Guilt and shame have long been a focal point of research across various disciplines, including psychology, behavioral economics, neuroscience, and psychiatry (e.g. Chang et al., 2011; Gao et al., 2018; Landers et al., 2024b; Mottershead et al., 2024; Schuster et al., 2021; Sznycer, 2019; Tangney et al., 2007). Extensive research has investigated the psychological and neural activities associated with the experience of guilt and shame to enhance emotion regulation and improve behavioral prediction (Bastin et al., 2016; Miceli and Castelfranchi, 2018; Michl et al., 2014; Piretti et al., 2023; Pulcu et al., 2014; Takahashi et al., 2004; Tangney and Dearing, 2003; Xu et al., 2022; Zhu et al., 2019a). We extended this research by shifting the focus from the mere states of guilt and shame to their associations with the cognitive antecedents and behavioral consequences. Our findings advance the understanding of the psychological and neural mechanisms that underlie both the formation of guilt and shame and their subsequent transformation into compensatory behavior.

Consistent with previous studies (Gao et al., 2021; Iyer et al., 2007; Koban et al., 2013; Li et al., 2020; Tangney, 1992; Yu et al., 2014), we observed that both harm and responsibility increase individuals’ feelings of guilt and shame. Of importance, for the first time, we discovered that harm exerts a stronger effect on guilt than on shame, whereas responsibility has a stronger effect on shame than on guilt. These findings provide empirical support for the proposition that guilt and shame serve distinct social functions. According to functionalist theories (e.g. Baumeister et al., 1994; Gilbert, 1997; Sznycer, 2019; Sznycer et al., 2021), guilt functions to curb the harm imposed on valued others, whereas shame functions to mitigate the cost of reputational damage to oneself. If these theories hold true, guilt should be more sensitive than shame to the information related to harm inflicted on valued others, where shame should be more sensitive than guilt to the information related to reputational damage to oneself (Sznycer et al., 2016). However, few studies have provided such direct evidence. An exception is the study by Landers et al., 2024b, which found that information related to harm inflicted on valued others (e.g., liking of a victim) and concerns about reputational damage (e.g. fear of a victim) were respectively predictive of guilt and shame. Notably, Landers et al., 2024b employed a vignette-based paradigm and assessed participants’ guilt and shame using items that reflected the motivational tendencies characteristic of these emotions (e.g. guilt: ‘I would go to him and apologize for it’; shame: ‘I would feel like avoiding him for a while’). In contrast, our study utilized a laboratory-based paradigm with real-time decision-making, directly measured participants’ state guilt and shame experiences, and tested new psychological factors (harm and responsibility). This approach offers greater ecological validity and provides novel evidence (Yu et al., 2024).

Existing findings suggest that guilt is more strongly linked to altruistic behavior than shame is (de Hooge et al., 2008; de Hooge et al., 2007; Declerck et al., 2014; Gausel and Leach, 2011; Graton and Ric, 2017; Ketelaar and Tung Au, 2003; Tangney et al., 2007; Tangney and Dearing, 2003). Nevertheless, because of methodological limitations, such as failing to compare guilt and shame directly or employing methods that may introduce confounding variables, conclusive evidence has been lacking. Overcoming these limitations, our study demonstrates that while both guilt and shame promote compensation, guilt is more effective in prompting compensatory behavior. The functionalist theories offer a framework for understanding why guilt exerts a stronger effect on compensation than shame does. Guilt corresponds to the adaptive problem of insufficiently valuing others (Sznycer, 2019). To address it, individuals in guilt must bring benefits to those they have harmed—typically through altruistic behavior—to correct the inequity caused by their wrongdoing. In contrast, shame is tied to reputational damage (Sznycer, 2019; Sznycer et al., 2021). Although altruistic behavior can also mitigate shame by demonstrating their social value, individuals may resort to other strategies—such as avoidance or aggression—to protect themselves from potential devaluation by others (de Hooge et al., 2018; Zhu et al., 2019c).

For a deeper understanding of social emotions, it is crucial to formally model their cognitive operations and investigate the neural underpinnings of these cognitive operations (Yu et al., 2024). Building on this research line, our computational modeling results reveal that individuals in guilt and shame disregard their self-interest, adopt a compensatory baseline, and mentally distribute harm across all wrongdoers. The findings not only offer a mechanistic explanation at the behavioral (algorithmic) level for guilt- and shame-driven compensatory decision-making, but also deepen the understanding of the phenomenon of responsibility diffusion by offering a formal mathematical formulation and linking it to compensatory behavior (Darley and Latané, 1968).

