, 1992; Tangney, Wagner, Fletcher, & Gramzow, in press; Zahn-Waxler et al., 1983). Hoffman (1982, p. 304) suggested that guilt is one of the two main prosocial motives. It is therefore important to look carefully for any evidence that guilt leads to antisocial responses.

The main antisocial consequence that we could find is a tendency for transgressors to want to avoid their victims. Freedman et al. (1967) showed that transgressors preferred to help in ways that enabled them to avoid direct contact with their victims. These authors suggested that guilty people suffer an inner conflict; that is, they want to help someone to expiate the guilt, but they also are reluctant to face their victims. Facial contact with one’s victim is presumably aversive because the victim is a guilt cue and so would intensify the emotional distress, including embarrassment; also, it may be awkward to interact with one’s victim if one is not sure what to say and does not know whether the victim might express anger or some other response that would increase one’s distress.

These findings are also consistent with recent evidence that the recipients of unwanted love, who often feel guilty even though they describe themselves as morally innocent, feel extremely uncomfortable facing their ardent admirers and often try to handle the situation by avoiding the would-be lover (Baumeister & Wotman, 1992). Apparently, their guilt makes them reluctant to interact with the person whom they, supposedly in all innocence, have caused to suffer.

Facing one’s victim may also entail acknowledging one’s guilt and hence accepting one’s obligation to make restitution or amends; insofar as people are reluctant to accept such unspecified debts, they may find it prudent to avoid their victims. Several findings support the hypothesis that some people are reluctant to acknowledge their guilt to others (although, as already noted, some are moved to confess). Notarius, Wemple, Ingraham, Burns, and Kollar (1982) linked high arousal after a guilt inducing procedure with lack of facial expressiveness. This study was perhaps particularly important because the accuser was overtly irate and unpleasant, which, as we have stated, should aggravate the guilty person’s reluctance to face the accuser.

In general, these antisocial responses could arise from the fact that the victim is a guilt cue and is avoided to prevent unpleasant feelings from arising. An additional factor is that of social awkwardness; it is difficult to interact smoothly or pleasantly with one’s victims (see Baumeister & Wotman, 1992). It is also plausible that the response of social withdrawal may derive from shame rather than guilt, because these investigations have not assiduously maintained the distinction and indeed people often report mixtures of the two affects (Lewis, 1971; Tangney, 1989, 1991). The reluctance to face one’s victim may therefore not be a proper result of guilt, although it could arise when guilt is mixed with shame. In any case, it is important for research to separate guilt and shame carefully and then ascertain whether guilt alone makes people reluctant to face those whom they have harmed.

It also deserves mention that avoiding the victim is not a universal result of guilt. In fact, as we have noted, many consequences of guilt (such as confessing and attempting to make reparations) involve approaching the victim. Possibly a more general formulation would be that guilt tends to create an approach–avoidance conflict with respect to the victim. Ferguson et al. (1991) concluded that guilt produces just such a conflict.

A last point is that diminished contact between transgressor and victim may not necessarily be bad for the relationship. If the social avoidance is limited in time, it could prevent the victim from retaliatory, aggressive, or other responses that could make it more difficult to repair the relationship. A temporary avoidance caused by guilt could therefore even be beneficial for the relationship in the long run.

Conclusion

There is little to support the view that guilt causes or is frequently accompanied by a desire to suffer or a desire to be punished. The fact that guilt is associated with learning lessons and changing subsequent behavior is consistent with the view that guilt creates effects that are beneficial to human relationships. Other positive results of guilt are also interpersonal, such as motivating confession, apology, and reparative attempts, as well as the fact that the transgressor’s visible guilt feelings seem to help mollify victims and observers. Meanwhile, antisocial consequences of guilt seem mainly to involve a preference for avoiding contact with one’s victim. This tendency may simply reflect the fact that the victim serves to revive the unpleasant feelings of guilt.

