Guilt

Abstract

Guilt is a type of emotion: the negatively valenced, self-conscious moral emotion that follows a person's own real or imagined transgression. This article covers how guilt is defined and measured, how it differs from shame in what is evaluated and in what it motivates, its interpersonal functions of repair and compensation, its early development between 22 and 45 months of age, the brain regions and the depression-specific disconnection associated with it, and its divergent relations to depression, anxiety and prosocial behaviour. The central dispute is whether guilt is inherently adaptive, a question that turns on which variant of guilt is measured and how. Three interactive demos show the behaviour-versus-self attribution that separates guilt from shame, the meta-analytic correlations of guilt and shame with prosocial orientation, and the discharge of interpersonal guilt as compensation or self-punishment.

Keywords: guilt, shame, self-conscious emotions, moral emotions

Guilt occupies an unusual position among the emotions. Moral emotions represent a key element of the human moral apparatus, influencing the link between moral standards and moral behaviour, and the field's attention has long been focused on a triad of negatively valenced self-conscious emotions, shame, guilt and embarrassment (Tangney et al., 2007). Of the three, guilt is the one that the research literature has both described as adaptive and blamed for suffering. A critical review of the recent literature puts the asymmetry plainly: on one side the destructiveness of guilt is emphasised, since it inflicts punishment and pain on individuals for their errors and can lead to psychopathology, while on the other side it is described as a friendly emotion that motivates behaviour adapted to social and cultural rules (Carnì et al., 2013). Much of the history of the topic is the attempt to say which guilt is which.

What Guilt Is

Guilt is defined, in the most common research usage, by its object. A review of 23 theory-based definitions of guilt and 25 measures of guilt found that two broad conceptualisations emerge, and that central to both is a focus on one's action or inactions involving real or imagined moral transgressions; what distinguishes the two is whether or not guilt is an inherently adaptive construct, generating remorse, augmenting a sense of responsibility, and motivating reparation (Tilghman-Osborne et al., 2010). The same review found that research on the relation of guilt to psychopathology is highly inconsistent, that measures of guilt do not correspond well to the definitions from which they derive, and that many definitions and measures reflect the intrusion of extraneous constructs that could confound guilt research (Tilghman-Osborne et al., 2010). The definitional disorder is therefore not incidental to the topic. It may be the reason the same emotion can be reported as protective in one study and pathogenic in the next.

The emotion is common. In a nationwide telephone survey of 1,003 German adults, the prevalence of current guilt feelings was 10.6%, about one fourth of those with current guilt feelings rated them as rather strong or very strong, and feeling guilty was unrelated to age, sex and education but significantly associated with depression (Luck & Luck-Sikorski, 2021). The weighted prevalence of guilt feelings in adults with major depression was 37.4%, compared with 8.1% in adults without (Luck & Luck-Sikorski, 2021). Guilt is thus an ordinary feature of adult life and, at the same time, several times as prevalent among adults with major depression as among those without.

Guilt belongs to the family of self-conscious emotions, which are fundamentally important to a wide range of psychological processes yet have received relatively little attention compared with the more basic emotions such as sadness and joy (Tracy & Robins, 2004). The theoretical model of self-conscious emotions outlines the features that distinguish them from basic emotions and argues that generally accepted models of basic emotions do not adequately capture the self-conscious emotion process (Tracy & Robins, 2004). What the self-conscious emotions have in common is that they require the person to represent the self, and often the mental states of others, which is why their study overlaps with social cognition as much as with the psychology of affect.

The scope of the construct has also widened. The review of moral emotions highlights several newer areas of research, including the phenomena of vicarious or collective experiences of shame and guilt, the link between childhood abuse and later proneness to shame, and the expansion of the concept of moral emotions to include the positive emotions of elevation, gratitude and the sometimes morally relevant experience of pride, alongside the morally relevant emotional process of other-oriented empathy (Tangney et al., 2007). Vicarious or collective guilt, felt on behalf of a group one belongs to, is thus listed in that review among the newer areas of research on the same emotion.

