Self - Regulatory Perspective on Shame and Guilt 1

A self-regulatory framework for distinguishing between shame and guilt was tested in three studies. Recently Janoff-Bulman, Sheikh, and Hepp (2009) proposed two forms of moral regulation based on approach versus avoidance motivation. Proscriptive regulation is sensitive to negative outcomes, inhibition-based, and focused on what we should not do. Prescriptive regulation is sensitive to positive outcomes, activation-based, and focused on what we should do. In the current research, consistent support was found for shame’s proscriptive and guilt’s prescriptive moral underpinnings. Study 1 found a positive association between avoidance orientation and shame-proneness and between approach orientation and guilt-proneness. In Study 2, priming a proscriptive orientation increased shame while priming a prescriptive orientation increased guilt. In Study 3 transgressions most apt to represent proscriptive and prescriptive violations predicted subsequent judgments of shame and guilt, respectively. This self-regulatory perspective provides a broad interpretive framework for understanding and extending past research findings. keywords: shame, guilt, moral emotions, morality, self-regulation Self-Regulatory Perspective on Shame and Guilt 3 The “Shoulds” and “Should Nots” of Moral Emotions: A Self-Regulatory Perspective on Shame and Guilt Shame and guilt are emotions central to people’s moral experiences. Whether anticipated or actually felt, they steer our course towards morality and away from immorality, guiding our self-evaluations and our conduct. In this sense, shame and guilt are emotions central to moral regulation in particular. Given that shame and guilt are both negative responses to transgressions, it is no surprise that colloquially the two terms are often used interchangeably and the two emotions are often experienced simultaneously. Past research has focused largely on descriptive analyses and has generated a compelling case for differences between shame and guilt in two primary domains: appraisals and action tendencies. We acknowledge the importance of these distinctions, but aim to provide a new perspective on shame and guilt by placing them within a broader theoretical context, one informed by work on selfregulation, or more specifically moral regulation. Our goal then is not to minimize the contributions of the important work to date, but rather to build on this base to further expand our understanding of these two moral emotions. Past Research: Appraisals and Action Tendencies “The currently most dominant basis for distinguishing between shame and guilt” is the “focus on self versus behavior” (Tangney, Stuewig, & Mashek, 2007, p. 349). This distinction has been most noticeably advanced by Tangney (e.g., Tangney, 1991; Tangney and Dearing, 2002; Tangney et al., 2007) based on her extensive research program on these two moral emotions. Here a negative evaluation of the global self is implicated in shame, whereas a negative evaluation of a specific behavior is implicated in Self-Regulatory Perspective on Shame and Guilt 4 guilt (also see Janoff-Bulman, 1978, on the attributional differences between characterological self-blame [global self] and behavioral self-blame [specific behavior]). From an appraisal perspective, when we experience guilt, we are focused on what we did—the “bad” act; in contrast, shame’s focus is on who we are—the “bad” self. And although both involve internal attributions, the former attributions are controllable and malleable whereas the latter are uncontrollable and much more persistent (Tracy and Robins, 2006). Similarly, within an intergroup context, group-based shame, but not guilt, is associated with threatened shared group identity (e.g., Lickel, Schmader, Curtis, Scarnier, & Ames, 2005; Lickel, Schmader, & Spanovic, 2006). Guilt and shame have also consistently been found to engage distinct action tendencies to amend or hide, respectively. More specifically, researchers have observed that guilt motivates reparative actions, including apology, confession and prosocial actions, whereas shame motivates denial, withdrawal, and escape from the shameinducing event (e.g., de Hooge, Zeelenberg, & Breugelmans, 2007; Hoffman, 1982; Holtzworth-Monroe, 1989; Lewis, 1971; Lindsay-Hartz, 1984; McGraw, 1987; Tangney, 1993; Tangney et al., 1996; Wicker et al., 1983; Tangney, 1991; Tangney, Miller, Flicker & Barlow, 1996; Tangney, Wagner, Fletcher, and Gramzow, 1992; Niedenthal, Tangney, and Gavanski, 1994; for reviews see Tangney and Dearing, 2002; Tangney, Stuewig, & Mashek, 2007; cf. de Hooge, Breugelmans, & Zeelenberg, 2008). Even among gradeschool children, guilt is associated with approaching others who were hurt and making amends, whereas shame is associated with attempts to escape and avoid looking at others (Ferguson, Stegge, and Damhuis, 1991). These differences are evident as well in people’s eagerness to talk about their experiences of guilt, manifesting in turn an urge “to Self-Regulatory Perspective on Shame and Guilt 5 confess and atone...—the desire to set things right” in contrast to their hesitance to talk about shame, reflecting an “urgency to hide these shameful experiences from others, as well as from themselves...” (Lindsay-Hartz, 1984, p. 691-693). The self versus behavior attributions for shame and guilt are most commonly used to account for the differences in motivational tendencies associated with the two emotions. Tangney and Dearing (2002, p.24) note “the fundamental difference between shame and guilt centers on the role of the self,” which has “far-reaching implications...for subsequent motivation.” Thus, guilt’s focus on behavior allows a person to more easily remedy and correct for the bad action, whereas in the case of shame no apology or reparation would remedy the inadequacies of the bad self. Rather, the subjective distress associated with negative evaluations of the global self motivate the individual to minimize these negative feelings through efforts to hide or withdraw. A Broader Framework: Self-Regulation Self-regulation refers to purposive (but often not self-controlled) processes that guide action motivated by end-states (e.g., Carver, 2008). More specifically, the selfregulatory literature focuses on distinct, orthogonal self-regulatory systems, one focused on approaching a positive or rewarding end-state and the other on avoiding a negative or punishing end-state. Thus Carver and Scheier’s (1990, 1998, 2008; see also, Carver, 2006) model of self-regulation conceptualizes behavior as movement to or from one of two internal referents; the approach system, regulated by a discrepancy-reducing feedback loop, involves movement towards a goal or desired referent. The avoidance system, regulated by a discrepancy-enlarging feedback loop, involves