Getting Rid of Guilt

Insofar as guilt is an acutely unpleasant state, it seems likely that people may want to escape from it. People apparently use a variety of strategies to reduce their guilt feelings.

One important strategy for dealing with guilt is to reduce fellow feeling with one’s victims (e.g., by dehumanizing them). In essence, one denies that any communal relationship exists between victim and transgressor. To regard a sufferer as an outgroup member with whom one has no social ties removes any danger that one’s transgression will break social bonds and minimizes the basis for empathic distress (because people mainly empathize with similar others). Evidence in support of this hypothesis can be gleaned by examining psychological processes associated with large-scale atrocities (e.g., Becker, 1986; Conquest, 1986; Lifton, 1986). To prepare oneself for harming others, it is apparently helpful to come to regard others as very different from oneself and having little in common. The Nazis’ attempt to cast Jews as vermin or germs, the Americans’ depictions of Amerindians as heathen savages and Vietnamese as gooks, slaveholders’ discussions of slaves as subhuman creatures, Communist cadres’ treatment of their former neighbors as alien members of a villain class, and other examples appear to follow this strategy. Meanwhile, it is noteworthy that protest movements against these atrocious persecutions often stressed that the victims were people too and sought to make commonalities salient. These observations are thus consistent with the view that guilt is linked to communal relationships and their inherent norms of mutual concern for each other’s welfare. Guilt apparently rises and falls with the perception of a communal relationship between transgressor and victim.

Derogating one’s victim may be a guilt-reduction strategy related to the denial of fellow feeling. In our own research, some subjects were asked to furnish accounts of perceived transgressions about which they did or did not feel guilty. One significant difference pertained to the relationship to the offended party (i.e., victim). Holding this person in high esteem was linked to higher feelings of guilt. When the victim was someone whom the subject held in low regard, guilt was minimal or absent (Baumeister et al., in press).

It must be kept in mind that these results are correlational. There is no way of assessing whether disrespect eliminated guilt or lack of guilt led to disrespect, although the former seems far more plausible. Our results do suggest, however, that derogating a victim would be an effective way of minimizing guilt. Previous work has shown a tendency for people to derogate their victims (Lerner & Matthews, 1967), although it has not indicated that such derogation actually reduces guilt. It is also noteworthy that Noel (1973) failed to replicate the transgression–compliance pattern when the transgression involved derogating another person; possibly, the derogation of the victim removed the guilt that often mediates subsequent altruistic behavior.

An experimental study conducted by Katz et al. (1973) is particularly relevant because subjects derogated victims from a different race more than victims from their own race, which seems to suggest that severing the tie of fellow feeling is easier when one’s victim is from another race. This suggestion seems to fit several observational studies that have proposed that perpetrators of crimes and atrocities tend to derogate their victims to remove any sense of fellow feeling, even sometimes regarding their victims as subhuman, especially when the victims belong to an ethnic or social group that can be clearly separated from one’s own (Conquest, 1986; Lifton, 1986).

An interpersonal approach suggests several reasons for derogating the victim. Such derogation may help make one’s relationship to the victim trivial, expendable, or undesirable. By thus severing the social bond, one removes an important basis for guilt. In addition, derogating the victim reduces his or her deservingness, and so the victim’s negative outcomes begin to seem more equitable and appropriate. For example, in the Soviet Great Terror, those accused of being enemies of the people were deprived of their rights and even of trial. The denial of the humanity of these unjustly accused individuals escalated over time. When one citizen protested the treatment of scientists sent to a Siberian labor camp by saying “These people might die!” the camp administrator replied, “What people? These are enemies of the people” (Conquest, 1990, p. 320).

