Do Bad People Deserve Empathy? Selective Empathy Based on Targets’ Moral Characteristics

Abstract

The relation between empathy and morality is a widely discussed topic. However, previous discussions mainly focused on whether and how empathy influences moral cognition and moral behaviors, with limited attention to the reverse influence of morality on empathy. This review summarized how morality influences empathy by drawing together a number of hitherto scattered studies illustrating the influence of targets’ moral characteristics on empathy. To explain why empathy is morally selective, we discuss its ultimate cause, to increase survival rates, and five proximate causes based on similarity, affective bonds, the appraisal of deservingness, dehumanization, and potential group membership. To explain how empathy becomes morally selective, we consider three different pathways (automatic, regulative, and mixed) based on previous findings. Finally, we discuss future directions, including the reverse influence of selective empathy on moral cognition, the moral selectivity of positive empathy, and the role of selective empathy in selective helping and third-party punishment.

Imagine the following scenarios and think about how you would feel: a basketball player, famous for his flagrant fouls, accidentally sprains his ankle in a game; a selfish co-worker, who always takes advantage of others’ contributions, is getting a divorce; a prisoner who has sexually assaulted dozens of children is suffering from an incurable disease. When observing others suffer misfortune, people usually show empathy toward them. However, if immoral people suffer—as in the above scenarios—do people still show empathy?

Empathy is usually perceived as a moral virtue (e.g., Goldstein et al., 2014; Nussbaum, 1996), and equal empathy for all people (regardless of whether the target is socially distant or close) is attributed greater moral value than selective empathy (Bloom, 2017; Fowler et al., 2021). However, in the above scenarios, both intuition and rationality might lead us to show less empathy for immoral people. In other words, it is widely accepted that empathy is selective depending on targets’ moral characteristics (Cameron et al., 2022; Wang et al., 2019). Indeed, a recent study found that showing empathy toward a negative target (e.g., a White supremacist) would bring the empathizer less respect and liking, compared with not showing empathy (Wang & Todd, 2021).

Although such selective empathy is ubiquitous in daily life, there is a lack of comprehensive research and no theoretical model to explain this phenomenon. In this paper, we provide, for the first time, a comprehensive overview of selective empathy based on targets’ moral characteristics. We draw together a number of hitherto scattered studies illustrating this phenomenon and explain why and how empathy becomes selective in response to the moral characteristics of targets. Previous discussion has mainly focused on the way that empathy influences moral cognition and moral behavior (for reviews, see Decety & Cowell, 2014; Eisenberg, 2000; Eisenberg & Miller, 1987; Hoffman, 1987, 2001). By contrast, this review focuses on the influence of moral cognition on empathy and points to new directions for research.

Before discussing whether individuals show varied empathic responses toward moral and immoral individuals, we need to clarify what counts as moral vs. immoral behavior. Although variation within and across cultures makes it hard to provide a completely consistent definition, there is a consensus about what acts (and agents) are considered moral or immoral (Atari et al., 2022; Brambilla & Leach, 2014; Brambilla et al., 2011; Curry et al., 2019; Fiske et al., 2007; Graham et al., 2011, 2013; Haidt & Graham, 2007; Haidt & Joseph, 2004; Schein & Gray, 2018), and these societal values emerge early in life (Decety & Cowell, 2018; Hamlin et al., 2007; Surian et al., 2018; Ting & Baillargeon, 2021).

What Is Empathy?

Empathy refers to the process of sharing and understanding the feelings of others (Cohen & Strayer, 1996; Decety & Jackson, 2004; Decety & Lamm, 2006; Singer & Lamm, 2009). It is generally seen as an important, moral emotion (Decety & Cowell, 2014; Eisenberg, 2000; Eisenberg & Miller, 1987; Hoffman, 2001; Hume, 1896; Slote, 2007; Smith, 2010). With increasing research on empathy, the concept has been variously defined by different researchers (for a review, see Cuff et al., 2016). Given the complexity of these definitions, more and more researchers recommend a focus on specific sub-concepts or components when discussing empathy, rather than relying on the word empathy in general (Decety & Cowell, 2014; Hall & Schwartz, 2019; Weisz & Cikara, 2021). Here, we focus on three components of empathy in discussing responses based on the moral characteristics of targets.

The emotional component of empathy refers to the process of sharing another person’s emotion, also called emotional contagion or emotional matching (e.g., feeling sad when seeing others feeling sad or feeling happy when seeing others feeling happy; Hatfield et al., 1994). The cognitive component of empathy refers to the understanding of others’ emotions, also called affective perspective-taking (Decety & Cowell, 2014; Decety & Jackson, 2004). The motivational component of empathy refers to concern about sufferers, including the motivation to alleviate their pain (de Waal, 2008; de Waal & Preston, 2017)—often called empathic concern or compassion or sympathy by some researchers (Batson et al., 1983; Davis, 1983; Eisenberg, 2000; Goetz et al., 2010; Hoffman, 2001; Klimecki, 2019; for the claim that compassion and sympathy are different, see Dutton et al., 2006). When seeing others suffer, individuals usually exhibit one or more of these three components of empathy (emotional sharing, affective perspective-taking, and empathic concern).

Yet people do not always show empathy toward another’s misfortune. Indeed, sometimes, people feel pleasure at another’s misfortune, defined as Schadenfreude, a form of counter-empathy (Lanzetta & Englis, 1989; Portmann, 2000; Takahashi et al., 2009). Counter-empathy can be viewed as the opposite of empathy. Research has shown that people often show counter-empathy toward competitors, superiors, or immoral individuals (Cikara et al., 2014; Lanzetta & Englis, 1989; Singer et al., 2006; Takahashi et al., 2009).