Notably, in many computational models of social decision-making, self-interest plays a crucial role (e.g. Wu et al., 2024). However, our computational findings suggest that participants disregarded self-interest during compensatory decision-making. A possible explanation is that the personal stakes in our task were relatively small (a maximum loss of 5 Chinese yuan), whereas the harm inflicted on the receiver was highly stigmatized (i.e. an electric shock). Under conditions where the harm is highly salient and the cost of compensation is low, participants may be inclined to disregard self-interest and focus solely on making appropriate compensation.

At the neural level, our findings demonstrate the involvement of the posterior insula and striatum in representing the cognitive antecedents of guilt and shame. Specifically, the activation in these brain regions decreased as the quotient of harm divided by the number of wrongdoers increased. Harm inflicted on the victim, particularly the portion for which a wrongdoer is responsible, creates a sense of inequity between them. Beyond its well-established role in interoceptive awareness (Craig, 2009; Craig, 2011), the pINS has been implicated in processing economic inequity in allocation tasks (Gao et al., 2018; Hsu et al., 2008). For instance, Hsu et al., 2008 reported that pINS activation negatively correlates with the degree of inequity, suggesting that greater inequity elicits lower pINS activation. Our results extend this role of the pINS beyond economic inequity to encompass harm inequity. Given that the striatum is implicated in value computation (Bartra et al., 2013; Crockett et al., 2017; Rilling et al., 2008), we believe that its activity reflects individuals’ perception of the loss (i.e. harm) inflicted on the victim. In contrast, no brain region had significant responses to the product of harm and responsibility. Thus, the fMRI and computational modeling findings offer convergent evidence indicating that individuals are more likely to integrate these cognitive antecedents in a form of quotient.

In addition, no brain region exhibited significant responses to harm. Only the sensorimotor areas showed significant responses to the number of wrongdoers (i.e. the complement of responsibility, 5 – responsibility level). Although the fMRI findings revealed that no brain region associated with social cognition showed significant responses to harm or responsibility, this does not suggest that the human brain encodes only a unified measure integrating harm and responsibility and does not process them as separate entities. Using more fine-grained techniques, such as intracranial electrophysiological recordings, it may still be possible to observe independent neural representations of harm and responsibility.

As to emotion sensitivity, our findings show that individuals who tend to convert responsibility into shame exhibit reduced activation in brain regions associated with other-oriented theory-of-mind processing, specifically the TPJ and STS. The TPJ and STS have been implicated in inferring others’ mental states (Schurz et al., 2014). Lower activation in these regions indicates that individuals with higher responsibility-driven shame sensitivity may be less engaged in considering the victim’s experiences and thoughts. This aligns with existing research on shame, which, compared to guilt, is associated with less concerns on one’s actions on others (Tangney and Dearing, 2003) and weaker activation in the TPJ (Zhu et al., 2019a).

Regarding compensatory sensitivity, our results show that both individuals with higher guilt-driven and shame-driven compensatory sensitivity have stronger activation in the DMPFC. This region is central to both theory-of-mind processing (Schurz et al., 2014) and self-referential processing (Northoff et al., 2006), playing a crucial role in combining others’ thoughts and feelings with one’s own (D’Argembeau et al., 2007; Saxe et al., 2006). Zhu et al., 2019a have identified the DMPFC’s involvement in the experience of both guilt and shame. Our findings here further highlight its role in translating these emotions into compensatory behavior.

We found that the TP’s activity is positively related to individuals’ guilt-driven compensatory sensitivity. This region is considered a core part of the theory-of-mind network (Frith and Frith, 2003). Numerous studies suggest that the activation in this region reflects retrieval of both general conceptual knowledge (Lambon Ralph and Patterson, 2008) and social conceptual knowledge (e.g. social rules; Ross and Olson, 2010; Sugiura et al., 2006; Tsukiura et al., 2010; Zahn et al., 2007). The retrieval of such information likely facilitates understanding others’ thoughts and empathizing with their suffering (Olson et al., 2007; Schurz et al., 2014). Empathy, in turn, has been widely established as a significant driver of altruistic behavior, including compensatory behavior (Ding and Lu, 2016 ; Eisenberg and Miller, 1987). Our findings confirm the involvement of the TP in translating guilt into compensation.

The IPL’s activity has a positive correlation with individuals’ shame-driven compensatory sensitivity. This region is associated with various non-social and social cognitive functions, including number processing (Pinel et al., 2004; Sandrini et al., 2004), salience processing (Kahnt et al., 2014), and theory-of-mind processing (Igelström and Graziano, 2017; Tusche et al., 2016). Two recent studies provided direct evidence showing that the IPL plays a role in encoding others’ benefits during altruistic decision making (Hu et al., 2017; Hu et al., 2021). Our findings about IPL can be explained by its involvement in generating other-regarding motives (Hu et al., 2021) that facilitate the conversion from shame to compensation.