Guilt and Shame

The modern literature on guilt is, to a large degree, a literature on how guilt differs from shame. The distinction is usually credited to Helen Block Lewis, whose 1971 book set shame as a negative evaluation of the global self against guilt as a negative evaluation of a specific behaviour, and whose further notion that the two emotions are differentially related to unique symptom clusters was tested and not supported in two studies of 245 and 234 undergraduates (Tangney et al., 1992; Tangney et al., 2007). The behaviour-versus-self formulation has since been described as the currently most dominant basis for distinguishing between shame and guilt (Tangney et al., 2007). It is the basis of the Guilt and Shame Proneness scale, whose two guilt subscales assess negative behaviour-evaluations and repair action tendencies following private transgressions, and whose two shame subscales assess negative self-evaluations and withdrawal action tendencies following publicly exposed transgressions (Cohen et al., 2011). In the scale's validation, both guilt subscales were highly correlated with one another and negatively correlated with unethical decision making, whereas the two shame subscales were only weakly correlated with one another and their relationships with unethical decision making diverged: shame as negative self-evaluation constrained unethical decision making, but shame as withdrawal did not (Cohen et al., 2011).

Figure 1

The Behaviour-Versus-Self Distinction Between Guilt and Shame

The folk distinction, that shame is public and guilt private, does not survive the data. When 182 undergraduates described personal embarrassment, shame and guilt experiences, shame was no more likely than guilt to be experienced in public situations; all three emotions typically occurred in social contexts, but a significant proportion of shame and guilt events occurred when respondents were alone (Tangney et al., 1996). Phenomenological ratings in the same study demonstrated that shame, guilt and embarrassment are not merely different terms for the same affective experience, that embarrassment was a relatively distant neighbour of shame and guilt, and that the differences among the three could not be explained simply by intensity of affect or by degree of moral transgression; participants were generally their own harshest critics in each type of event (Tangney et al., 1996).

The behaviour-versus-self account has rivals. One recent reconsideration argues that some of the current distinctions between shame and guilt need to be redrawn, and proposes instead the kind of self-evaluation involved, inadequacy versus harmfulness; one's focus on the perceived discrepancy between actual and ideal self versus one's focus on the perceived responsibility for one's fault; and consequently the different domains of self-esteem involved (Miceli & Castelfranchi, 2018). The authors note that these criteria have in part been suggested in the literature and that integrating them allows a better account of the two emotions' possible coexistence or the shift from one to the other (Miceli & Castelfranchi, 2018). Whether the decisive variable is the object of evaluation or the content of the evaluation remains an open question, and the two accounts make different predictions about cases in which a harmful act is also read as evidence of an inadequate self.

Types of Guilt

The Medical Subject Headings thesaurus files guilt under its parent descriptor Emotions (tree F01.470) and gives it one direct subtype, listed in Table 1. The placement is an indexing artefact rather than a theoretical claim: MeSH is an indexing classification for the biomedical literature, not a taxonomy of the mind, and the psychological literature reviewed above treats shame as a distinct emotion that shares variance with guilt rather than as a kind of guilt. Subtypes in MeSH need not be mutually exclusive, and a paper may be indexed under both terms.

The research literature draws its own divisions within guilt, and these matter more for prediction than the thesaurus does. A meta-analysis of the relation between guilt and depressive symptoms distinguished two maladaptive variants: contextual-maladaptive guilt, involving exaggerated responsibility for uncontrollable events, and generalised guilt, involving free-floating guilt divorced from specific contexts (Kim et al., 2011). A meta-analysis of anxiety symptoms reached the same partition, finding that two types of guilt seem to be equally maladaptive as shame, generalised guilt, involving a free-floating guilt separated from specific contexts, and contextual-maladaptive guilt, involving an inappropriate or exaggerated feeling of responsibility (Cândea & Szentagotai-Tătar, 2018). Guilt that is tied to a specific, controllable act is the variant the adaptive accounts describe; guilt that floats free of any act, or attaches to events the person could not control, behaves statistically like shame.

Guilt as an Interpersonal Emotion

A second organising idea treats guilt as something that happens between people as much as inside them. A review of multiple sets of empirical findings concluded that guilt appears to arise from interpersonal transactions, including transgressions and positive inequities, and to vary significantly with the interpersonal context; guilt patterns appear to be strongest, most common and most consistent in communal relationships, which are characterised by expectations of mutual concern (Baumeister et al., 1994). On this account guilt serves various relationship-enhancing functions, including motivating people to treat partners well and avoid transgressions, minimising inequities and enabling less powerful partners to get their way, and redistributing emotional distress (Baumeister et al., 1994). Guilt, in other words, is part of the machinery by which close relationships police themselves.