Evidence for Selective Empathy Based on Targets’ Moral Characteristics

Behavioral, physiological, and neural studies all demonstrate that empathy is selective, depending on targets’ moral characteristics. Behavioral studies have found that when watching an immoral person experience bad things, people feel less empathic than when watching a moral person experience the same things. By manipulating the description of characters’ moral traits (such as honesty, sincerity, and trustworthiness) or morally-related acts (such as helping or harming), researchers have found that individuals report less compassion or more pleasure when learning that immoral others experience a misfortune as opposed to learning that moral others experience a misfortune (Brambilla & Riva, 2017; Rodriguez-Gomez et al., 2020). Individuals perceive the social pain of immoral others as less painful than that of moral or neutral others (Riva et al., 2016). Furthermore, individuals feel happier at the punishment of a person who has committed multiple immoral acts than at the punishment of a person who has committed a single immoral act, and individuals’ evaluation of the person’s moral characters plays a mediating role in this process (Berndsen & Tiggemann, 2020). These findings confirm that others’ moral characteristics can modulate adults’ empathy, and that the influence of others’ morally related acts on empathy is mediated by the evaluation of their moral character. Additionally, studies of “moral circles” have found that individuals often place “villains” outside the scope of moral concern (Crimston et al., 2016, 2018), in line with the above findings that individuals show reduced empathy or even counter-empathy toward immoral people.

However, it should be acknowledged that social desirability can also influence people’s self-reported feelings. Under the pressure of the norm of not showing empathy toward negative targets (Wang & Todd, 2021), people may actually feel equally empathic, but report feeling less empathic toward immoral as compared to moral others. Accordingly, self-report in combination with physiological recordings or brain imaging can provide more convincing evidence for whether individuals show varied empathy toward moral and immoral others.

Stellar et al. (2014) compared people’s emotional and physiological responses when they learned of the misfortune of a moral or immoral target. The targets’ moral behaviors were manipulated by giving the participants messages describing the targets’ selfish or cooperative behaviors. Then, participants watched a video clip of the target talking about his or her misfortunes, and participants’ physiological responses were recorded as they watched. Participants reported feeling less compassionate and showed physiological responses indicating less compassion (greater heart rate and reduced respiratory sinus arrhythmia activity, Stellar et al., 2015) toward the selfish target than toward the cooperative target.

Event-related potential (ERP) studies were also conducted to compare individuals’ empathy toward moral and immoral others. Cui et al. (2016) compared individuals’ brain responses toward painful or non-painful pictures (i.e., bodies with wounded or non-wounded parts, such as a finger cut or not cut by scissors) of blood donors, killers, or unidentified targets. Painful pictures elicited larger amplitude of N2 than non-painful pictures when the target was a blood donor or was unidentified. However, this difference disappeared when the target was a killer. The difference in N2 to painful vs. non-painful pictures was localized in the ventral medial prefrontal cortex (vmPFC) and the rostral anterior cingulate cortex (rACC) areas. Given that N2 is believed to reflect individuals’ emotional arousal (Fan & Han, 2008), these findings imply that individuals showed less emotional sharing toward immoral others than to moral others.

Li et al. (2018) compared individuals’ brain responses when watching painful and non-painful pictures of different targets. Unlike the study by Cui et al. (2016), participants were asked to imagine the body part in the picture as that of a moral or immoral person. Although N2 responses did not differ for moral as compared to immoral targets, the responses of P3 did differ, with a larger amplitude for painful pictures than nonpainful pictures when the target was imagined as moral but no amplitude difference when the target was imagined as immoral. Given that P3 is believed to reflect attention redistribution driven by the top-down regulation of empathy (Cheng et al., 2012, 2014; Coll, 2018; Fan & Han, 2008), these findings suggest that targets’ moral characters influence the top-down regulation of empathy. More specifically, participants may reduce their attention to painful stimuli when imagining the target as an immoral person (MacNamara et al., 2009).

There are two plausible explanations for the contrasting findings of Cui et al. (2016) and Li et al. (2018). First, whether the identity of the target is specified or imagined broadly can influence the modulation of empathy. In Cui et al. (2016), the identity of the target was specified as a blood donor or a killer, leading to more direct selectivity at the early stage of empathy. In Li et al. (2018), participants needed to imagine the target as moral or immoral themselves, and each participant might have chosen different prototypes for moral or immoral others, leading to selectivity at a later stage of empathy. Second, the gravity of target immorality might influence the modulation of empathy. A killer is usually perceived as extremely immoral, whereas a general description of “immoral” can refer to various transgressions, including minor wrongdoings, such as taking advantage of others. More extreme immorality might lead to a more immediate selectivity of empathy during the early emotion arousal process without top-down modulation, as in Cui et al. (2016).

Additionally, a functional magnetic resonance imaging (fMRI) study demonstrated that males (but not females) showed less empathy toward an unfair confederate who had damaged their benefits as compared to a fair confederate who ensured their benefits (Singer et al., 2006). Males’ empathy-related activation in pain-related brain areas (fronto-insular and anterior cingulate cortices) was weaker, and their brain activation in reward-related areas (left ventral striatum and nucleus accumbens) was stronger, toward unfair as compared to fair confederates. The gender effect found in this study also suggested that males might be more likely to show reduced empathy toward immoral others than females, a possibility that warrants further examination.

Taken together, using self-report, physiological, and brain imaging measures, prior research has found that the three components of adult empathy are all influenced by targets’ moral characteristics. More specifically, individuals are less likely to share the feelings of immoral as compared to moral targets (Berndsen & Tiggemann, 2020; Brambilla & Riva, 2017; Rodriguez-Gomez et al., 2020), tend to perceive the social pain of immoral targets as less painful than that of moral targets (Riva et al., 2016), and show less empathic concern for immoral as compared to moral targets (Berndsen & Tiggemann, 2020; Brambilla & Riva, 2017; Rodriguez-Gomez et al., 2020; Stellar et al., 2014). Furthermore, the ERP findings suggest that the influence of targets’ moral characteristics on empathy can be achieved either at the emotional arousal stage (Cui et al., 2016) or via the top-down regulation of empathy (Li et al., 2018), depending on the specific context.