We did not find a significant difference in the correlations between TP activity and guilt-driven versus shame-driven compensatory sensitivities. Similarly, no significant difference was observed in the correlations between IPL activity and shame-driven versus guilt-driven compensatory sensitivities. These findings suggest that neither of these regions plays a domain-specific role in compensation driven by guilt or shame.

In contrast, LPFC activity exhibited a stronger correlation with shame-driven compensatory sensitivity than with guilt-driven compensatory sensitivity, indicating a domain-specific role of the LPFC in shame-related compensation. The LPFC is implicated in cognitive control (Koechlin et al., 2003) and the optimization of social decision-making (Buckholtz and Marois, 2012; Feng et al., 2015). Some brain stimulation studies have demonstrated that disrupting LPFC activity impairs individuals’ ability to inhibit selfish or aggressive impulses, which can incur social devaluation and punishment from others (Knoch et al., 2006; Knoch et al., 2009; Riva et al., 2015). Further research has extended these findings by emphasizing the LPFC’s role in strategic social behavior (Ruff et al., 2013; Strang et al., 2015). For instance, Ruff et al., 2013 found that when individuals face the possibility of being punished for selfish behavior, enhancing LPFC activity suppresses selfish impulses and promotes altruistic behavior. However, in the absence of punishment risk, enhancing LPFC activity instead reduces altruistic behavior and promotes self-interest. Considering that guilt is typically alleviated through altruistic behavior (Tang et al., 2019), whereas coping strategies for shame are more varied—ranging from altruistic behavior to aggression and avoidance (Sznycer, 2019; Sznycer et al., 2016)—shame appears to be more closely linked to strategic thinking than guilt. This explains why LPFC activity, which is associated with strategic behavior, is more strongly related to shame-driven compensatory sensitivity than to guilt-driven compensatory sensitivity.

The neural correlates of the parameter κ largely overlapped with those linked to compensatory sensitivities driven by guilt and shame. Intriguingly, beyond that, κ also showed a strong association with aINS activity. Insula, known as a key node in the salience network (Uddin, 2015), engages in during experiencing various negative emotions, including sadness (Wagner et al., 2011), disgust (Craig, 2009), guilt (Yu et al., 2014), shame (Piretti et al., 2023; Zhu et al., 2019a), and indebtedness (Gao et al., 2024). Social neuroscience research highlights the critical role of the aINS in processing norm violations and guiding behavior accordingly (Bellucci et al., 2018; Zinchenko and Arsalidou, 2018). For example, Chang et al., 2011 found that aINS serves to mitigate anticipated negative feelings triggered by norm violations by facilitating individuals’ reciprocity toward their partners’ investments in a trust game, thereby maintaining adherence to the norm of reciprocity. Consistently, numerous studies on the ultimatum game reveal aINS’s involvement in rejecting unfair offers and upholding the norm of fairness (Feng et al., 2015; Gabay et al., 2014). In the same line, the findings on the involvement of aINS in social conformity also manifest its role in monitoring norm violations and reinforcing adherence to social norms (Berns et al., 2010; Klucharev et al., 2009). Given the involvement of the aINS in various social-affective processes, our findings suggest that the motivation to uphold social norms might directly shape individuals’ compensatory behavior or indirectly influence it through emotions beyond guilt and shame, with aINS activity playing a pivotal role in this process.

Interestingly, we found that the sensorimotor areas were associated with the representation of a shame-related cognitive antecedent (i.e. responsibility) and emotional sensitivity. Our findings align with the result from a fMRI meta-analysis, which identified the involvement of sensorimotor regions in processing shame (Piretti et al., 2023). It has been suggested that sensorimotor activation may reflect typical shame-related action tendencies, such as reduced social presence, speech, and movement (Piretti et al., 2023). However, in our study, participants were required to remain completely still throughout the experiment to maintain MRI data quality and were continuously monitored, eliminating the possibility of physical withdrawal. Therefore, the observed sensorimotor activation may reflect motor preparation for subsequent button-press responses associated with compensation rather than a general tendency toward shame-related avoidance. Future studies that permit participants to engage in actual avoidance behaviors could further clarify the role of sensorimotor areas in shame processing.