Other guilt-reduction strategies appear to focus on the subjective interpretation of the event rather than on the interpersonal relationship. Thus, one means of avoiding guilt feelings is apparently to deconstruct the incident (i.e., to shift awareness to low levels of action identification and minimal meaningfulness); in such a mental state, one does not consider the meanings or implications of one’s actions, and so one escapes guilt (see Baumeister, 1991; Vallacher & Wegner, 1985, 1987). Wegner and Vallacher (1986) suggested that criminals focus attention on details and procedures rather than on implications and, in this way, avoid being disturbed by moral concerns and presumably avoid guilt. A similar suggestion emerged from Lifton’s (1986) observations about the Nazi concentration camp guards and staff, who preferred to dwell on checking lists and developing technical procedures instead of reflecting on the broader moral implications such as mass murder.

A related form of deconstruction can be called temporal bracketing, that is, treating a prior transgression as an isolated incident unrelated to ongoing issues in one’s life or to one’s identity. Baumeister et al. (1990) showed that interpersonal transgressors described such incidents in isolated, temporally bracketed terms, even explicitly insisting that the incident had had no relation to subsequent events; conversely, victims described transgressions as having lasting implications and continued relevance.

Another common means of evading guilt is to minimize the consequences of one’s actions (e.g., Baumeister et al., 1990). The severity of one’s misdeed is diminished by portraying its consequences as minor. Likewise, people make excuses and external attributions for their misdeeds, such as pointing to external causes and mitigating circumstances (e.g., Baumeister et al., 1990; Fischer, Schoeneman, & Rubanowitz, 1987; Kitson & Sussman, 1982; Snyder, Higgins, & Stucky, 1983). These strategies may or may not have interpersonal aspects. In particular, the denial of intentionality appears to be central to many strategies for minimizing guilt (e.g., Baumeister & Wotman, 1992) and indicates that people’s beliefs about guilt assign a prominent place to intentionality. Although people clearly do feel guilty over unintentional transgressions (McGraw, 1987), they also respond as if guilt can be reduced by proving a lack of intent.

Thus, people use a variety of strategies to get rid of guilt. Some of these strategies involve intrapsychic responses, including avoiding meaningful thought, minimizing the offense or their responsibility for it, and bracketing it in the past. Others are interpersonal, particularly the pattern of derogating (even dehumanizing) one’s victim.

Discussion

The empirical literature on guilt has repeatedly found evidence of interpersonal patterns and aspects. Although feelings of guilt may occur in the privacy of an individual psyche, they appear to be pervasively and multiply related to interpersonal motivations and attachments, particularly involving communal bonds between significant others. The social dimension of guilt is not limited to socialization and the teaching of moral standards; rather, the causes, forms, and consequences of guilt appear to derive from a fundamental concern with communal relatedness. We begin by reviewing the main points.

Autobiographical accounts of guilt episodes refer almost exclusively to interpersonal contexts, inequities, and transgressions. Interpersonal harm, rather than the intent to harm, appears to be crucial for determining guilt. Neither self-perceived innocence nor lack of harmful intent is proof against guilt feelings, particularly when a relationship partner seeks to induce them.

Research on transgression and compliance showed multiple interpersonal patterns. After doing something bad to another person (in most studies, this was an accident), people are motivated to help that person or comply with that person’s wishes, apparently to rectify any inequity and to repair any damage to the relationship. Various interpersonal factors, including sense of interpersonal obligation, an empathic response to the victim, and other people’s awareness of the transgression, all contribute substantially to the link between transgression and compliance. Additional evidence suggests that transgression creates a motivation to affirm social bonds and attachments, possibly along with derogating out-groups or rival groups.

Guilt can even occur without transgression; these patterns likewise reflect interpersonal equity and relationships. People feel guilty when they suffer less or benefit more than others do. People also feel guilty when they hurt someone unintentionally or when another person is seen as ultimately responsible for the hurt that they inflict on him or her (as in the case of unrequited love). These patterns of guilt appear to be strongest and most common in the context of (actual or possible) communal relationships.