withdrawal from an “anti-goal” or undesired referent. Self-Regulatory Perspective on Shame and Guilt 6 Carver and Scheier (e.g., 2008) build on Gray’s (1982, 1994) distinction between the Behavioral Activation System (BAS) and the Behavioral Inhibition System (BIS). The BAS, the appetitive motivational system, is sensitive to reward, while the BIS, the aversive motivational system, is sensitive to punishment. Importantly, the BAS is responsible for the activation of movement towards positive goals while the BIS causes the inhibition of movement towards goals that may lead to negative or painful outcomes (Carver & Scheier, 2008; Carver and White, 1994; also see Gray, 1982, 1994). Research has found that the BAS and BIS are associated with different neural substrates (e.g., Davidson, 1995, 1998; Davidson, Ekman, Saron, Senulis, & Friesen, 1990; Sutton & Davidson, 1997). The approach system, then, involves positive end-states and an activation action tendency, whereas the avoidance system involves negative end-states and an inhibition action tendency. The distinction between these two regulatory systems has been used to promote our understanding of phenomena as diverse as attention (e.g., Forster, Friedman, Ozelsel, & Denzler, 2006), achievement (e.g., Elliot & Church, 1997), interpersonal relationships (e.g., Grable & Strachman, 2008), and power (e.g., Keltner, Gruenfeld, & Anderson, 2003). Most recently, Janoff-Bulman, Sheikh, and Hepp (2009) have applied this general dual regulatory system to moral regulation. The Two Systems of Moral Regulation Janoff-Bulman et al. (2009) have distinguished between two systems that regulate distinct moral conduct. Proscriptive morality restrains immoral conduct; it is based in an avoidance system, in that it is sensitive to negative end-states (e.g., anti-goals, punishments, threats) and motivates behavioral inhibition. Prescriptive morality Self-Regulatory Perspective on Shame and Guilt 7 promotes moral conduct; it is based in an approach system, sensitive to positive endstates (e.g., goals, rewards, incentives) and motivates behavioral activation. In guiding moral conduct, proscriptive regulation focuses on what we should not do, whereas prescriptive regulation focuses on what we should do. More specifically, the former engages negative internal referents (e.g., “I should not harm others”) and the latter engages positive internal referents (e.g., “I should help others”). The proscriptive system regulates morality by inhibiting negative (immoral) behaviors—by curbing negative desires and temptations to engage in wrong conduct. In contrast, the prescriptive system regulates the activation of positive (moral) behaviors—by catalyzing the positive desire to engage in right conduct. Kochanska and colleagues’ (Aksan & Kochanska, 2005; Kochanska, 2002; Kochanska, Coy, & Murray, 2001) longitudinal studies on the early development of morality in children provides support for differentiating between these two forms of moral self-regulation. The researchers distinguished between “do’s,” behaviors involving activating and sustaining an activity (e.g., toy cleanups) and “don’ts,” involving prohibitions and suppressing behaviors (e.g., not playing with a forbidden attractive toy). Kochanska et al.’s (2001) research demonstrates that “do’s” are more challenging than “don’ts” for children at all ages studied (i.e., 14, 22, 33, and 45 months), and fearfulness is associated with “don’ts,” but not with the “do’s.” They conclude that their data provide “impressive evidence of substantial differences” between do’s and don’ts in early self-regulation; and it appears that these differences reflect distinct appetitive and aversive motivational underpinnings (e.g., fear has been proposed as part of the BIS; Gray, 1972; Carver and White, 1994; Sutton & Davidson, 1997). Self-Regulatory Perspective on Shame and Guilt 8 The two modes of moral regulation--proscriptive and prescriptive--are not simply mirror images, but operate differently and have distinct features. For instance, psychologists have shown that the motivation to avoid a negative entity is stronger than that to approach a positive entity, and failure to avoid incurs greater psychological distress than the failure to approach (see Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001; Rozin & Royzman, 2001). Similarly, Janoff-Bulman et al. (2009) provide support for the greater potency and dominance of the proscriptive system over the prescriptive system; they find moral beliefs and judgments related to the proscriptive system are generally more condemnatory, strict, and mandatory than those related to the prescriptive system, and proscriptive immorality engenders greater blame than prescriptive immorality. Moral Regulation and Its Relevance to Shame and Guilt How might this self-regulatory distinction between prescriptive and proscriptive morality inform guilt and shame? We propose that guilt, more than shame, is associated with prescriptive moral regulation, whereas shame, more than guilt, is associated with proscriptive moral regulation. We experience guilt when we interpret our transgression via the prescriptive system—by focusing on positive end-states and moral “shoulds.” Shame, in contrast, is felt when we interpret our transgression via the proscriptive system—by focusing on negative end-states and moral “should nots.” In the case of guilt, our experience reflects a primary focus on the recognition that we didn’t act morally by doing the right thing; that is, we failed to approach a positive referent, for we did not act like a caring friend, honest person, or loving partner. In the case of shame, our predominant response reflects the recognition that we acted immorally by doing the Self-Regulatory Perspective on Shame and Guilt 9 wrong thing. Shame acknowledges our failure to avoid a negative referent, for we acted like a cheat, a thief, or a liar (also see Tracy & Robins, 2004, 2007 on the importance of self-representations in shame and guilt). Guilt highlights positive (rewarding) moral referents, what we should do, and pushes us forward, towards the possibility of redemption and further future moral outcomes; shame highlights negative (punishing) moral referents, what we should not do, and leaves us confronting our own immorality. In both instances we recognize our transgression, but in the case of guilt we are able to focus on the right, good, or moral. The approach-activation action tendency associated with prescriptive morality is evident, for this focus motivates us to move towards the desired goal by “righting the wrong.” In contrast, the focus for shame is on the wrong, bad, or immoral. Here, the avoidance-inhibition tendency associated with proscriptive morality is evident when we feel shame, for now this focus motivates us to escape the negative