In line with previous research (Cohen et al., 2011), our findings reveal that both dimensions of trait guilt were significantly associated with compensatory behavior, whereas neither dimension of trait shame exhibited such an association. Furthermore, we found neural responses in the aMCC mediated the relationship between repair action tendencies (one dimension of trait guilt) and compensation. A substantial body of research has revealed that guilt processing consistently activates the aMCC (see a meta-analysis, Gifuni et al., 2017). It is suggested that, in the context of guilt, the aMCC plays a role in detecting the conflict between social norms and actual behavior and signaling this conflict via generating negative emotions (Bastin et al., 2016; Gifuni et al., 2017). In addition, Yu et al., 2014 linked the aMCC activity with compensatory behavior. Accordingly, our fMRI findings suggest that individuals with stronger tendency to engage in compensation across various moral violation scenarios (indicated by their repair action tendencies) are more sensitive to the severity of the violation and therefore engage in greater compensatory behavior. However, the neural correlates of negative behavior evaluations (another dimension of trait guilt) were absent. The reasons underlying the non-significant neural finding may be multifaceted. One possibility is that negative behavior evaluations influence neural responses indirectly through intermediate processes not captured in our study (e.g. specific motivational states).

Although previous research has found that trait gratitude and SVO are significantly associated with altruistic behavior in contexts such as donation (Van Lange et al., 2007; Yost-Dubrow and Dunham, 2018) and reciprocity (Ma et al., 2017; Yost-Dubrow and Dunham, 2018), their associations with compensatory decisions in the present study were not significant. This suggests that the effects of trait gratitude and SVO on altruistic behavior are context-dependent and may not predict all forms of altruistic behavior.

This research has several limitations. First, post-task assessments of guilt and shame, unlike in-task assessments, rely on memory and may thus be less precise, although in-task assessments could have heightened participants’ awareness of these emotions and biased their decisions. Second, our measures of guilt and shame depend on participants’ conceptual understanding of the two emotions. While this is common practice in studies with adult participants (Michl et al., 2014; Wagner et al., 2011; Zhu et al., 2019a), it may be less appropriate for research involving children. Third, although we aimed to construct a conceptually comprehensive computational model space informed by prior research and our own understanding, it does not encompass all plausible models. Future research is encouraged to explore additional possibilities. Fourth, fMRI cannot establish causality. Future studies using brain stimulation techniques (e.g. transcranial magnetic stimulation) are needed to clarify the causal role of brain regions in guilt-driven and shame-driven altruistic behavior. Fifth, we did not explicitly measure emotions similar to guilt and shame (e.g. indebtedness), which would have been helpful for understanding their distinct contributions. Sixth, marginally significant results should be viewed cautiously and warrant further examination in future studies with larger sample sizes.

Our study makes several innovative contributions. First, we developed a novel paradigm that effectively elicits guilt and shame at comparable intensities, enabling researchers to systematically explore the associations among guilt, shame, their cognitive antecedents, and behavioral consequences. Future research could combine this paradigm with other cognitive neuroscience methods, such as electroencephalography (EEG) or magnetoencephalography (MEG), and adapt it to investigate additional behaviors linked to guilt and shame, including donation (Xu, 2022), avoidance (Shen et al., 2023), and aggression (Velotti et al., 2014). Second, our behavioral findings provide high-quality empirical evidence for functionalist theory, aligning with the contemporary trend of comprehending emotions through their adaptive functions (Landers et al., 2024b; Sznycer et al., 2021). Third, our computational and neural findings offer a clear delineation of the neurocognitive mechanisms underlying guilt and shame. Building on knowledge that harm and responsibility are related to guilt and shame, our results further reveal how these cognitive antecedents are integrated. While previous studies have broadly identified brain regions associated with guilt and shame processing as a whole (Bastin et al., 2016; Gifuni et al., 2017; Piretti et al., 2023), our study advances this understanding by breaking down guilt and shame processing into distinct processes and precisely mapping the neural correlates of each process.

Our study has potential practical implications. The behavioral findings may help counselors understand how cognitive interventions targeting perceptions of harm and responsibility could influence experiences of guilt and shame. The neural findings highlight specific brain regions (e.g. TPJ) as potential intervention targets for regulating these emotions. Given the close links between guilt, shame, and various psychological disorders (e.g. Kim et al., 2011; Lee et al., 2001; Schuster et al., 2021), strategies to regulate these emotions may contribute to symptom alleviation. Nevertheless, because this study was conducted with healthy adults, caution is warranted when considering applications to other populations.

In conclusion, our findings support the functionalist theory by demonstrating distinct effects of harm and responsibility on guilt and shame, as well as differences in the efficiency with which guilt and shame translate into compensatory behaviors. Notably, harm and responsibility are integrated in a manner consistent with responsibility diffusion prior to influencing guilt-driven and shame-driven compensation. Furthermore, the distinct stages involved in guilt and shame processing correspond to activities in specific neural regions related to value computation, salience processing, theory-of-mind processing, self-referential processing, and cognitive control. By simultaneously providing computational, algorithmic, and neural accounts of guilt and shame (Yu et al., 2024), our study advances the holistic understanding of these emotions, which provides insights into how guilt and shame can be regulated and informs the treatment of guilt- and shame-related mental disorders.