Personality and developmental data reveal positive links between guilt and empathy, which is a form of interpersonal sensitivity, and between guilt and other interpersonal traits. Guilt also varies with the relationship context and often depends on a convergence among the transgressor’s own standards, the victim’s standards, and the victim’s evaluation of the transgressor’s actions. Guilt depends on the interpersonal consequences of one’s actions; identical actions and identical intentions can lead to different levels of guilt, depending on how other people are affected. Guilt is strongest and most common when one transgresses against valued relationship partners. The interpersonal causation of guilt is further indicated by the fact that people can and do make other people feel guilty and by the fact that these guilt manipulations seem mainly to occur in the context of close, communal, ongoing relationships.

Not only the causes but the consequences of guilt are interpersonal. Guilt motivates people to apologize, to attempt to make amends, to try to repair damage to relationships, to confess and seek forgiveness, and to change their behavior so as to be more pleasing and satisfactory to relationship partners. On the other hand, when no communal relationship exists, guilt motivates people to distance themselves from victims, (e.g., by reducing contact with them, derogating them, and even dehumanizing them).

Assessment of Interpersonal Analysis

An interpersonal approach to understanding guilt is consistent with a large number of findings and offers a useful basis for interpreting many empirical results. Interpersonal factors, including relationship context and roles, empathy, public surveillance, and victim reaction, appear to be influential in generating guilt. Furthermore, guilt appears to motivate several interpersonal patterns of behavior, including reparation and confession.

The fact that people sometimes feel guilty vis-à-vis strangers raises the largest objection to the view of guilt as rooted in close relationships. It may be that people have most guilt feelings in the context of ongoing, close relationships but that their response patterns sometimes generalize to interactions with strangers, particularly strangers with whom a minimal in-group bond exists. Indeed, one may consider it to be a major function of socialization to take the basic capacity for empathic guilt and extend it beyond the close relationships in which it originates. Still, the problem of guilt vis-à-vis strangers was an important reason to shift the theoretical emphasis from close to communal relationships as the primary source of guilt. People will adopt a communal stance toward a stranger with whom they have something in common and with whom a communal relationship is regarded as possible (e.g., Clark et al., 1987). Respect and concern for the welfare of others, regardless of self-interest, is the essence of communal relatedness, and in our view such concern is a vital factor in making guilt possible.

We have proposed several interpersonal functions of guilt. First, guilt motivates people to adopt relationship-enhancing behavior patterns. Ample evidence was consistent with this view. Guilty people affirm their social bonds, pay increased attention to their partners, apologize and make amends for transgressions, repair relationship damage, and change their behavior to suit the partner. Indeed, there is some evidence that guilt itself can help strengthen a relationship, because it conveys a positive message of concern about the partner and about the relationship that may overcome any implication (of the transgression) that one does not care. And people apparently induce guilt in others to increase their attachment.

The second function of guilt is as a means of influencing others; in particular, it can increase equity in a relationship by allowing the less powerful person to get his or her way. This was also well supported. People do respond to guilt as a form of influence, in particular changing their behavior to suit the other person’s wishes. Conversely, people induce guilt in others to get their way. People feel guilty over positive inequities within relationships, and they will avoid or rectify such inequities to avoid guilt. To the extent that guilt punishes and prevents people from enjoying positive inequities, it will serve as a valuable aid to less powerful relationship partners to help them receive fair and equitable treatment. Indeed, evidence suggests that positive inequities mainly produce guilt in an interpersonal context: Simply being overpaid does not appear to make people feel guilty, but being overrewarded in comparison with other people does induce guilt. All of these patterns appear most strongly and consistently within close relationships. The link between guilt and empathy may be partly based on the fact that guilt occurs primarily within communal relationships, in which empathic ties are likely to be maximal.

Indeed, communal relationships may need an influence mechanism like guilt. In exchange relationships, one gets one’s way by making a deal, so to speak: One offers rewards or withholds punishments in exchange for the benefits one wants. In communal relationships, on the other hand, there is no such implicit structure of social exchange, and, in principle, each person depends on the other’s concern to motivate the other person to do what one wants. If that fails, even because one’s partner simply does not realize what one wants, one needs some way to remotivate the partner, and guilt may serve this function effectively. Guilt invokes the broad assumption of pervasive mutual concern to motivate the partner to do the particular thing one wants. Also, it appears that guilt induction is especially favored by nonassertive people who are presumably unwilling or unable to get their way with more direct or confrontational means.