implications of our misdeeds, or what Carver (2006) calls the “antigoal.” Thus, guilt is associated with motivations to amend, whereas shame is associated with efforts to hide. A self-regulatory approach to shame and guilt reactions is consistent with differences in the behavior versus self focus of guilt and shame noted in previous research. Guilt’s emphasis on rewarding, moral outcomes limits the negative evaluation associated with transgressing to the behavioral act in order to foster the activation of alternative, positive behaviors. Moreover, following the vast literature on the negativity bias and an asymmetry in morality discussed above, proscriptive regulation is far harsher and more condemnatory than prescriptive regulation. When the proscriptive system is the basis for moral evaluation, transgressions reflect punishment-worthy moral failure. Self-Regulatory Perspective on Shame and Guilt 10 We perceive what we did wrong in terms of punishing, demanding referents, and are thus likely to experience harsher, more negative evaluations that generalize to the global self. Not surprisingly, then, shame is considerably more painful than guilt (e.g., Lewis, 1971; Lindsay-Hartz, 1984; Morrison, 1996; Tangney et al. 1992). To date shame and guilt have not been considered in terms of broader moral regulation or been distinguished in accordance with approach-avoidance regulatory systems more generally. However, some past research suggests the relevance of this perspective for shame and guilt. A series of developmental studies by Zahn-Waxler and Kochanska (Zahn-Waxler and Kochanska, 1990; Zahn-Waxler, Kochanska, Krupnick, & McKnew, 1990), for instance, found that guilt in particular arises from recurrent positive and rewarding interactions with socializing agents. The authors observed that parental warmth and affection, but not punishment, are positively associated with the development of guilt. Tangney and Dearing (2002) also have noted that adolescents of punitive parents, who use putdowns and are emotionally abusive, exhibit high proneness to shame (e.g., Gilbert, Allan, & Goss, 1996; Hoglund & Nicholas, 1995). Further, recent research suggests guilt is related to the neural correlates of the BAS: Amodio, Devine, and Harmon-Jones (2007) found that guilt-eliciting bogus feedback regarding displays of prejudice produced increases in subjects’ left-sided frontal cortical asymmetry, supporting guilt’s association with an appetitive system. Current Studies The research studies that follow are a first attempt to test the proposed regulatory underpinnings of shame versus guilt, and in particular the proposed positive association between shame and proscriptive morality, and between guilt and prescriptive morality. Self-Regulatory Perspective on Shame and Guilt 11 Three studies employed distinct methods for assessing the two regulatory systems and the subsequent reports of shame versus guilt. Study 1 explored the relationship between dispositional approach versus avoidance orientation and the two self-evaluative emotions. We expected that for the same transgressions, those higher in avoidance orientation would report greater shame, whereas those higher in approach orientation would report greater guilt. In Study 2, we situationally manipulated the prescriptive and proscriptive moral systems and examined the impact on experienced shame and guilt. We expected that priming the proscriptive system would result in greater shame, whereas priming the prescriptive system would result in greater guilt. And in Study 3, we employed transgressions most apt to represent prescriptive and proscriptive violations in order to investigate subsequent judgments of shame and guilt; in this study we also addressed the question of whether some types of eliciting events are more likely to be associated with shame versus guilt. Study 1: Shame, Guilt, and Individual Differences in Self-Regulation As a first study, we investigated whether individual differences in general approach-avoidance orientation serve as unique predictors of guilt-proneness and shameproneness. Carver and White’s (1994) Behavioral Activation System (BAS) and Behavioral Inhibition System (BIS) Scales are ideal to investigate the respective general appetitive and aversive-based regulatory underpinnings of guilt and shame. The BIS measures dispositional differences in behavioral inhibition, or avoidance motivation, and the BAS measures dispositional differences in behavioral activation, or approach motivation. Indeed, as Carver (2006, p.107) notes, these scales are best used to “investigate whether a given phenomenon pertains to approach or avoidance.” Self-Regulatory Perspective on Shame and Guilt 12 Janoff-Bulman et al. (2009) have previously demonstrated a unique and positive association between BIS scores and proscriptive moral judgments, and between BAS scores and prescriptive moral judgments, supporting the self-regulatory differences between these types of morality. We therefore expected the BAS to positively predict responses indicating guilt-proneness and the BIS to positively predict responses indicating shame-proneness, thereby providing support for differential regulatory sensitivities underlying shame and guilt. Method Participants. A total of 120 undergraduate students (99 women and 21 men) participated in the study for extra course credit. Materials and Procedure. Participants first completed Carver and White’s (1994) 7-item BIS and 13-item BAS scales. A sample BAS item is “When I want something, I usually go all-out to get it.” A sample BIS item is “Criticism or scolding hurts me quite a bit.” Participants indicated their extent of agreement on 4-point scales (1 = “strongly disagree” and 4 = “strongly agree”) and then completed the Rosenberg Self-Esteem Scale (RSE; Rosenberg, 1965). The Test of Self-Conscious Affect-3R (TOSCA-3R), developed by Tangney, Dearing, Wagner, & Gramzow (2000) was then given to participants as an outcome measure of proneness to shame and guilt. The TOSCA-3R consists of 16 scenarios with the following prompt (Tangney & Dearing, 2002; p. 207): “Below are situations that people are likely to encounter in day-to-day life, followed by several common reactions to those situations. As you read each scenario, try to imagine yourself in that situation. Then indicate how likely you would be to react in each of the ways described.” Self-Regulatory Perspective on Shame and Guilt 13 Participants are then presented with 16 scenarios describing “common day-to-day situations” such as “You make a big mistake on an important project at work. People were depending on you and your boss criticizes you.” The scenario is followed by descriptions of shame and guilt without using the terms “shame” and “guilt” themselves. Items measuring shame-proneness and guilt-proneness include: “You would feel like