The behavioural evidence is consistent with this. Using a dyadic social dilemma game and an everyday cooperation measure, one pair of experiments replicated the finding that guilt increases cooperation for proselfs, and found no such effect for shame, concluding that the effects of moral emotions on cooperative behaviour can only be understood if the specific moral emotion is known (de Hooge et al., 2007). The link between guilt and altruism is, however, smaller and more method-dependent than the theory suggests. A meta-analysis of dispositional guilt and shame and pro-social orientation found that, among studies employing zero-order correlations, dispositional guilt was positively correlated with pro-social orientation (k = 63, mean r = .13), whereas dispositional shame was negatively correlated (k = 47, mean r = -.05); test format was a significant moderator for guilt studies only, with scenario measures producing significantly stronger effects, and semi-partial correlations resulted in significantly stronger effects among guilt and shame studies (Tignor & Colvin, 2017). The authors conclude that although dispositional guilt and shame are differentially related to pro-social orientation, such relationships depend largely on the methodological choices of the researcher, particularly in the case of guilt (Tignor & Colvin, 2017).

The repair tendency has a dark twin. The scientific literature has mainly highlighted the beneficial interpersonal consequences of guilt, showing repeatedly that guilt motivates compensatory pro-social behaviour to repair social bonds, but when opportunities for compensation are not present, guilt may evoke self-punishment, demonstrated through self-denied pleasure in a scenario study and self-enforced penalties in an experimental study (Nelissen & Zeelenberg, 2009). The authors named this tendency the Dobby Effect and discussed it as an explanation for the widely held conviction that atonement absolves sins, for its contribution to some types of psychopathology, and for its possible functional relevance (Nelissen & Zeelenberg, 2009). The finding matters because it suggests that the motivational core of guilt is not repair as such but the discharge of a debt, which repair normally serves and self-punishment can substitute for.

Guilt's social function has also been tested longitudinally. In a study of 476 jail inmates assessed for shame proneness, guilt proneness and externalisation of blame shortly after incarceration, with 332 interviewed one year after release and official arrest records accessed for 446, guilt proneness negatively and directly predicted reoffence in the first year after release, and shame proneness did not (Tangney et al., 2014). Mediational modelling showed that shame proneness positively predicted recidivism via its robust link to externalisation of blame, while there remained a marginal direct effect (p = .052) in which, unimpeded by defensive externalisation of blame, shame inhibited recidivism, a pattern driven primarily by items assessing a motivation to hide (Tangney et al., 2014). Guilt, on this evidence, does the inhibitory work that shame is popularly credited with.

The Development of Guilt

Guilt is observable before children can describe it. In a longitudinal study of the early development of guilt, 106 children were observed in laboratory paradigms in which they were led to believe that they had damaged valuable objects, during two separate sessions at each of the assessments at 22, 33 and 45 months (Kochanska et al., 2002). The behavioural and affective components of guilt cohered significantly across the sessions, converged with each other, were stable across all the assessments, and corresponded modestly with maternal reports; most components decreased with age, except for bodily tension, which increased, and at 33 and 45 months girls displayed more guilt than did boys (Kochanska et al., 2002).

The study also located guilt within temperament and parenting. Children who were more fearful in typical fear-inducing paradigms also displayed more guilt, and children of mothers who relied on more power-assertive discipline, observed and self-reported, displayed less guilt (Kochanska et al., 2002). Guilt related positively to the development of self at 18 months and to moral self at 56 months, children who displayed more guilt were less likely to violate rules of conduct at 56 months, and a mediational model was supported in which fearful temperament contributed to guilt proneness, which in turn served to inhibit children's tendency to violate rules (Kochanska et al., 2002). The developmental finding and the recidivism finding thus agree across five decades of age: guilt proneness predicts rule-compatible conduct, and it does so through the behaviour rather than around it.

The Neural Basis of Guilt

Imaging studies of guilt began by asking whether the self-conscious emotions share a neural basis with social cognition. In an early study, 19 healthy volunteers read sentences carrying neutral, guilty or embarrassing contents during scanning; both guilt and embarrassment conditions commonly activated the medial prefrontal cortex, left posterior superior temporal sulcus and visual cortex, while compared with guilt the embarrassment condition produced greater activation in the right anterior temporal cortex, bilateral hippocampus and visual cortex (Takahashi et al., 2004). Most of these regions have been implicated in the neural substrate of social cognition or theory of mind, and the authors read their results as supporting the idea that both are self-conscious emotions, social emotions requiring the ability to represent the mental states of others (Takahashi et al., 2004).