The third function of guilt is to redistribute emotional distress. Undoubtedly, guilt makes the transgressor feel worse, and a small amount of evidence showed that victims feel better when transgressors feel or express guilt. Although evidence is consistent with this third function, it is not extensive, and more research is needed.

We also have proposed two bases for guilt feelings. Regarding the first, the evidence appears to be consistent with Hoffman’s (1982) contention that guilt begins with empathic distress; children feel bad when they see another person hurt, and their distress may indeed be the developmental precursor of guilt. Moreover, empathy appears to be linked to guilt even into adulthood. The second proposed basis for guilt was anxiety over the breaking of social bonds, such as separation and exclusion anxiety (Baumeister & Tice, 1990; Bowlby, 1969, 1973). Doing things that may jeopardize social attachments may bring distress, and this distress too may take the form of guilt. The fact that many patterns of guilt are experienced mainly in the context of close, communal relationships fits the view that guilt is substantially based on anxiety over possible loss of attachment, and the empirical links to anxiety, loneliness, and other forms of interpersonal distress also fit this view of guilt.

In sum, it seems reasonable and even necessary to invoke an interpersonal perspective for understanding guilt. Guilt seems well designed to regulate behavior in ways that will strengthen and preserve social relationships. It is found most strongly and consistently in communal relationship contexts, it depends heavily on interpersonal interaction patterns and relationships, and it serves useful, relationship-enhancing functions.

Reconciling Interpersonal and Intrapsychic Perspectives

Although interpersonal and intrapsychic processes were long considered as rival forms of explanation, recent years have witnessed increasing efforts to integrate them (e.g., Tetlock & Manstead, 1985). Such integration seems necessary for an adequate theory of guilt, because both dimensions appear essential to a full understanding of the concept.

Thus, our disagreement is not with the assertion of intrapsychic aspects and processes in guilt but with the reductionistic assertion that guilt consists solely of such aspects and processes. Guilt is not evoked only from within the self, contrary to Lewis’s (1971, p. 85) claim. Our review has repeatedly indicated reasons for considering a purely intrapsychic account of guilt to be incomplete. These reasons can be briefly recapitulated as follows. First, it is readily apparent that people induce and manipulate guilt feelings in others. Guilt is not invariably a private, inner response generated wholly within the self through evaluation of one’s own actions. Rather, interpersonal communication can be an effective and powerful means of generating guilt. Second, proneness to guilt is empirically correlated with empathy. Empathy is, by definition, a form of interpersonal sensitivity (e.g., Eisenberg & Miller, 1987; Hoffman, 1981, 1982), and if guilt were wholly a matter of intrapsychic self-evaluation, there would be little reason for a link to empathy. Third, the causes of guilt appear to be heavily interpersonal. Fourth, guilt often motivates people to make reparations or amends or at least to apologize; in other words, the motivational consequences of guilt are interpersonal. Fifth, the consequences of guilt do not seem to encompass a wish for punishment, contrary to what some intrapsychic theories (e.g., Piers & Singer, 1953/1971) have insisted. Sixth, the emphasis of some intrapsychic accounts of guilt on intention and voluntary choice appears to be fallacious. People feel guilty for accidental transgressions as well as for voluntary ones. The interpersonal consequence of the act, rather than the intrapsychic intention, must be emphasized in the genesis of guilt. Seventh, guilt appears to be significantly linked to close and communal relationship contexts. Finally, people sometimes feel guilty without doing anything wrong. Thus, perceiving oneself as responsible for a transgression is neither necessary nor sufficient to produce guilt feelings. An intrapsychic account of guilt as the outcome of a self-evaluation process is therefore incomplete.