you wanted to hide” (shame-proneness) and “You would think ‘I should have recognized the problem and done a better job’” (guilt-proneness). Participants then rate their likelihood of responding in each manner on a 5-point scale (1= “not likely” and 5 = “very likely”). Participants receive separate shame and guilt scores, which are averaged across the scenarios, indicating shame-proneness and guilt-proneness. Results and Discussion Means for the BIS and BAS scales were 2.91 and 3.15 respectively, and scores on these scales were not correlated, r(120) = -.12, n.s. Means for shame and guilt responses were 2.84 and 3.89, respectively, and as reported in past research (e.g., Tangney et al., 1992), these scores were highly correlated, r(120) = .52, p < .001. To analyze the relationship between the BIS and BAS and shame and guilt-proneness, two linear regressions were conducted. When examining the differences between shame and guilt, past researchers have statistically controlled for the high correlation between the two emotions. Similarly, the following analyses also controlled for guilt-proneness when investigating the predictors of shame-proneness, and vice versa. The first linear regression was run to test the predictive value of the BIS on shame responses; BIS, BAS, RSE, and guilt were entered as independent variables. The BIS significantly predicted shame, β = .267, p = .004, even Self-Regulatory Perspective on Shame and Guilt 14 when considering the predictive value of the RSE, β = -.023, p =.014. The BAS did not significantly predict shame, β = -1.58, p = n.s. The second regression was run to test the predictive value of the BAS on guilt responses; BAS, BIS, RSE, and shame were entered as independent variables. The BAS significantly predicted guilt responses, β = .279, p = .009. However, the BIS had no significant predictive value, β = .113, p = .ns. The predictive value of the BAS was found even when considering the RSE in the regression, which also significantly predicted guilt, β = .023, p =.004. These findings support our predictions that an avoidance-based regulatory orientation would be positively associated with shame-proneness, whereas an approachbased regulatory orientation would be positively associated with guilt-proneness. Those with a strong BIS orientation, with an increased sensitivity to negative outcomes, were most likely to report greater levels of shame, but not guilt, after reading the scenarios. In contrast, those participants with a strong BAS orientation, with an increased sensitivity to positive outcomes, were most likely to report greater levels of guilt, but not shame, after reading the scenario. Moreover, the BIS and the BAS predicted shame-proneness and guilt-proneness, respectively, over and above self-esteem, which has been found in previous research to positively predict guilt and negatively predict shame responses (see Tangney & Dearing, 2002). These results provide support for understanding guilt and shame responses as a function of different approach and avoidance regulatory sensitivities. Study 2: Priming Proscriptive and Proscriptive Morality Study 2 further investigated the moral regulatory underpinnings of shame and guilt. As moral emotions, shame and guilt should be associated with moral regulation in Self-Regulatory Perspective on Shame and Guilt 15 particular, beyond their association with approach versus avoidance orientation more generally (as demonstrated in Study 1). Would situationally priming a prescriptive versus proscriptive moral orientation affect subsequent guilt versus shame experiences? In this study we explored whether activating the proscriptive (versus prescriptive) system would increase experiences of shame, whereas activating the prescriptive (versus proscriptive) system would increase experiences of guilt. Method Participants. A total of 101 undergraduate students (66 males and 35 females) participated in this study; 49 were in the prescriptive and 52 in the proscriptive condition of the study. Materials and Procedure. Participants were randomly assigned to either a proscriptive morality prime (ProM) or a prescriptive morality prime condition (PreM). In both cases participants were informed that we were interested in morality: “Each of us has our own way of understanding right and wrong. We are interested in your views. What comes to mind when you think about how to be moral or not be immoral?” PreM respondents were asked to indicate what they should do if their goal is to be moral and not immoral, whereas ProM participants were asked what they should not do. Each group was then provided the phrase “To be moral or not be immoral” followed by 10 lines, each preceded by the stem “I should” (PreM condition) or “I should not” (ProM condition). The State Shame and Guilt Scale (SSGS), developed by Marschall, Sanftner, & Tangney (1994), was given to respondents following the PreM/ProM manipulation. The SSGS is a 15-item self-report scale of “in-the moment (state) feelings” (Tangney and Self-Regulatory Perspective on Shame and Guilt 16 Dearing, 2002, p. 240) of shame, guilt, and pride. The prompt states, “The following are some statements which may or may not describe how you are feeling right now. Please rate each statement using the 5-point scale below. Remember to rate each statement based on how you are feeling right at this moment.” Shame items include “I want to sink into the floor and disappear,” and “I feel small.” Guilt items include “I feel remorse, regret,” and “I feel tension about something I have done.” Participants indicated the extent to which they felt each way on a 5-point scale (1= “not feeling this way at all” and 5= “feeling this way very strongly”). Separate shame and guilt subscores were calculated by averaging relevant responses. Not of interest in this study, the pride items were not analyzed. It is important to note that in this study participants were reporting on felt guilt and shame in the absence of actual transgressive behavior or stimuli priming such reactions. Means on the SSGS were therefore expected to be low (see Marschall et al., 1994) Results and Discussion A mixed-design ANOVA with ProM versus PreM condition as a betweensubjects variable and shame versus guilt as a within-subjects variable produced a significant interaction, F(1,101) = 4.38, p = .04, ηp 2 = .042. A follow-up univariate ANCOVA tested for the effect of condition on shame, controlling for guilt, and revealed that shame was higher in the ProM condition (mean = 1.72, SD = .96) than the PreM condition (mean = 1.54, SD = .65), F(1,101) = 4.52, p = .036, ηp 2 = .044. On the other hand, there was more guilt in the in the PreM condition (mean = 2.08, SD = .99) than the ProM condition (mean =1.96, SD = .98), controlling for shame, F(1,101) = 3.63, p = .06, ηp 2 = .036. As in Study 1 and in previous