The above studies were all conducted with adults. Studies on children have also found that their empathy is selective, depending on targets’ moral characteristics. When presented with picture-stories, 4- to 8-year-old children reported feeling more pleasant and less sorry for a protagonist’s misfortune when the protagonist intended to harm rather than to help others, and this effect was more pronounced among older children (Schindler et al., 2015; Schulz et al., 2013). Another study examined children’s facial expressions when watching the punishment of an antisocial puppet who tricked them and a prosocial puppet who gave toys to them (Mendes et al., 2018). Whereas 4- and 5-year-olds mainly produced frowns when watching either puppet being hit, six-year-olds produced more smiles with frowns when watching the antisocial puppet being hit than when watching the prosocial puppet being hit. By implication, 6-year-olds showed selective empathy based on targets’ moral characteristics.

Overall, we see that the evidence mainly supports the selectivity of empathy based on targets’ moral characteristics that are related to the violation of harm/care (e.g., Cui et al., 2016; Schindler et al., 2015; Schulz et al., 2013) and fairness/reciprocity principles (e.g., Berndsen & Tiggemann, 2020; Brambilla & Riva, 2017; Riva et al., 2016; Singer et al., 2006; Stellar et al., 2014). It remains unclear whether targets’ violation of moral values in other domains, such as ingroup/loyalty and purity/sanctity, can also influence empathy toward them. For example, it is unclear whether individuals will show reduced empathy toward a man who is fired for disclosing the secrets of his own company to competitors or a man who has a stomachache after eating his dead pet dog.

Justice in a Broad Sense

One important basis for the moral selectivity of empathy is the evaluation of how much the target deserves to suffer (Gibbs, 2019; Goetz et al., 2010; Hein & Singer, 2008; Simpson et al., 2014). Justice in a broad sense refers to the principle that people receive what they deserve (Lerner, 1980). More specifically, bad things happen to immoral people, and good things happen to moral people. Driven by such justice beliefs (Hafer & Rubel, 2015; Lerner, 1980; Vermunt, 2014), individuals tend to believe that immoral targets deserve to suffer, thereby showing reduced empathy (for more details on this process, see the deservingness-appraisal account in the next section). When the targets’ suffering is linked to their immoral behavior, for example, a murderer is sentenced to death for murder, it is widely accepted that the murderer’s suffering is what he or she deserves, and that people will show less empathy for the murderer. However, if the targets’ suffering is not linked to their immoral behaviors, for example, a murderer suffers from cancer, how far will people respond with empathy?

A recent study of way that empathizers are evaluated provides a preliminary answer. People showed less respect or liking toward empathizers who expressed empathy toward a negative target whose suffering was causally linked to the target’s negative behaviors (e.g., a White supremacist suffered from the stress of working in an organization peddling racist views), whereas people showed more respect or liking toward empathizers who expressed empathy toward a negative target whose suffering was not linked to the target’s negative behaviors (e.g., a White supremacist suffering from cancer; Wang & Todd, 2021). These findings demonstrate that the way in which third-party observers evaluate empathizers is influenced by the causal links between the targets’ negative characteristics and the targets’ suffering. However, things might be different when people are empathizers themselves.

Indeed, previous findings suggested that individuals show less empathy toward immoral targets than moral targets, regardless of whether the targets’ misfortune was directly brought about by their immoral behaviors (e.g., Berndsen & Tiggemann, 2020; Rodriguez-Gomez et al., 2020; Schindler et al., 2015; Schulz et al., 2013), or by accidents unrelated to their immoral behaviors (e.g., Brambilla & Riva, 2017; Stellar et al., 2014). Limited empathy for targets’ suffering, even when the suffering was unrelated to their immoral characteristics, might indicate a kind of broad justice belief, namely that bad people deserve bad outcomes anyhow. This kind of justice belief may not be rational, but it is widespread. However, given that previous studies did not directly compare the moral selectivity of empathy when targets’ morally related acts are related or unrelated to their misfortunes, future studies should explore this issue more systematically.

Person-Based or Act-Based Selectivity

Previous studies suggest that the selectivity of empathy can be either person-based (i.e., a response toward targets’ moral characters) or act-based (i.e., a response toward targets’ moral behaviors). Some studies manipulated the description of characters’ moral traits (such as honesty, sincerity, and trustworthiness) and found that individuals show less empathy toward targets described as immoral than targets described as moral (Brambilla & Riva, 2017; Li et al., 2018; Riva et al., 2016). Other studies manipulated targets’ morally related acts (such as helping or harming) without directly describing the targets as moral or immoral and found that individuals show less empathy toward targets who have behaved immorally than targets who have behaved morally (Rodriguez-Gomez et al., 2020; Singer et al., 2006; Stellar et al., 2014). Therefore, both the moral character and the morally related acts of a target can influence empathy, which is also why we frame our paper as a review of selective empathy based on targets’ moral characteristics, rather than on moral character or moral behavior.

The person-based selectivity and act-based selectivity of empathy are not completely independent. A target’s moral character can be inferred from the target’s moral behavior. More specifically, the severity and frequency of a target’s immoral behavior can lead to different assessments of immoral character, which further impacts the level of empathy. For example, one study manipulated the frequency of targets’ immoral acts and found that individuals feel happier about the punishment of a person who has committed multiple immoral acts as compared to a single immoral act (Berndsen & Tiggemann, 2020). Moreover, individuals’ evaluation of the person’s moral characters mediated the influence of the frequency of immoral acts on individuals’ Schadenfreude. These findings suggest that the act-based selectivity of empathy might come from a person-based inference regarding morally-related acts.

Differences also exist between the person-based and act-based selective empathy. When the target’s moral character is described as immoral in general, individuals might be more likely to show reduced empathy toward the immoral target, regardless of whether the target’s misfortune is linked to their immorality or not, driven by the broad justice belief mentioned in the above section. In contrast, when a target’s immoral behavior is described, empathy might be more influenced by whether or not the target’s misfortune is linked to that immoral behavior. More studies are needed to examine these possibilities.