Later work asked what is specific to guilt. Using an autobiographical memory paradigm in which participants relived situations from their own past associated with strong feelings of guilt, shame or sadness, one study found that compared with the control emotions, guilt episodes specifically recruited a region of right orbitofrontal cortex, which was also highly correlated with individual propensity to experience guilt, and guilt-specific activity in the paracingulate dorsomedial prefrontal cortex, a critical theory-of-mind region that overlapped with brain areas of self-referential processing identified in an independent task (Wagner et al., 2011). A systematic review of 21 functional and structural imaging studies of shame, embarrassment and guilt found that, although studies differed considerably in methodology, guilt was more likely to be associated with activity in ventral anterior cingulate cortex, posterior temporal regions and the precuneus, shame with the dorsolateral prefrontal cortex, posterior cingulate cortex and sensorimotor cortex, and embarrassment with the ventrolateral prefrontal cortex and amygdala (Table 2); the review concluded that results point to some common and some distinct neural underpinnings of these emotions and that further research is required to replicate findings (Bastin et al., 2016).

Table 2. Human imaging findings on guilt, as reported in the cited abstracts.

Note. Each row lists only the regions named in that paper's abstract for the guilt condition or contrast (Takahashi et al., 2004; Wagner et al., 2011; Yu et al., 2014; Green et al., 2012; Bastin et al., 2016). Paradigms, samples and contrasts differ, so the rows are not directly comparable.

Interpersonal guilt has been studied with tasks in which a participant's error actually harms another person. In one such study, participants played multiple rounds of a dot-estimation task with anonymous partners while undergoing scanning; the partner would receive pain stimulation if the partner or the participant or both responded incorrectly, and the participant was then given the option to intervene and bear a proportion of pain for the partner (Yu et al., 2014). The level of pain voluntarily taken and the activations in anterior middle cingulate cortex and bilateral anterior insula were higher when the participant was solely responsible for the stimulation than when both committed an error, grey matter volume in the anterior middle cingulate predicted the individual's compensation behaviour, and a mediation pathway analysis revealed that activation in a midbrain region mediated the relationship between anterior middle cingulate activation and the tendency to compensate (Yu et al., 2014).

The most recent step is a multivariate one. Across two independent neuroimaging datasets that used interpersonal interactions to evoke guilt, a distributed guilt-related brain signature was identified that discriminated conditions associated with interpersonal guilt from closely matched control conditions in a cross-validated training sample of 24 Chinese participants and in an independent test sample of 19 Swiss participants (Yu et al., 2020). The signature did not respond to observed or experienced pain, or to recalled guilt, and exhibited only weak spatial similarity with other brain signatures of social-affective processes, which the authors take as indicating the specificity of the brain state it represents (Yu et al., 2020). That the signature for interpersonal guilt did not respond to recalled guilt is itself a finding: the guilt evoked by harming a partner in the scanner and the guilt evoked by remembering a past wrong may not be one brain state.

Guilt in Psychopathology

Guilt has a long clinical reputation, and the quantitative literature has spent three decades trimming it. In the two undergraduate studies that tested the proposal of differential symptom clusters, shame-proneness was strongly related to psychological maladjustment in general, whereas guilt-proneness was only moderately related to psychopathology and its correlations were ascribable entirely to the shared variance between shame and guilt; shame accounted for substantial variance in depression above and beyond a depressogenic attributional style (Tangney et al., 1992). The meta-analytic picture is the same. Across 242 effect sizes from 108 studies employing 22,411 participants, shame showed significantly stronger associations with depressive symptoms (r = .43) than guilt (r = .28), although the association of shame with depressive symptoms was statistically indistinguishable from the associations of two maladaptive variants of guilt (r = .39 and r = .42); external shame was associated with larger effect sizes (r = .56) than internal shame (r = .42), and depressive symptom measures that invoked the term guilt yielded stronger associations between guilt and depressive symptoms (r = .33) than those that did not (r = .21) (Kim et al., 2011). The authors note that these correlational data are ambiguous with respect to their causal interpretation, and conclude that shame should figure more prominently in understandings of the emotional underpinnings of depressive symptoms (Kim et al., 2011).

The same asymmetry holds for anxiety. A meta-analysis of shame-proneness, guilt-proneness and anxiety symptoms found that in most cases shame was more strongly associated with anxiety symptoms, with in general medium effect sizes, than guilt, with in general small effect sizes, and that when controlling for the shared variance of shame and guilt, in most cases only shame remained significantly associated with anxiety symptoms; external shame seemed to be more strongly associated with social anxiety symptoms than internal shame (Cândea & Szentagotai-Tătar, 2018). Taken together, the meta-analyses say that ordinary guilt is a weak correlate of internalising symptoms once shame is controlled, and that the guilt which does predict them is the free-floating or exaggerated-responsibility kind.