On the other hand, adults do consistently act as if guilt depends heavily on intrapsychic factors such as self-appraisal, controllable decisions, and malicious intent. One line of potential integration is that guilt may originate in interpersonal processes such as empathy and harming a relationship partner but that people’s understanding of guilt is gradually assimilated to models based on intentional violation of abstract standards. People’s beliefs about guilt may, of course, differ substantially from the actual phenomena and causes of guilt. People’s efforts to respond to guilt and to escape from it may be substantially conditioned by these beliefs. Thus, we noted the paradox of people who feel guilty for rejecting another’s romantic overtures and who seek to justify their behavior by insisting that they never intended to lead the other person on or encourage unrequited love. The interpersonal harm causes the guilt, but people invoke intentionality and other features of intrapsychic, self-evaluation models to escape from guilt feelings.

In our view, guilt can best be conceptualized as rooted in social relationships rather than theological principles or abstract ethical conceptions. Guilt feelings originate in interpersonal attachments and social exchange. People appear to feel guilty when they hurt, neglect, or disappoint others and when they benefit unfairly vis-à-vis others or at others’ expense. Communal relationships, based on expectations of mutual concern for each other’s welfare, are particularly relevant to causing guilt. This view of guilt is consistent with recent attempts to reconceptualize emotions as social phenomena rather than as strictly intrapsychic occurrences (e.g., De Rivera, 1984; Frijda, 1986).

Meanwhile, responsibility, intention, external or mitigating circumstances, and other tenets of the self-evaluation view are often important determinants of the magnitude of guilt. Intentionality and responsibility are not prerequisites for guilt, but some evidence does suggest that they increase guilt (e.g., Schmitt et al., 1989). This suggests how the interpersonal and self-evaluation theories might be integrated. Deliberate, intentional transgressions are more likely to be repeated than accidental ones and are more plausibly interpreted as reflecting an indifference about the relationship partner (and about the relationship itself). Therefore, voluntary, intentional transgressions represent a greater threat to the relationship than do involuntary or accidental ones, and so a stronger guilt response would be warranted. Other possibilities for integrating the interpersonal and self-evaluation views were suggested by Millar and Tesser (1988), who showed that people feel guilty when their expectations for proper behavior match the expectations of another person and they violate those expectations within the relationship with that person.

Hoffman (1982) suggested that as children grow, they learn to make attributions that restrict guilt to occurrences in which they feel responsible for another’s suffering. This may be regarded as the first developmental step toward constructing the principled, intentional gloss that eventually comes to describe people’s beliefs about guilt. To Hoffman’s view may be added the development of communal relationships that bring an understanding of mutual obligation. These relationships furnish a basis for equity calculations, which may initially be crude and primitive but can eventually evolve into complex, sophisticated forms. In an exchange relationship, fairness can be negotiated explicitly; in communal relationships, however, norms of mutual concern are supposed to replace them. Yet equity and fairness cannot be irrelevant to communal relationships either. In an important sense, therefore, guilt may be a vital mechanism for maintaining fairness and equity in communal relationships. Guilt may arise whenever positive inequities arise within the relationship. Hence, one may feel guilty without doing anything wrong, although interpersonal transgressions probably still form the largest source of guilt.

The developmental research is consistent with the view that guilt has its roots in social relatedness, although not in superego identification with parents in the fashion that Freudian theory might suggest. Instead, it appears that empathic distress over others’ suffering, strong affectional bonds with parents, and concern over loss of love provide the foundations of guilt feelings. As people mature into adulthood, however, they do refine their notions of guilt to incorporate conceptions of intentionality, responsibility, and abstract moral standards. Guilt among adults may therefore consist of emotional responses linked to empathic distress and desire for social attachment but overlaid with conceptions derived from ethical standards and abstract conceptions of equity.