However, the belief in a just world can also lead to some unreasonable attributions in daily life. People might be likely to blame a victim based on their beliefs that “good things usually happen to good people” and that “bad things usually happen to bad people” (Lerner, 1980). For example, when watching a target suffering from painful electric shocks due to their errors in a learning task, participants tended to reject and devalue the target when they could not avoid seeing the target suffer (Lerner & Simmons, 1966). In this way, people rationalized the target’s suffering and no longer needed to feel empathy for the target. In such circumstances, people’s reduced empathy toward innocent victims may redound on their evaluation of the victims (for more details, see the future direction part).

Why Empathy Is Morally Selective?

The previous section has summarized findings of selective empathy based on targets’ moral characteristics. In general, individuals tend to show less empathy toward immoral targets than toward moral targets. This section will consider why selective empathy is morally based from both ultimate and proximate perspectives, based on previous theories and models of selective empathy. The ultimate cause may be consistent with that of selective prosociality. On this view, the moral selectivity of empathy is the product of evolution and can increase survival rates at the individual and group levels. The proximate causes may be more diverse. Below, we summarize five different accounts of proximate causes: the perceived-similarity account, the affective-link account, the deservingness-appraisal account, the dehumanization account, and the potential group-membership account.

Ultimate Causes

Empathy entails both cognitive and emotional costs. When feeling empathic toward others, individuals experience a cognitive struggle (Cameron et al., 2019) and risk experiencing personal distress (Cameron et al., 2016). Indiscriminate empathy might cause fatigue and/or financial costs (Cameron et al., 2019). The moral selectivity of empathy ensures that individuals spend more empathy-related resources on moral targets and less on immoral targets, which could increase the survival rates of both individuals and groups. For individuals, such selectivity is likely to prevent resources being wasted on undeserving persons and gives empathizers more opportunities to build reciprocal relationships with moral, rather than immoral, others who are less likely to reciprocate when the empathizers need help. The selectivity of empathy is broadly consistent with the approach-avoidance tendency to approach rewarding and avoid threatening stimuli (Kaldewaij et al., 2017).

For groups, the moral selectivity of empathy can promote indirect reciprocity within a group by shaping targets’ behaviors. More prosocial behaviors based on empathy directed at moral rather than immoral targets can encourage moral behaviors and discourage immoral behaviors by both targets and other observers (Van de Vondervoort et al., 2018). This can increase indirect reciprocity within a group. For example, C is more likely to empathize with A after learning that A is a daily helper. Then C will be more likely to help A, even though C has not been directly helped by A before. After receiving positive feedback from C, A will also become more likely to continue helping others, including C. Such indirect reciprocal relationships can promote prosociality and cooperation within a group and further increase the survival rates of the whole group (Nowak, 2006; Nowak & Sigmund, 2005; Van de Vondervoort et al., 2018). In other words, selective empathy based on targets’ moral characteristics serves as a signal of moral standards within a group. This also explains why individuals make more negative evaluations of people who empathize with negative targets (e.g., White supremacist) than people who empathize with positive targets (e.g., children’s hospital worker; Wang & Todd, 2021); empathizing with negative targets indicates tolerance of such behaviors.

Overall, the moral selectivity of empathy could increase the survival rates of both individuals and groups, which may be its ultimate cause.

Proximate Causes

We propose the five accounts below to tentatively explain why empathy becomes morally selective in specific situations.

Perceived-Similarity Account

The extent to which individuals perceive similarity between themselves and targets is an important modulator of empathy (for a review, see Preston & de Waal, 2002). The perception–action model (PAM) of empathy proposes that individuals’ perception of the state of a target can activate corresponding representations and further activate somatic and autonomic responses, leading to emotional sharing with the target (Preston & de Waal, 2002). According to this model, the more similar empathizers and targets are, the more similar empathizers’ and targets’ representations are, which leads to more similarity in state-matching (Preston & de Waal, 2002).

Variation in the perceived similarity of moral and immoral targets might provide one proximate explanation for the influence of targets’ moral characteristics on empathy. Individuals may perceive moral persons as more similar to themselves than immoral persons, given that individuals tend to view themselves as more moral than others (e.g., Allison et al., 1989). In this way, individuals would show more empathy toward moral persons than immoral persons. Riva et al. (2016) were the first to explore the possible role of perceived similarity in selective empathy toward moral and immoral targets. However, they failed to find a significant mediating role for the perceived similarity of the target’s moral standing (moral versus immoral) and evaluations of the target’s social pain, an indicator of cognitive empathy. It remains to be seen whether perceived similarity could play a mediating role between the moral characteristics of the target and other components of empathy, such as empathic concern. In another perspective on the issue of mediation, some researchers have argued that perceived similarity is actually based on liking or disliking the target (e.g., Batson et al., 2005). Individuals show more empathy toward similar others because they like them more than dissimilar others. Therefore, affective links between empathizers and targets may play a more important role in the moral selectivity of empathy than perceived similarity in itself.

Affective-Link Account

The liking or disliking of targets is an important factor that can influence individuals’ emotional reactions to targets’ misfortune (Singer et al., 2006; Smith & van Dijk, 2018). According to the balance theory of emotion, “three-element units are balanced when all relations between the elements are positive or two are negative” (Smith & van Dijk, 2018, p.295). More specifically, when individuals dislike a person (a negative element), the person’s negative feelings about his or her misfortune (a negative element) will be balanced by the observer’s positive feelings about the misfortune (a positive element). In this way, individuals tend to show reduced empathy or counter-empathy toward the misfortune of those they dislike.

With respect to the influence of targets’ moral characteristics on empathy, Singer et al. (2006) were the first to propose that individuals’ affective links with moral or immoral others can explain the selectivity of empathy. This view was initially used to explain selective empathy in a second-party situation where immoral others directly undermined participants’ benefits and were disliked by participants. A similar logic could also apply to a third-party moral situation. Individuals prefer persons who have behaved morally toward others, because these persons may also benefit them in the future; meanwhile, individuals dislike persons who have done immoral things to others, because these persons are more likely to hurt them or undermine their benefits in the future (e.g., Brambilla & Leach, 2014; Hamlin et al., 2011). In this way, individuals’ dislike of immoral persons and the negative experience of immoral persons can be balanced by individuals’ positive feelings toward the negative experience of immoral persons, leading to reduced empathy or even counter-empathy toward them. One study provided direct evidence for this affective-link account. Participants’ relative dislike of the immoral target as compared to the moral target mediated the relation between the manipulation of targets’ moral characteristics and Schadenfreude at their misfortune (Brambilla & Riva, 2017).