Depression research has nonetheless kept guilt at its centre, because the cognitive model of depressive vulnerability is a model of self-blame. Cognitive models predict that vulnerability to major depressive disorder is due to a bias to blame oneself for failure in a global way, resulting in excessive self-blaming emotions, decreased self-worth, hopelessness and depressed mood (Zahn et al., 2015). In 132 patients with remitted major depression assessed with a phenomenological interview, with moral-emotion items administered to 94 of them, feelings of inadequacy, depressed mood and hopelessness emerged as the most closely co-occurring and consistent symptoms, self-blaming emotions occurred in most patients, with self-disgust or contempt being more frequent than guilt, followed by shame, while anger or disgust towards others was experienced by only 26%; 85% of patients reported feelings of inadequacy and self-blaming emotions as their most bothering symptoms (Zahn et al., 2015). The authors conclude that self-blaming emotions were highly frequent and bothering but not restricted to guilt, which calls for a refined assessment of self-blaming emotions (Zahn et al., 2015).

The neural correlate of that vulnerability is guilt-selective. Recent imaging studies in healthy participants had shown that moral feelings such as guilt activate representations of social meaning within the right superior anterior temporal lobe, and that this area was selectively coupled with the subgenual cingulate cortex and adjacent septal region during the experience of guilt compared with indignation (Green et al., 2012). In a case-control study of 25 patients with remitted major depression and 22 controls, the prediction of a guilt-selective reduction in coupling between the anterior temporal lobe and the subgenual cingulate in depression was corroborated, with additional medial frontopolar, right hippocampal and lateral hypothalamic areas of decoupling while controlling for medication status and intensity of negative emotions, and lower levels of coupling were associated with higher scores on a validated measure of overgeneralised self-blame (Green et al., 2012). Vulnerability to major depression, the authors conclude, is associated with temporofrontolimbic decoupling that is selective for self-blaming feelings (Green et al., 2012). Guilt in depression is thus not simply more guilt; it is guilt that has lost its connection to the representation of what, specifically, was done to whom.

Worked Example

The meta-analytic and epidemiological numbers in this literature repay conversion into shares of variance and relative risks. The depressive-symptom meta-analysis reports a correlation of r = .43 for shame and r = .28 for guilt across 242 effect sizes from 108 studies employing 22,411 participants (Kim et al., 2011). Squaring the correlations gives the shared variance: 0.43 squared is 0.185, so shame shares about 18% of its variance with depressive symptoms, while 0.28 squared is 0.078, about 8%. The two maladaptive variants of guilt, at r = .39 and r = .42, share about 15% and 18% respectively, which is why their associations were statistically indistinguishable from shame's (Kim et al., 2011). The pro-social meta-analysis reports mean correlations of .13 for guilt over 63 studies and -.05 for shame over 47 (Tignor & Colvin, 2017); squared, these are 1.7% and 0.25% of variance, a reminder that dispositional guilt's much-cited link to pro-social orientation is reliable but small.

The epidemiological survey reports a weighted prevalence of current guilt feelings of 37.4% in adults with major depression and 8.1% in adults without, in a sample of 1,003 adults (Luck & Luck-Sikorski, 2021). The ratio 37.4 / 8.1 is 4.6, so current guilt feelings were roughly four and a half times as prevalent among depressed as among non-depressed adults; the difference of 29.3 percentage points is the absolute excess. The recidivism study assessed 476 inmates, interviewed 332 one year after release and accessed arrest records for 446 (Tangney et al., 2014): 332 / 476 is 69.7% retention for the interview and 446 / 476 is 93.7% for records, a retention rate the authors report as comparing very favourably with other longitudinal inmate studies. The imaging case-control study compared 25 patients with remitted depression and 22 controls (Green et al., 2012), a total of 47 participants; the guilt-signature study trained on 24 and tested on 19 (Yu et al., 2020), a total of 43. The systematic review's call for replication should be read against sample sizes of this order (Bastin et al., 2016).