research, guilt was overall higher than shame Self-Regulatory Perspective on Shame and Guilt 17 across conditions (Tangney & Dearing, 2002, report the SSGS means for guilt are higher than shame). Nevertheless, guilt was significantly higher when participants’ prescriptive rather than proscriptive system was primed. In contrast, shame was significantly higher when participants’ proscriptive system rather than prescriptive system was primed. In this study, participants were not told to focus on any specific transgressions, nor were they focused on the action tendencies engendered by guilt or shame. Mean scores indicating experiences of guilt and shame were therefore low as expected. Nevertheless, simply orienting on what they should or should not do produced differences in reported guilt and shame. Considered with the findings from Study 1 indicating positive associations between approach orientation and guilt and avoidance orientation and shame, Study 2 provides additional support for a moral regulatory perspective on the differences between shame and guilt. Study 3: Transgressions and Their Regulatory Functions So far we have shown that people’s experiences of one moral emotion over the other are influenced by their dispositional self-regulatory orientation and affected by the situational activation of the proscriptive versus prescriptive moral regulatory system. Yet a crucial aspect of shame and guilt is that the two emotions result from a recognized transgression. Thus, Study 3 focused on the regulatory role of specific types of transgressions in engendering either shame or guilt. Transgressions may engender one motivational system over the other: an avoidance motivation increasing distance from the anti-goal, or an approach motivation decreasing distance from a desired goal. When a transgression occurs, either regulatory system can be dominant—it is now a matter of whether the person focuses on immorality, Self-Regulatory Perspective on Shame and Guilt 18 specifically the “should nots” directly reflected in our transgressions themselves, or on morality, and in particular the “shoulds” that represent moral alternatives to our transgressions. The focus is influenced not only by individual differences (Study 1) and by situational activation (Study 2), but may also depend on the type of transgression itself. Some transgressions may be more likely to be part of one moral system over the other, depending on the presence of an anti-goal versus a desired goal. In particular, proscriptive moral regulation focuses particularly on the inhibition of harm-doing and immoral temptations. It includes the domain of many of the popularized “seven deadly sins” that one should not engage in; these are presumably excesses or indulgences that call for inhibition via self-control (Janoff-Bulman, Sheikh, & Baldacci, 2008). In contrast, prescriptive moral regulation promotes positive behaviors; this is the moral system that guides prosocial acts of helping, benevolence and generosity, as well as conduct reflecting the “Protestant ethic”—industriousness, self-reliance, and hard-work, all of which involve the activation of positive behaviors rather than the inhibition of immoral acts and indulgences (see Janoff-Bulman et al., 2008, for further discussion of “approach” and “avoidance” moral domains). The question remains, however, whether transgressions in one or the other domain would be more likely to result in one moral emotion over the other. Any transgression can engage the proscriptive or prescriptive system, and thus shame or guilt, depending on the interpretation by the person. For instance, the act of lying, once enacted, can engage the proscriptive system and a focus on the immorality of lying, resulting in shame and the motivation to inhibit that immorality. Or it can engage Self-Regulatory Perspective on Shame and Guilt 19 the prescriptive system and a focus on the failure to be honest in the interaction, resulting in guilt and the motivation to advance towards honesty. Both systems may be engaged to some extent following a transgression (no doubt accounting in part for the correlations found between guilt and shame), but one system is likely to be activated more strongly, depending on individual differences in motivational orientation or situational activation. In addition, certain types of transgressions themselves may be more likely to engage one system over the other. Thus, engaging in behaviors we think of as temptations, a central domain of proscriptive morality, should be more apt to produce shame than guilt. Violations of this sort are likely to reference a negative outcome (an “anti-goal”) and focus the individual on the immorality of the transgression; by failing to control temptations and restrain negative impulses, there is no desired goal to approach, no “should” to activate. Conversely, a central domain of prescriptive morality is prosocial, benevolent behavior. What transgressions are most apt to engage the prescriptive system via these concerns? Damaged interpersonal relationships (and particularly close relationships; see Baumeister, Stillwell & Heatherton, 1995) are likely to activate the desire to help and care for others. When we transgress by directly hurting others or not helping others, our recognition of what we did wrong is likely to reference a positive outcome or goal—and motivate actions to repair the relationship or make amends for the wrong. Additionally, we propose that failures to advance ourselves via hard work and self-reliance also engender more guilt because a desired goal is present to motivate us into action. The third study used a previously developed scale (see Janoff-Bulman et al., 2009) that measured different domains reflecting either proscriptive or prescriptive Self-Regulatory Perspective on Shame and Guilt 20 regulation to determine whether transgressions regarding the former behavioral domain elicit shame over guilt whereas those regarding the latter domain elicit guilt over shame. We felt a forced choice manipulation would best tease apart and help clarify a number of domain differences between shame and guilt based on a moral regulation framework. Methods Participants provided forced-choice responses indicating shame or guilt to a questionnaire that presented scenarios describing prescriptive and proscriptive transgressions. We expected shame to be greater following proscriptive wrongs and guilt to be greater following prescriptive wrongs. Participants. Participants were 48 (29 women, 17 men, and 2 unreported) undergraduates who received experimental credit for participation. Procedure. Participants completed the Moralisms Scale, a 20-item measure developed by Janoff-Bulman et al. (2009) containing 10 Proscriptive Morality (ProM) scenarios and 10 Prescriptive Morality (PreM) items. In the original