In sum, the feeling of guilt can be regarded as an intrapsychic outcome and mediator of interpersonal processes. The progress of individual socialization may increase the capacity of intrapsychic processes to substitute for interpersonal ones, but typically guilt continues to originate and be guided by interpersonal factors. In the course of development and socialization, people do acquire a body of beliefs about guilt, many of which invoke models of abstract principles and standards, as well as intrapsychic factors such as intention. This cognitive overlay will exert an influence over guilt processes among adults, in particular guiding how they respond to guilt and seek to defend themselves against guilt feelings (e.g., by arguing that they never intended harm). The original link to interpersonal processes remains strong and vital in adulthood, however.

Problems and Pathologies

Consistent with the main thrust of recent evidence, this article has emphasized the constructive aspects of guilt, but these aspects should not be overstated. The popular impression of guilt as a potentially destructive and dysfunctional emotion may have substantial validity in some cases. In the absence of compelling data on the destructive effects of guilt, we offer a speculative extension of our interpersonal discussion to encompass how such effects might occur.

Guilt may be considered a tool that is well designed to benefit relationships but can be destructive if misused. The simple fact that guilt is aversive suggests a first destructive possibility: Too much guilt may cause people to abandon a relationship simply to avoid unpleasant feeling states. In other words, whereas occasional and limited doses of guilt may induce people to do what is best for a relationship, persistent and large doses may make them abandon it. We noted that a tendency to avoid one’s victim may arise from the mere fact that the victim serves as a stimulus or cue to guilt feelings. If this pattern were to become deeply ingrained so that the person became more or less permanently unable to encounter the victim without feeling guilty, avoidance could become the rule.

We also suggested that one person’s guilt feelings may enhance a relationship by conveying concern and commitment to the relationship partner whom he or she has wronged, particularly if guilt feelings arise from emphathic concern and anxiety over losing the relationship. One can thus use the degree of the partner’s guilt as a test of the partner’s love and commitment. Unfortunately, this response pattern may invite abuse: Insecurely attached people may try to induce guilt feelings in their partners to gain reassurance whenever they begin to doubt the partner’s love and commitment. Once the partner begins to see these efforts as manipulative, he or she may respond with resentment and refusal to play along, which will send the opposite signal from that desired by the insecure member. The result will be an upsetting experience for both partners that if repeated, could begin to undermine the attachment.

More generally, we have noted that inducing guilt may often breed resentment. In this, to be sure, guilt does not differ from other, more direct forms of influence. It differs from other methods, however, in that it is based directly on the other person’s feelings of love and concern; in a sense, guilt elicits compliance by using the partner’s attachment feelings against him or her. In some cases, the partner may begin to think that the only way to resist the other person’s influence is to reduce those positive feelings. We reiterate that inducing guilt appears to be a potentially costly technique for getting one’s way, and overdoing it can be extremely destructive.

Yet another kind of problem could arise if two relationship partners are mismatched in their understanding of guilt. Undoubtedly, people differ in their proneness and vulnerability to guilt. If guilt is a mechanism for regulating relationships, then serious relational problems could arise if the partners differ widely in their guilt proneness. The more guilt-prone person could easily end up being exploited as a result of inequitable perceptions of transgressions and acceptable acts. The more guilt-prone person may also be more likely to perceive the relationship as unfair and to accumulate a global resentment toward the partner. Thus, for example, if women are indeed more prone to guilt than men (e.g., Tangney, 1990), then many marriages might readily drift toward arrangements in which wives perform more than their equitable share of joint duties and resent their husbands for this (see Hochschild, 1989). Meanwhile, having a partner with a stronger sense of guilt than oneself could also have its drawbacks, such as a recurrent demoralizing experience of feeling that one is falling short of the partner’s expectations and hence eliciting disapproval.

Ultimately, guilt may be most beneficial if people alter their actions so as to prevent it from ever actually occurring. Although we have emphasized the beneficial functions of guilt, they must be recognized as responses to problems, and so the existence of frequent guilt feelings is not a healthy sign; rather, it is a sign of trouble. On an occasional basis, guilt can help resolve such troubles, but large, frequent, or chronic episodes of guilt may become part of the problem rather than the solution. The ideal relationship partner would presumably be neither the person who is immune to guilt nor the person who is frequently consumed with guilt feelings. Rather, it would perhaps be the person who is so sensitive to guilt that he or she carefully and consistently acts in ways that will prevent guilt.