Discussion

Three lessons follow for cognitive psychology. First, guilt shows how much of an emotion's reputation depends on how it is measured. The review of definitions found that measures of guilt do not correspond well to the definitions from which they derive and that definitions and measures change with developmental level (Tilghman-Osborne et al., 2010); the pro-social meta-analysis found that, for guilt studies only, scenario measures produced significantly stronger effects than checklists and that semi-partial correlations produced stronger effects than zero-order ones (Tignor & Colvin, 2017); and the depression meta-analysis found that symptom measures which invoked the term guilt yielded stronger guilt-depression associations than those that did not (Kim et al., 2011). A substantial part of the disagreement about whether guilt is adaptive is disagreement about instruments. The sharpest form of that disagreement concerns the scenario-based Test of Self-Conscious Affect, which is used in many studies addressing shame-proneness versus guilt-proneness and relies on the self-versus-behaviour distinction: its items designed to assess shame-proneness mainly refer to global negative evaluations of the self and the tendency to withdraw, whereas those designed to assess guilt-proneness focus on specific negative evaluations of one's behaviour and the tendency to make amends, and its shame scale has been found to correlate strongly with both chronic shame and chronic guilt, whereas its guilt scale related to reparative action tendencies but showed no or weak relation to guilt feelings (Miceli & Castelfranchi, 2018). Whether the instrument measures maladaptive aspects of guilt and adaptive aspects of shame at all was the question posed by an early empirical investigation of it (Luyten et al., 2002), and the recidivism, psychopathology and pro-social findings reviewed above rest largely on measures of this family.

Second, the behaviour-versus-self distinction has earned its place by prediction, not only by definition. Guilt-proneness negatively and directly predicted reoffence while shame-proneness predicted recidivism via externalisation of blame (Tangney et al., 2014); guilt increased cooperation among proselfs where shame did not (de Hooge et al., 2007); and the guilt subscales of a proneness measure were negatively correlated with unethical decision making while the shame subscales diverged (Cohen et al., 2011). The rival proposal that the decisive criterion is harmfulness versus inadequacy (Miceli & Castelfranchi, 2018) holds that guilt is not necessarily circumscribed to one's behaviour and that the self-versus-behaviour view is questionable, and it awaits tests of the same predictive kind.

Third, the neuroscience of guilt has moved from localisation to connectivity and to pattern. The early contrasts named regions shared with theory of mind (Takahashi et al., 2004) and a guilt-specific orbitofrontal region (Wagner et al., 2011); the depression work found that what distinguishes vulnerability is not average activity in the anterior temporal lobe or the subgenual cingulate but the coupling between them, selective for guilt (Green et al., 2012); and the signature work found a distributed pattern that generalised across a Chinese and a Swiss sample but did not respond to recalled guilt (Yu et al., 2020). The last result is a caution for the whole literature: the guilt evoked by harming someone in the scanner and the guilt evoked by remembering a past wrong may be different enough that a single neural account of the emotion is premature.

Current Directions

The research front is clinical and methodological. In depression, the finding that self-blaming emotions were highly frequent but not restricted to guilt, with self-disgust or contempt more frequent than guilt, has led to a call for a refined assessment of self-blaming emotions to improve the diagnosis and stratification of major depressive disorder (Zahn et al., 2015), and the guilt-selective decoupling of the anterior temporal lobe from the subgenual cingulate is described as the first neural mechanism of depressive vulnerability that accounts for self-blaming biases (Green et al., 2012). In the neuroscience of social emotion, the guilt-related brain signature is presented as a step toward developing biological markers of social emotions that could serve as tools to investigate guilt-related brain processes in both healthy and clinical populations (Yu et al., 2020). In measurement, the meta-analytic demonstration that the relation between guilt and pro-social orientation depends largely on the methodological choices of the researcher (Tignor & Colvin, 2017) and the finding that the Guilt and Shame Proneness scale separates negative self-evaluation from withdrawal within shame (Cohen et al., 2011) set the agenda for the next generation of instruments. In theory, the proposal to redraw the shame-guilt boundary around inadequacy versus harmfulness and around discrepancy from the ideal self versus responsibility for one's fault (Miceli & Castelfranchi, 2018) is the live alternative to the behaviour-versus-self account. And in epidemiology, the observation that a substantial part of the German adult population is confronted with guilt feelings has prompted a call to identify the specific circumstances under which feelings of guilt lead to adverse health-related outcomes (Luck & Luck-Sikorski, 2021).

Common Misconceptions

“Guilt and shame are two words for the same feeling.”

Phenomenological ratings of personal experiences demonstrated that shame, guilt and embarrassment are not merely different terms for the same affective experience, and that the differences could not be explained simply by intensity of affect or by degree of moral transgression (Tangney et al., 1996). The two also predict differently: guilt increased cooperation among proselfs and shame did not (de Hooge et al., 2007).