It is important to note that the affective-link account is based on the premise that empathizers value morality. However, how much individuals value morality can vary, as reflected in a person’s moral identity. Moral identity refers to “the degree to which being a moral person is important to an individual’s identity” (Hardy & Carlo, 2011, p. 212). If a person has a high moral identity, the person will perceive moral targets as more similar to themselves and prefer moral targets over immoral targets, thereby showing less empathy toward immoral targets than moral targets. However, if a person has a low moral identity, meaning that the person does not care to be moral or might be immoral themselves, the perceived-similarity account might not be able to explain their selective empathy, and the person might not even show selective empathy toward moral and immoral targets.

Deservingness-Appraisal Account

Deservingness is also widely believed to modulate the level of empathy, or more specifically, it influences the motivational component of empathy, namely compassion or empathic concern (for reviews, see Gibbs, 2019; Goetz et al., 2010; Hein & Singer, 2008; Simpson et al., 2014). Before showing compassion, individuals tend to evaluate how much the target deserves compassion. Simpson et al. (2014) summarized four rules used to determine whether the suffering person is worthy of compassion. If the person is “(1) responsible for their suffering; (2) had prior knowledge of the risk or danger; (3) has the means to address the situation; and/or (4) their distress has no valid organizational or social explanation” (Simpson et al., 2014, p. 480), then the person deserves to suffer and is not worthy of compassion. By implication, when the misfortune is beyond the sufferer’s control, individuals feel empathic toward the sufferer’s misfortune. However, when the misfortune is the fault of the sufferer, individuals feel less empathic or even show counter-empathy (Gibbs, 2019). For example, individuals showed more empathy toward patients who contracted AIDS through blood transfusion than those who contracted AIDS through drug-abuse (Decety et al., 2010).

The influence of deservingness also applies to morally charged situations. Driven by a belief in justice, individuals tend to believe that bad people deserve bad consequences (Hafer & Rubel, 2015; Lerner, 1980; Vermunt, 2014). Such expectations were confirmed by a recent ERP study. Rodriguez-Gomez et al. (2020) compared individuals’ ERP responses when reading sentences describing a fortunate or unfortunate outcome for agents previously described as prosocial or antisocial. They found that the brainwaves related to semantic fit (N400) were stronger when participants read sentences stating that antisocial agents met an unfortunate outcome as compared to when prosocial agents did so. This suggests that individuals expect misfortune to befall antisocial rather than prosocial agents.

When a person who has harmed others suffers, the belief that this person deserves this misfortune can reduce empathy. Such a reaction is quite common in everyday life. For example, when a serial murderer is sentenced to death, most people do not care about the murderer’s negative feelings and may even feel happy about the ending of his life. This deservingness-appraisal account of the moral selectivity of empathy is supported by previous findings. Participants’ evaluation of the deservingness of targets’ misfortune mediated the relation between the manipulation of targets’ moral characteristics and Schadenfreude at their misfortune (Brambilla & Riva, 2017). In addition, dislike of targets mediated the relation between the manipulation of moral characteristics and the evaluation of deservingness (Brambilla & Riva, 2017), suggesting that the deservingness-appraisal process can also be influenced by affective link (Smith & van Dijk, 2018).

Dehumanization Account

Dehumanization is another factor that might explain the moral selectivity of empathy. When people perceive less humanity in targets, they are less likely to protect the targets’ moral rights (Bandura, 1999; Bastian et al., 2011; Castano & Giner-Sorolla, 2006) and show less empathy toward the targets’ misfortune (Cehajic et al., 2009). Individuals who engaged in immoral behaviors are usually perceived as showing less humanity (Bastian et al., 2011). Given the negative relation between empathy and dehumanization (Cehajic et al., 2009), people might show less empathy toward immoral others due to their perception of less humanity in immoral as compared to moral others.

The dehumanization account was supported by Riva et al. (2016). They found that participants’ dehumanization of targets mediated the relation between the manipulation of targets’ moral characteristics and participants’ evaluation of targets’ social pain. Future studies could examine whether dehumanization can explain the moral selectivity of other components of empathy, such as emotional sharing and empathic concern.

Potential Group-Membership Account

Group membership is also a widely studied modulator of empathy. People show greater empathy and empathy-related brain responses toward ingroup members than outgroup members (for reviews, see Cikara et al., 2014; Eres & Molenberghs, 2013; Montalan et al., 2012; Vanman, 2016; Vollberg & Cikara, 2018). The modulation of empathy by group membership might explain the moral selectivity of empathy.

Individuals tend to cooperate with prosocial others and avoid cooperating with antisocial others (for reviews, see Kurzban et al., 2015; Nowak, 2006; Nowak & Sigmund, 2005). Therefore, moral persons can be viewed as potential ingroup members, while immoral persons can be viewed as potential outgroup members. For example, after watching someone help others on several occasions, we will form a good moral impression of this person and also want to be the person’s friend. By contrast, after watching someone harm others on several occasions, we will form a bad moral impression of this person and view the person as a potential enemy. Given that individuals usually have more empathy toward ingroup members than outgroup members, moral targets (potential ingroup members) might provoke more empathic feelings than immoral targets (potential outgroup members). However, more evidence is needed for this assumption.

In addition, the group membership of targets might interact with targets’ moral characteristics to influence empathy. It would be interesting to examine empathy responses when there is a conflict between the group membership of targets and their moral characteristics. For example, researchers could compare individuals’ empathy for the suffering of an ingroup member who engaged in immoral acts and the suffering of an outgroup member who engaged in moral acts (for a similar design, see Meidenbauer et al., 2018).