questionnaire, each item consisted of a scenario in which the target person is deciding whether or not to engage in a particular behavior. For the current study the items were re-written so that the target person actually did what s/he should not have (ProM) or did not do what s/he should have (PreM). The prompt was, “We all feel bad to some extent after doing something wrong or failing to do something right. We are interested in the different ways that people feel bad. In each of the following situations, please think carefully about how the person is likely to feel. Then check the emotion--either shame or guilt--that you think is most likely felt. Please choose only one emotion for each scenario.” Self-Regulatory Perspective on Shame and Guilt 21 In the development of the questionnaire, the ProM and PreM items were created so that there were no differences in the moral weight of the two scales; that is, participants reported that the ProM targets should not engage in the undesired behaviors to the same extent that they reported that the PreM targets should engage in the desired behaviors (Janoff-Bulman et al., 2009). Although the PreM and ProM items reflect their respective domains (see below), the behaviors selected were intentionally not clearly required or prohibited in order to generate greater variability in respondents across participants in the initial studies. The items were also balanced in terms of self versus interpersonal contexts; that is, half of each scenario set involved behaviors that had an impact on the target person alone, and the other half of each set involved behaviors that clearly affected other people; none of the items involved close, personal relationships. ProM scenarios represented behavioral “failures” involving indulging in personal temptations or behaviors that indicated a desire or willingness to disregard social norms with regard to others in the community. Examples included “excessive” gambling, wearing a skimpy dress to a funeral, painting a house bright pink and purple in a modest, well-kept neighborhood, and going into greater debt to purchase an expensive TV. As a specific example, the latter debt scenario was written as follows: “Sarah is getting more and more into debt with her credit card. She recently bought lots of expensive new clothes and costly furniture for her apartment. She could start saving her money but instead goes out and buys a very expensive hi-definition TV and goes into even deeper debt.” PreM scenarios represented behavioral “failures” involving benevolence or industriousness, and included stiffing an excellent waiter, regularly skipping an important Self-Regulatory Perspective on Shame and Guilt 22 early morning class, not helping an elderly person who is having trouble carrying her groceries, not working hard enough to meet an important work deadline, and walking past a homeless person who asks for some money. The latter scenario, for example, was written as follows: “Mary walks by a homeless man on the street, and he asks if she can spare some change. Mary knows there’s a local shelter that costs $2.00 a night. She could give him $2.00, but walks past the homeless man instead.” In each case, participants were asked whether the target person in the scenario was more likely to feel shame or guilt. For the ProM example, “Sarah now feels really bad. Which emotion do you think she feels the most?” and for the PreM example, “Mary now feels really bad. Which emotion do you think she feels the most?” Participants were provided a forced choice between guilt and shame. Results The number of times participants chose guilt and the number of times they chose shame were aggregated separately for the PreM and the ProM vignettes. In other words, each participant ended up with two scores: a proportion for guilt to shame responses for the PreM and a proportion for guilt to shame responses for the ProM vignettes. A single analysis therefore compared these two proportions (which would be the same whether run on the guilt or the shame proportions). As expected, participants indicated that more judgments of guilt would be elicited in the prescriptive than proscriptive vignettes, or, alternatively, more judgments of shame would be elicited in the proscriptive than prescriptive vignettes, t(45) = 4.560, p < .001. More specifically, guilt reactions were chosen more often for the PreM than ProM items, .67 to .43; and shame reactions were chosen more often for the ProM than PreM items, .57 to .33. In other words, Self-Regulatory Perspective on Shame and Guilt 23 transgressions reflecting failures of restraint and self-control in the face of temptations (i.e., body-related transgressions, gambling, over-spending) were particularly likely to elicit shame, whereas transgressions reflecting failures to work hard or help needy others were particularly likely to elicit guilt. The differences found in this study could not be attributed to a distinction between interpersonal versus non-interpersonal contexts, nor to the severity of the transgressions (see development of questionnaire above) . However, the scenarios depicted specific behaviors that may not reflect people’s everyday experiences of guilt and shame. Would transgressions generated by participants themselves reflect differences found by the Moralisms Scale? Generalizability of Study 3 to everyday experiences would be increased if participants indeed generate proscriptive versus prescriptive behaviors for shame versus guilt experiences, respectively. Additionally, the results from Study 3 would be replicated using a very different method. In a brief followup study, we therefore used autobiographical micro-narratives of guilt and shame to investigate if peoples’ accounts of guilt and shame experiences reflect the domain differences predicted by proscriptive versus prescriptive regulation. Follow-up: Micro-Narratives.Participants were 145 undergraduates (101 women and 44 men) who were randomly assigned to one of two study conditions: a shame condition or a guilt condition. They were asked, “We all fall short of our moral standards at times. We would like you to think back on a time when you felt shame. With this situation in mind, please answer the following questions. What happened to make you feel shame? What did you think or feel in the situation?...” Participants in the guilt condition were given the same prompt with “guilt” substituted for “shame.” Two Self-Regulatory Perspective on Shame and Guilt 24 independent raters coded participants’ responses (all kappas above .85) for whether the participants’ response fit into each of the coding categories noted below (either “yes” or “no”). Twelve responses fit more than one coding category. Raters found 33 (11 shame, 22 guilt) omission responses in all, but more were elicited in the guilt than shame condition, χ2(1, N = 33) = 3.67, p = 0.056. In general these responses began with “I didn’t,” “I forgot to,” “I couldn’t” followed by the desirable behavior that the respondent failed to activate. It is not surprising that omissions would be more apt to activate the prescriptive system; the recognition that one did not do what one should have done (omission) more readily suggests a focus on approaching a desired goal (engaging in the omitted action) and reinstating one’s own morality. The vast majority of the transgressions recounted by participants were commissions, which were reported in both shame and guilt conditions. Seven types of transgressions emerged most frequently and accounted for almost all of the responses: the most common transgression was lying (total=32, shame=15, guilt=17), followed by hurting others emotionally (total=23, shame=8, guilt=15), and cheating on a partner (total=14, shame=4, guilt=10), all of which involved individuals whom the participants knew personally. Engaging in sexual behaviors (total=13, shame=10, guilt=3), drug or alcohol use (total=12, shame=11, guilt=1), stealing (total=9, shame=2, guilt=7), and academic cheating (total=6, shame=3, guilt=3) were also reported. The relationship transgressions (by aggregating lying, hurting others’ emotionally, and cheating on a partner) were reported more in the guilt condition than shame condition, χ2(1, N = 61) = 3.69, p = 0.055. This difference was accounted for in Self-Regulatory Perspective on Shame and Guilt 25 particular by cheating on a partner and emotionally hurting others, behaviors that generally involved close others. Moreover, engaging in sexual behaviors and drug or alcohol use were reported more in the shame condition, χ2(1, N = 13) = 3.77, p = 0.05, and χ2(1, N = 12) = 6.23, p = 0.01, respectively. Stealing and academic cheating, situations that do not entail people close to the participants (none of the stealing responses involved people the participants knew personally) were elicited least frequently, but in both shame and guilt conditions. Although Study 3 found that failures of advancing oneself academically elicited more guilt, academic cheating in this study was reported in both guilt and shame conditions. We suggest academic cheating in particular may elicit both shame and guilt, whereas other behaviors concerning academic performance may elicit more guilt than shame. Discussion Study 3 supports the findings in Studies 1 and 2 of associations between proscriptive regulation and shame and between prescriptive regulation and guilt. In addition, it extends the findings of Studies 1 and 2 by implicating different domains regulated by prescriptive versus proscriptive morality: transgressions reflecting failures of inhibition and lack of self-control—such as body-related transgressions and failures to restrain from eating excessively, gambling, and overspending—appear to be associated more with shame. On the other hand, transgressions reflecting failures of prosociality and industriousness—such as not helping others, letting oneself and others down, and even failures in the work and academic environment—as well as transgressive omissions appear to be associated more with guilt. Self-Regulatory Perspective on Shame and Guilt 26 Tangney and colleagues (e.g., Tangney and Dearing, 2002, p.17) have noted that they find “very few, if any, ‘classic’ shame-inducing or guilt-inducing situations...” We agree that whether shame or guilt is experienced after a transgression depends on the interpretation by the individual. However, it appears that some transgressions are more likely to engender one interpretation over another and are thus more likely to elicit one moral emotion over the other. Fredrickson and Roberts (1997) have shown that cultural practices of sexual objectification increase women’s self-objectification, resulting in “body shame,” and eating disorders such as anorexia nervosa and bulimia have also been positively associated with shame (e.g., Calogero, Davis, & Thompson, 2005; Sanftner, Barlow, Marschall, & Tangney, 1995); Nussbaum (2005) has more generally noted shame’s focus on one’s own body. Kurtz (e.g., 2007) has argued that recovering alcoholics experience specifically high levels of shame, and alcohol and/or drug abuse has been found to be positively correlated with shame, while negatively or negligibly correlated with guilt (Dearing, Stuewig, & Tangney, 2005). Tavares, Martins, Zilberman, & el-Guabaly (2002) provide evidence for high shame among gamblers seeking clinical treatment. In contrast, Baumeister et al. (1994) have shown that guilt arises from the attachment system and is especially experienced in close, communal relationships. The prescriptive and proscriptive systems provide a self-regulatory framework to understand why these domain differences occur. Given that proscriptive regulation involves overcoming negative temptations and restraint from “bad” behavior, shamerelated domains often involve self-perceived “excesses” and “indulgences” related to drinking, eating, gambling, and the body more generally (e.g., Fredrickson & Roberts, 1997; Sanftner et al., 1995). On the other hand, given that prescriptive regulation entails Self-Regulatory Perspective on Shame and Guilt 27 overcoming inertia in order to activate positive behaviors, it is most readily enacted in moral behaviors concerning the well-being of others, particularly those we care about, and thus guilt is experienced especially in relationship contexts (Baumeister et al., 1994). It appears that transgressions are more apt to be associated with one or the other moral emotion depending on the focus on desired goals or anti-goals, “shoulds” or “should nots.” General Discussion Overall the results of these three studies provide support for a self-regulatory perspective on shame and guilt. Although the two emotions are highly correlated, we found self-regulation to tease the two apart: In Study 1, individual differences in approach versus avoidance orientation differentially predicted guilt versus shameproneness respectively, over and above the predictive value of self-esteem. Study 2 found that situationally priming proscriptive versus prescriptive morality increased state shame but not guilt, whereas priming prescriptive versus proscriptive morality increased state guilt but not shame. And Study 3 and its micro-narrative follow-up showed that transgressions that engaged proscriptive regulation engendered more shame than guilt while transgressions that engaged prescriptive regulation engendered more guilt than shame. Together the studies provide evidence for self-regulatory differences across personality and situational aspects of shame and guilt (see Fontaine, Luyten, De Boeck, Corveleyn, Fernandez, Herrera, et al., 2006, for the importance of capturing the personality and the situational