“Shame is felt in public; guilt is felt in private.”

Contrary to popular belief, shame was no more likely than guilt to be experienced in public situations; all three self-conscious emotions typically occurred in social contexts, but a significant proportion of shame and guilt events occurred when respondents were alone (Tangney et al., 1996).

“Guilt is bad for mental health.”

Guilt-proneness was only moderately related to psychopathology, with correlations ascribable entirely to the shared variance between shame and guilt (Tangney et al., 1992), and in meta-analysis shame showed significantly stronger associations with depressive symptoms than guilt did; only contextual-maladaptive and generalised guilt matched shame (Kim et al., 2011).

“Guilt always leads to making amends.”

Guilt motivates compensatory pro-social behaviour to repair social bonds, but when opportunities for compensation are not present, guilt may evoke self-punishment, shown as self-denied pleasure and self-enforced penalties (Nelissen & Zeelenberg, 2009).

Glossary

Anterior middle cingulate cortex.

A medial frontal region whose activation and grey matter volume tracked interpersonal guilt and compensation when a participant's error caused pain to a partner.

Anterior temporal lobe.

A region holding representations of social meaning whose coupling with the subgenual cingulate is reduced, selectively for guilt, in remitted major depression.

Communal relationship.

A relationship characterised by expectations of mutual concern, the context in which guilt patterns are strongest, most common and most consistent.

Contextual-maladaptive guilt.

Guilt involving exaggerated or inappropriate responsibility for events, including uncontrollable ones; as strongly associated with depressive and anxiety symptoms as shame.

Dobby Effect.

The tendency for guilt to evoke self-punishment when opportunities for compensation are not present.

Embarrassment.

The third negatively valenced self-conscious emotion, a relatively distant neighbour of shame and guilt in phenomenological ratings.

External shame.

Negative views of the self as seen through the eyes of others; more strongly associated with depressive and social anxiety symptoms than internal shame.

Externalisation of blame.

The defensive attribution of fault to others, through which shame proneness predicted recidivism.

Generalised guilt.

Free-floating guilt divorced from any specific context; a maladaptive variant that behaves statistically like shame.

Guilt and Shame Proneness scale (GASP).

A measure with two guilt subscales (negative behaviour-evaluation, repair) and two shame subscales (negative self-evaluation, withdrawal) across a range of personal transgressions.

Guilt-proneness.

The dispositional tendency to experience guilt across transgressions; predicts lower reoffence, higher pro-social orientation and less unethical decision making.

Guilt-related brain signature.

A distributed multivariate pattern of brain activity that discriminates interpersonal guilt from matched control conditions across independent samples.

Interpersonal guilt.

The bad feeling that follows inflicting harm upon another person; the form of guilt evoked by interactive tasks and the object of the interpersonal approach.

Moral emotions.

Emotions that influence the link between moral standards and moral behaviour, including shame, guilt, embarrassment and the positive emotions of elevation, gratitude and pride.

Power-assertive discipline.

Parental control by force or threat; children of mothers who relied on it displayed less guilt.

Self-blame.

The global attribution of failure to oneself that cognitive models place at the root of depressive vulnerability; guilt is one of its emotions, alongside self-disgust, contempt and shame.

Self-conscious emotions.

Emotions such as shame, guilt, embarrassment and pride that require a representation of the self and are not captured by models of basic emotions.

Shame.

The self-conscious emotion that follows a negative evaluation of the whole self and motivates withdrawal; more strongly associated with depression and anxiety than guilt.

Subgenual cingulate cortex.

A ventromedial frontal region, with the adjacent septal area, that couples with the anterior temporal lobe during guilt and decouples from it in remitted depression.

Key Researchers

Roy F. Baumeister

. Emeritus Professor in the School of Psychology at the University of Queensland, formerly at Florida State University and Case Western Reserve University; proposed the interpersonal approach to guilt, in which guilt arises within communal relationships and serves to repair them. ORCID - Wikipedia - Faculty Page

Grazyna Kochanska

. Professor in the Department of Psychological and Brain Sciences at the University of Iowa; traced the early development of guilt in longitudinal studies of toddlers, linking children's guilt displays to fearful temperament, parental discipline and later rule-compatible conduct. ORCID - Wikipedia - Google Scholar