The five accounts above explain the proximate causes of moral selectivity of empathy from different perspectives. The affective-link account, the dehumanization account, and the potential group-member account can explain the selectivity of all three components of empathy, including emotional arousal, cognitive empathy, and empathic concern or compassion. The other accounts focus more on a particular component. The deservingness account explains the selectivity of empathic concern or compassion (Gibbs, 2019; Goetz et al., 2010; Hein & Singer, 2008; Simpson et al., 2014), and the similarity account explains the selectivity of emotional arousal (Preston & de Waal, 2002). The relative importance of these five accounts for the moral selectivity of empathy remains unclear. Affective links may play a more fundamental role in the moral selectivity of empathy because they mediate the effects of perceived similarity and deservingness appraisal (Batson et al., 2005; Brambilla & Riva, 2017). It is also possible, however, that the importance of different accounts varies across situations and individuals. Future research could examine these accounts of the moral selectivity of different components of empathy. These accounts also provide several possible ways to link the moral selectivity of empathy with other types of selective empathy.

How Does Empathy Become Morally Selective?

In this section, we discuss in more detail the process by which empathy becomes selective in response to targets’ moral characteristics. Decety and Meyer (2008) propose a dual-processing model in which empathy involves both an automatic, affective, bottom-up process, and a controlled, cognitive, top-down process. First, individuals automatically share the feelings of others by matching their perception of others’ feelings with their own feeling-related reactions via a bottom-up process. Then, self-other awareness enables individuals to distinguish their own feelings states from those of others. Next, individuals take the perspective of the other and regulate their empathic levels through executive control and emotion regulation via a top-down process. This latter regulation process may be closely related to empathic concern (Decety & Meyer, 2008). Based on previous findings and the dual-processing model, we propose three possible explanations for the process of selective empathy: automatic, regulative, and mixed.

An Automatic Pathway

The automatic explanation is that individuals automatically show less empathy for antisocial targets than prosocial targets without going through some top-down regulation. If empathy becomes morally selective through this automatic route, individuals will show less emotional arousal toward antisocial targets than prosocial targets, following by less empathic concern and cognitive empathy. This pathway is supported by the ERP findings of Cui et al. (2016) that N2, an early ERP component indicating emotional arousal, showed larger amplitude for painful pictures than non-painful pictures when the target was a blood donor or was unidentified, but showed no difference for painful as compared to non-painful pictures when the target was a killer. The selective empathic responses appearing at around 200 ms after the appearance of the stimuli suggest that the selectivity of empathy is quick and thus very likely to be automatic (for a similar logic in deciding whether a response is automatic, see Kahneman, 2011).

However, this automatic explanation cannot explain all the findings of selective empathy based on targets’ moral characteristics. For example, it cannot explain why the P3 component, the indicator of top-down regulation of empathy, showed varied responses toward moral and immoral targets (Li et al., 2018). A pathway that can deal with more complex situations is needed.

A Regulative Pathway

With the gradual deepening of empathy research, more and more researchers view the production of empathy from a regulative perspective (Cameron, 2018; Cameron et al., 2022; Gross, 2015; Thompson et al., 2019; Zaki, 2014). Previous studies have found that people can regulate their empathy and show insensitivity to others’ suffering (Cameron & Payne, 2011). Such a process is similar to the regulation of other emotions (for a review, see Tamir, 2016). Similarly, in the case of the moral selectivity of empathy, it is possible that individuals might first show equal levels of empathy for both moral and immoral targets, and then down-regulate their empathy for antisocial targets.

Empathy is usually modulated in three different ways: via situation selection, attention modulation, and appraisal. Each type of modulation can help to explain individuals’ selective empathy toward moral vs. immoral people. The first way to modulate empathy is via situation selection (Gross, 1998). Individuals can avoid or increase empathy by avoiding or approaching empathy-inducing cues (Cameron, 2018; Thompson et al., 2019; Zaki, 2014). If people do not want to show empathy for immoral targets, they might be more likely to leave situations that could arouse their empathy for immoral targets. For example, we might be less willing to listen to a selfish colleague’s complaint about his divorce from his wife than to listen to a kind colleague’s similar complaint.

The second way to modulate empathy is via attention modulation. People can regulate their empathy by modulating their attention to empathy-inducing cues (Cameron, 2018; Thompson et al., 2019; Todd et al., 2012; Zaki 2014). Individuals might allocate less attention to the misfortune of immoral as compared to moral targets. The ERP study by Li et al. (2018) provided some support for this proposal. The late ERP component of empathy (P3) showed larger amplitude toward painful pictures than nonpainful pictures when the target was imagined as moral but showed no such amplitude difference when the target was imagined as immoral. P3 is believed to be associated with the top-down processing of stimulus significance (MacNamara et al., 2009). Although the pictures of suffering were actually the same sets of pictures of wounded body parts, participants viewed the suffering of an immoral person as less significant than that of a moral person, suggesting that attention modulation happens during the processing of painful pictures of immoral targets.

If individuals have attended to the suffering of targets, they can still modulate their empathy via reappraisal, the third way to modulate empathy. Reappraisal has three different forms (Zaki 2014). First, modulation of empathy can occur by changing the perception of targets’ feelings. When processing others’ suffering, individuals might shift their beliefs about the extent of an immoral target’s suffering. For example, when listening to your selfish colleague’s sad story about his divorce, you might think that he is not actually as sad as he claims to be. This type of reappraisal is supported by previous findings that individuals tend to perceive the social suffering of an immoral person as less acute than that of a moral person (Riva et al., 2016), suggesting that individuals’ perception of antisocial targets’ feelings have shifted. This interpretation might also explain why the perception of the physical pain of an immoral as compared to a moral person does not differ (Riva et al., 2016). Beliefs about physical pain based on objective descriptions are likely harder to shift. Second, appraisal processes might operate via a re-evaluation of the deservingness of the target’s misfortune. Individuals show less empathy when they believe that the sufferers are responsible for their own suffering, or in other words, they deserve their suffering (Bloom, 2017; Goetz et al., 2010; Zaki, 2014). By reappraising the misfortune of immoral persons, people come to believe that immoral persons deserve their misfortune, which is possibly driven by the justice-based belief that those who behave antisocially will be punished and that those who behave prosocially will be rewarded (Hafer & Rubel, 2015; Lerner, 1980). One study supporting this deservingness explanation found that targets’ moral characteristics influenced participants’ feelings about the misfortune of targets via the mediation of participants’ evaluation of whether those misfortunes were deserved (Brambilla & Riva, 2017). Apart from changing the perception of targets’ feelings and deservingness appraisal, another form of reappraisal is to reconstruct views of the misfortune itself and come to view it as a good thing (Zaki, 2014). When immoral persons suffer, people can view the immoral persons’ misfortune as a lesson to them, teaching them not to do bad things in the future. Even if the immoral persons cannot receive or benefit from the lesson, observers can be led to believe that doing bad things brings misfortune, such that empathizers have fewer negative feelings about the misfortunes of immoral targets.