Directions for Further Research

Our comments throughout the review have made it clear that further research is needed to build on the solid foundations of empirical information about guilt and to facilitate the emergence and refinement of new theory. New research needs to avoid several of the problems that have plagued past work, including the distinction between intentional and unintentional transgressions and, in particular, the uncertainty as to whether laboratory subjects do, in fact, feel guilty. It is no longer adequate to infer the existence of guilt simply because, in the researcher’s way of thinking, some standard is violated.

The interpersonal context and causation of guilt must be recognized as a top priority for further empirical work. One useful direction would be to examine directly the hypothesized interpersonal functions of guilt. Does guilt indeed make people do relationship-enhancing things and avoid relationship-harming acts? Is guilt an effective form of interpersonal influence, and, if so, who uses it and what are the costs (e.g., resentment)? We have already noted that the third function of guilt, as a mechanism for redistributing emotional distress within the dyad, has the smallest quantity of relevant empirical evidence, and so further study of this function is particularly needed. Evidence for this would be that victims feel better when they learn that transgressors feel guilty.

Furthermore, if guilt is indeed caused by making a relationship partner feel bad, and guilt is a form of feeling bad, then it seems likely that people will sometimes feel guilty about inducing guilt in relationship partners. This guilt over inducing guilt (“metaguilt”) could be a significant added cost of using guilt-induction strategies to influence other people.

A particular focus of the interpersonal view was that guilt should be especially marked and influential in close relationships. Calculations of equity may turn out to be very precise and highly relevant to guilt in close, long-term relationships. The past decade has witnessed a rapid growth in the empirical study of such close relationships, and researchers in this area may be well placed to pursue research on guilt. One issue deserving investigation is the long-term accounting of guilt within the context of such relationships. Couples may keep track of each other’s transgressions and reparations, and they may believe that one person owes the other nonspecific obligations as a result of transgressions. Negotiating guilt may become an ongoing process, requiring partners to point out all their sufferings and disappointments at the other’s hands and to remind the other person of past transgressions and inequities so as to sustain the sense of obligation.

One last and extremely important direction for guilt research is to explore the boundaries of guilt, that is, the limiting conditions that decide whether a given action will or will not generate guilt. This direction especially needs to move beyond the notion of transgression to examine other sources and moderators of guilt. Survivor guilt needs further study, and the notion that victims feel guilty is, at present, limited mostly to clinical observations. Systematic study of such phenomena could contribute significantly to a clear understanding of what events produce guilt.

Conclusion

Guilt is an important and pervasive phenomenon, and some people report that it is an almost constant factor in their decisions, feelings, and actions. It seems highly desirable to make a place for guilt in the psychology of motivation. Meanwhile, research on emotion, affect regulation, close relationships, and similar issues can likewise profit from studying guilt.

Our review has suggested a fundamentally social view of the essential nature of guilt. The sources of guilt should be conceptualized in terms of communal relations between significant others. The emotional roots of guilt apear to involve interpersonal relatedness, including empathy, attachment, and exclusion anxiety. Guilt appears to serve multiple relationship-enhancing functions, including motivating people to treat their partners well and avoid transgressions, rectifying inequities and allowing less powerful partners to exert influence, and possibly redistributing emotional distress so as to bring the relationship partners into harmony. Guilt appears to arise out of interpersonal transactions and to vary substantially with the interpersonal context.

Communal relationships are based on assumptions of mutual concern and nurturance, and guilt may be a major affective mechanism for ensuring that people adhere to those patterns. If so, then guilt, despite its unsavory reputation and aversiveness, may be valuable in helping people live together and maintain successful interpersonal relationships.