Helen Block Lewis

(1913-1987). Psychoanalyst and clinical psychologist who taught at Yale University; drew the modern distinction between shame as a global negative evaluation of the self and guilt as a negative evaluation of a specific behaviour in her 1971 book Shame and Guilt in Neurosis, the proposal against which the later empirical literature was tested. Wikipedia

June Price Tangney

. Distinguished University Professor of Psychology at George Mason University, retired in 2025; established the empirical distinction between shame-proneness and guilt-proneness with the Test of Self-Conscious Affect and showed their divergent relations to psychopathology, moral behaviour and recidivism. Faculty Page

Jessica L. Tracy

. Professor of Psychology at the University of British Columbia and director of the Emotion and Self Lab; co-authored the theoretical model of self-conscious emotions that distinguishes them from basic emotions and specifies the appraisals through which guilt, shame and pride arise. Faculty Page

Hongbo Yu

. Faculty member in the Department of Psychological and Brain Sciences at the University of California, Santa Barbara (Yu Emotion Science Lab); mapped the neural bases of interpersonal guilt and compensation and led the derivation of a generalisable multivariate brain signature for interpersonal guilt. ORCID - Faculty Page

Roland Zahn

. Clinical Professor of Mood Disorders and Cognitive Neuroscience at the Institute of Psychiatry, Psychology and Neuroscience, King's College London; identified the guilt-selective decoupling of the anterior temporal lobe from the subgenual cingulate in remitted depression and characterised the self-blaming emotions of the depressive syndrome. ORCID - Faculty Page

Marcel Zeelenberg

. Professor of Economic Psychology at Tilburg University and Professor of Behavioral Research in Marketing at Vrije Universiteit Amsterdam; showed that guilt and shame have distinct effects on cooperation and that guilt evokes self-punishment when compensation is unavailable. ORCID - Faculty Page

Frequently Asked Questions

What is guilt in psychology?

Guilt is a negatively valenced self-conscious moral emotion; the definitions that researchers use share a focus on one's action or inactions involving real or imagined moral transgressions, and divide over whether guilt is an inherently adaptive construct that generates remorse, augments a sense of responsibility and motivates reparation (Tilghman-Osborne et al., 2010; Tangney et al., 2007).

What is the difference between guilt and shame?

In the dominant account, guilt involves negative evaluations of a behaviour and a tendency to repair, whereas shame involves negative evaluations of the self and a tendency to withdraw (Cohen et al., 2011); the two are not merely different terms for the same affective experience, and shame is no more public than guilt (Tangney et al., 1996).

Is guilt a healthy emotion?

Guilt-proneness was only moderately related to psychopathology, with the correlations ascribable entirely to its shared variance with shame (Tangney et al., 1992); in meta-analysis guilt was positively correlated with pro-social orientation (Tignor & Colvin, 2017), but generalised and contextual-maladaptive guilt were as strongly tied to depressive symptoms as shame (Kim et al., 2011).

Why does guilt make people want to make amends?

Guilt appears to arise from interpersonal transactions and is strongest in communal relationships, where it motivates people to treat partners well, minimises inequities and redistributes emotional distress (Baumeister et al., 1994); when compensation is not possible, the same motive can turn into self-punishment (Nelissen & Zeelenberg, 2009).

Does guilt stop people from reoffending?

In a longitudinal study of 476 jail inmates, guilt proneness negatively and directly predicted reoffence in the first year after release, whereas shame proneness did not and predicted recidivism via externalisation of blame (Tangney et al., 2014).

When do children start to feel guilt?

Behavioural and affective components of guilt were observed in children led to believe they had damaged valuable objects at 22, 33 and 45 months; they cohered across sessions, were stable across assessments, and children who displayed more guilt were less likely to violate rules of conduct at 56 months (Kochanska et al., 2002).

Which brain regions are involved in guilt?

Guilt has been associated with medial prefrontal and posterior superior temporal regions shared with embarrassment (Takahashi et al., 2004), a guilt-specific region of right orbitofrontal cortex (Wagner et al., 2011), the anterior middle cingulate and anterior insula during interpersonal guilt (Yu et al., 2014), and, across 21 studies, the ventral anterior cingulate, posterior temporal regions and precuneus (Bastin et al., 2016).

How is guilt related to depression?

Current guilt feelings were reported by 37.4% of adults with major depression against 8.1% without (Luck & Luck-Sikorski, 2021); shame nonetheless shows stronger associations with depressive symptoms than guilt (Kim et al., 2011), and remitted depression is marked by a guilt-selective decoupling of the anterior temporal lobe from the subgenual cingulate (Green et al., 2012).

References