In sum, through a reappraisal of the misfortunes of immoral targets (by changing the perception of the targets’ feelings, via a deservingness appraisal, or by constructing a positive view), individuals can reduce their empathy toward immoral targets. Each of the above processes (situation selection, attention modulation, and reappraisal) can explain how empathy can be modulated. Some evidence supports the attention-modulation and appraisal explanations, but more evidence is needed for the situation-selection explanation.

The specific context might also influence the process by which empathy is modulated. When individuals already have a stable impression of the moral character of a target and can choose the situation themselves, their selective empathy might be more likely to occur via situation selection. When individuals only know some morally related acts of a target, they might need more time to reappraise the target’s misfortune. In the course of gradually forming a stable impression of a target’s immoral character, empathy regulation might gradually change from reappraisal to attention modulation and to situation selection. Finally, it remains unclear whether different components of empathy are regulated in different ways (Weisz & Cikara, 2021), an issue that warrants further investigation.

A Mixed Pathway

Empathy modulation might not be necessary for the moral selectivity of empathy. Individuals’ reduced empathy toward antisocial targets might be the mixed product of empathy and schadenfreude. Individuals show equal levels of empathy for both prosocial and antisocial targets, but meanwhile they also feel happy (or more specifically, schadenfreude) at the misfortune of antisocial targets. This mixture of empathy and schadenfreude explains the outcomes of reduced empathy for antisocial targets. This potential pathway is hard to verify based on current findings because most studies only measured the outcomes of individuals’ empathic responses, such as asking them to report on their feelings after learning about the misfortunes of antisocial or prosocial targets (Berndsen & Tiggemann, 2020; Brambilla & Riva, 2017; Riva et al., 2016), without focusing on individuals’ negative and positive responses at the same time.

However, one observational study of children provides some suggestive evidence for this pathway. Mendes et al. (2018) coded children’s facial expressions when watching the punishment of an antisocial puppet who tricked them and a prosocial puppet who gave toys to them. Six-year-olds produced more smiles with frowns when watching the antisocial puppet being hit than when watching the prosocial puppet being hit. Children’s frowns might reflect their empathic responses toward the suffering of targets, while their smiles were presumably signs of Schadenfreude. Their smiles with frowns suggest that they felt both empathic and schadenfreude at the antisocial puppets’ suffering, supporting the mixed route.

Overall, we have proposed three ways to explain how empathy becomes morally selective. The first is that individuals automatically show less empathy for antisocial targets than prosocial targets. The second is that individuals need to regulate their empathy for targets with different moral characteristics via situation selection, attention modulation, or reappraisal. The third is that selective empathy is a mixture of empathy and schadenfreude. Each pathway has its supporting evidence. These three pathways might apply to different situations or age groups. When the immoral targets are extremely immoral (e.g., killer), individuals might show selective empathy automatically with little regulation. When the immoral targets are not so evil but still bad (e.g., people who take advantage of others), individuals might need to regulate their empathy toward such immoral targets. For children whose abilities to integrate their moral evaluation of targets and their empathy are still limited, their selective empathy is more likely to be a mixed response of empathy and schadenfreude, such as smiles with frowns (Mendes et al., 2018). Adults whose moral evaluation is well-developed and who can regulate their empathic responses are more likely to show selective empathy automatically toward extremely immoral others or show selective empathy via top-down regulations.

Future Directions

We provide the first comprehensive review of the influence of targets’ moral characteristics on empathy. The sections above draw together a number of hitherto scattered studies on the moral selectivity of empathy and explain why and how empathy becomes selective in response to the moral characteristics of targets. It enriches our understanding of the functions of selective empathy and opens up several avenues to further address the relation between empathy and morality.

The Reverse Influence of Selective Empathy on Moral Cognition

When reviewing the influence of targets’ moral characteristics on empathy, we mainly focused on situations where the targets’ moral characteristics are easy to identify and quite objective. However, moral judgments in daily life can be more complicated than the situations in experimental studies. When the moral evaluation of targets is uncertain, individuals’ moral evaluation might be influenced by the states of their empathy. For example, when people do not want to feel empathy, they might project immorality onto a target, just to rationalize their feelings.

This reverse pathway from (lack of) empathy to moral cognition is apparent in previous work on victim blaming (Johnson et al., 2002). For example, when a person surfs the Internet and finds a news report that a girl was sexually assaulted at a party, instead of empathizing with the girl, the person might comment that girls should pay attention to their clothes at a party or that the girl behaved provocatively, thereby implying that the victim was responsible for the sexual assault. Malicious comments like this might be the product of failed empathy. When the person cannot empathize with the victim, or feels that it would be burdensome to empathize, the person might rationalize their callous reactions by evaluating the victim as immoral or negative. It is conceivable that such unfeeling responses are more common in today’s society. The information explosion on the Internet, and more specifically, the explosion of reports of suffering and brutality, might deplete people’s empathic resources. To rationalize their reduced empathy toward others’ misfortunes, people may make inaccurate inferences about the character of victims. More studies are needed to examine this plausible reverse influence of selective empathy on moral cognition and to find ways to avoid such rationalization of empathy depletion.