variations of shame and guilt). Although previous researchers (e.g., Ferguson et al., 1991; Schmader and Lickel, 2006) have noted that action tendencies motivated by guilt and shame resemble approach Self-Regulatory Perspective on Shame and Guilt 28 versus avoidance behaviors, to date differences between the two moral emotions have yet to be informed by self-regulation theory. Proscriptive versus prescriptive regulation provides a theoretical framework--and the above studies the first empirical evidence--for distinguishing between shame and guilt using self-regulation theory (e.g., Carver and Scheier, 1990, 1998, 2008; Gray, 1982, 1994). Shame-prone individuals are likely to be highly avoidance-oriented, whereas guilt-prone individuals are likely to be highly approach-oriented. Situational factors that promote behavioral inhibition and the salience of punishments and threats will increase the likelihood of shame, whereas situational factors that promote behavioral activation and the salience of rewards and incentives will increase the likelihood of guilt. Further, a prescriptive-proscriptive distinction highlights regulatory differences inherent in different types of transgressions: giving in to prohibited temptations and a lack of self-control, behaviors from which there is no positive recourse (that is, the moral response is abstention, or no action), are most apt to elicit shame. Transgressions that focus one’s response on the failure to help or care for others, behaviors that provide the opportunity for moral redemption, are more apt to elicit guilt. Interestingly, however, both regulatory systems may be engaged after a given transgression such that shame and guilt are experienced simultaneously. For instance, lying to a partner may elicit both guilt and shame, for a person may feel guilt for failing to be a “good,” honest person and also feel shame for failing to restrain one’s “bad,” deceptive behavior. This self-regulatory perspective on shame and guilt does not contradict past work, but rather provides a broad interpretive framework for understanding and extending past findings. The proscriptive-prescriptive differences in moral regulation are compatible Self-Regulatory Perspective on Shame and Guilt 29 with the widely supported attributional perspective (e.g., Tangney et al., 2007), in which shame reflects a focus on one’s entire self while guilt reflects a focus on one’s behavior. 4 For instance, researchers note that shame is much more painful because it “results from actions that reveal the self to be flawed or defective” (Haidt, 2003, p. 860; see also, Lewis, 1971, Lindsay-Hartz, 1984; Tangney and Dearing, 2002). In terms of moral regulation, the proscriptive moral system is stricter, harsher, and more demanding, and violations engender greater punishment than prescriptive violations. Not surprisingly, this appears to be reflected in our responses to transgression; that is, proscriptive violations are more punishing and thereby engage the entire “bad” self rather than a specific behavior. The prescriptive moral system is less harsh, more discretionary, and more focused on positive outcomes; guilt’s focus on specific behavior thus provides an opportunity to readily advance subsequent morality. Concerning the broader implications of a self-regulatory perspective, shame and guilt may be associated with different orientations regarding moral regulation at a societal level. Proscriptive regulation entails the “should nots” that restrain people from behaviors that violate group norms and the social order while prescriptive regulation entails the “shoulds” that activate positive obligations towards others, thus fostering group interdependence (Janoff-Bulman et al., 2008). In this sense, the social functions of shame and guilt are likely to differ in some aspects as well: shame may partly function to maintain and promote existing social norms and order while guilt may partly function to advance prosocial interactions and communal interdependence. And indeed, Haidt (2003) has argued that shame functions to submit people to social hierarchy, while Baumeister et al. (1994) have provided evidence for guilt as fostering communal Self-Regulatory Perspective on Shame and Guilt 30 relationships (also see Lickel, et al, 2005, 2006). Thus, a self-regulatory account of shame and guilt implicates different moral regulatory functions at the societal level as well. Linking shame and guilt to self-regulation suggests the possibility of drawing broadly on this literature to more fully inform differences between the two moral emotions. Questions regarding developmental trajectories, neural substrates, associations with other emotions, broader self-evaluative implications, and behavioral strategies may benefit from this explanatory account. For instance, are there distinct neural substrates of these two moral emotions, and are they associated with differences already found in the self-regulation literature (e.g., Amodio et al., 2007; Davidson, 1995, 1998; Davidson, Ekman, Saron, Senulis, & Friesen, 1990; Sutton & Davidson, 1997)? We believe the distinction between proscriptive and prescriptive moral regulation, reflecting differences in approach and avoidance orientations, provides a framework that consolidates previous research and presents new questions for future investigations on shame and guilt. Self-Regulatory Perspective on Shame and Guilt 31 Endnotes 1. Other constructs measured in the TOSCA-3R are externalization, detachment, Alpha pride, and Beta pride, none of which were of interest in the present study and not included in the analyses. 2. Janoff-Bulman et al. (2009) also found responses to the prescriptive and proscriptive items to be uniquely and positively correlated with the BIS and the BAS Scales (Carver and White, 1994), respectively, such that the BIS Scale uniquely predicted the extent to which participants believed the actor should not engage in the ProM behavior, whereas the BAS Scale uniquely predicted the extent to which participants believed the actor should engage in the PreM behavior. 3. Participants reporting multiply coded transgressions were excluded from this analysis. 4. Study 2 in particular provides support for the regulatory bases of these self versus behavior differences. In the SSGS (Marschall et al., 1994) shame is assessed primarily by self-focused items such as “I feel like I am a bad person,” and guilt is measured primarily by behavior-focused items such as “I cannot stop thinking about something bad that I have done.” Increased shame following a proscriptive prime, and increased guilt following a prescriptive prime therefore suggests that attributional differences associated with shame and guilt are consistent with a self-regulatory account of these moral emotions. Self-Regulatory Perspective on Shame and Guilt 32