Such vilification of victims is similar to the process of dehumanization of victims in contexts of moral disengagement. To feel less responsible for their immoral behaviors, people tend to change how they view the victims of those behaviors (Bandura et al., 1996, 1999). For example, when an army brutally executes enemy civilians, they tend to view them as less than human, rather than as ordinary individuals with feelings and families.

The Moral Selectivity of Positive Empathy

All the studies reviewed above focused on empathy for the misfortune of others, but individuals can also share in the good fortune of others. Positive empathy refers to the sharing and understanding of others’ positive emotions (Morelli et al., 2015). Different from negative empathy, it activates brain regions associated with positive affect (e.g., ventromedial prefrontal cortex; Morelli et al., 2014). Similar to negative empathy, it also promotes prosocial behaviors (Morelli et al., 2015; Telle & Pfister, 2016) and is modulated by many factors such as group membership, degree of liking, and justice (Smith & van Dijk, 2018). Corresponding to Schadenfreude, there is a concept called Gluckschmerz, which refers to pain at others’ good fortune (Smith & van Dijk, 2018). Both Schadenfreude and Gluckschmerz reflect counter-empathy for others’ experiences.

Several findings suggest that individuals’ positive empathy also depends selectively on targets’ moral characteristics. For example, adults were found to express less happiness and more anger toward an antisocial person’s good fortune than that of a prosocial person (Rodriguez-Gomez et al., 2020). Additionally, 4- to 9-year-old children tend to evaluate antisocial targets as less happy than prosocial targets even when both antisocial and prosocial targets are described as having similarly happy experiences (Yang et al., 2021). Indeed, positive empathy might be more selective in terms of targets’ moral characteristics than negative empathy. Previous studies found that individuals would share the negative feelings of both strangers and ingroup members, but only share the positive feelings of ingroup members, implying that positive empathy is more sensitive to targets’ group identity (Molenberghs et al., 2014; Motomura et al., 2015).

The asymmetrical selectivity of positive and negative empathy might also reflect individuals’ biased responses to positive and negative information (i.e., a negativity bias, for reviews, see Peeters & Czapinski, 1990; Rozin & Royzman, 2001). More specifically, negative information is usually more salient than positive information for individuals. Based on this logic, negative experiences of targets might draw more attention from empathizers than positive experiences of targets. Given that others’ negative experiences can also be a danger signal (Schulz, 2017), empathic responses toward negative experiences may be less selective than toward positive experiences because the negative experiences of others can signal potential danger no matter whether they are moral or immoral. More research is needed to examine the moral selectivity of positive empathy and compare the moral selectivity of empathy for positive and negative events.

Finally, it is important to note that individuals’ reduced empathy or counter-empathy toward the fortune of immoral others might also stem from, or be mixed, with envy or jealousy. Envy refers to a painful feeling caused by the good fortune of others, and usually results from social comparisons (Crusius et al., 2020). Immoral targets are usually perceived as inferior in morality. The good fortune of immoral targets might provoke a sense of conflict, making individuals experience greater envy toward the good fortunes of immoral as compared to moral targets. These varied feelings of envy might also influence the selectivity of positive empathy.

The Role of Selective Empathy in Selective Helping and Third-Party Punishment

Previous studies of third-party punishment and selective helping found that individuals were more likely to punish immoral others (e.g., Fehr & Gächter, 2002; Jordan et al., 2014; Kenward & Östh, 2015; McAuliffe et al., 2015; Yudkin et al., 2019) or avoid helping immoral as compared to moral others, even though their own benefits and wellbeing were not impacted by others’ immoral behaviors (e.g., Dahl et al., 2013; Malti et al., 2016; Vaish et al., 2010; Van de Vondervoort et al., 2018; Wedekind & Milinski, 2000). Given the positive role of empathy in prosocial behaviors (de Waal, 2008; Eisenberg, 2000; Hoffman, 2001), selective empathy may be the emotional precursor of selective helping and third-party punishment. Previous studies have found that children’s Schadenfreude and compassion played a mediating role between the targets’ moral characteristics and children’s willingness to help them (Schindler et al., 2015; Schulz et al., 2013), thereby providing evidence for a mediating pathway from selective empathy to selective helping.

Meanwhile, a change in empathy may not be the only explanation for individuals’ selective behaviors toward prosocial and antisocial others. Social norms and inhibitory control abilities can also influence third-party punishment (Krueger & Hoffman, 2016; Su et al., 2019). Nevertheless, selective empathy is likely to play a key role in situations involving emotion expression, such as helping others in distress or punishing others to make them feel pain.

Conclusion

This paper has reviewed evidence for selective empathy based on targets’ moral characteristics from the perspectives of three components of empathy (emotional arousal, perspective taking, and empathic concern). It then discussed why empathy is morally selective with both ultimate and proximate causes and how it becomes morally selective via three possible pathways. We also discussed the possible bi-directional relation between selective empathy and the moral evaluation of targets (which can explain victim blaming) and the possible role of morally selective empathy in selective prosociality. A graphic summary of our review can be seen in Fig. 1.

Fig. 1.

The relation between selective empathy and moral evaluation

Previous studies of the selectivity of empathy have mainly focused on how it can be biased and irrational, i.e., selective empathy based on racial or group bias (de Vignemont & Singer, 2006; Decety, 2021; Han, 2018; Montalan et al., 2012). However, by documenting the impact of moral considerations on empathy, this paper highlights a more positive aspect of selective empathy. More generally, the paper enriches our understanding of the relation between morality and empathy and should inspire future research on the moral functions of empathy from new perspectives.

Additional Information

Funding

This research was supported by the National Natural Science Foundation of China (Grant Nos. 31872782, 32071075).

Conflict of Interest

The authors declare no conflict of interest.

Availability of data and material

Not applicable.

Code availability

Not applicable.

Authors' contributions

Y.W. conceived of the theoretical framework, drafted the manuscript, and revised the manuscript. P.H. contributed in critical manuscript revisions. M.P. revised the manuscript. Y.S. acquired funding and revised the manuscript. All authors reviewed and approved of the final manuscript.

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

The authors confirm that the work has not been published before and that its publication has been approved by all co-authors.