A Less Attractive Feature of Empathy: Intergroup Empathy Bias

Empathy with others’ successes and misfortunes is a critical component of group living that promotes social cohesion. Unfortunately, empathy is a malleable phenomenon in that its elicitation is not automatic, but modulated by multiple interlocking factors. This chapter explores the specific phenomenon of intergroup empathy bias—the difference in empathy for members of social ingroups versus outgroups—which poses profound chal ‐ lenges for our modern human world characterized by a multitude of groups, ethnicities, and cultures. The chapter frames the discussion by contextualizing empathy as consist‐ ing of three interacting component processes, namely experience sharing, perspective taking, and empathic concern. It then goes on to examine research describing the effects of intergroup bias on each of these component processes. Next, it explores the factors, both at the level of the group and at the level of the individual, which may contribute to empathic breakdown in intergroup contexts. Finally, it considers strategies that may have potential in mitigating intergroup empathy bias. Here, we draw on our own experi‐ ences in the South African context, which is characterized by pervasive racial inequality and legacies of apartheid violence, to suggest that intergroup empathy is best stimulated in a context of reciprocal mutual engagement with the other.


Introduction
Humans are ultra-social organisms, because they form and depend on organizations that extend beyond the individual [1]. This social interdependence arguably evolved because group living, and its associated social functions, offered several reproductive and long-term survival advantages, compared to going solo [2]. In turn, these social functions necessitated the evolution of more sophisticated cognitive and emotional capacities, like theory of mind and moral emotions, to enable thriving within a group context [3]. Critical among these is our ability to empathize: the capacity to think and feel oneself into the inner reality of another person, while recognizing that their emotional experience is separate from our own [4]. Empathizing with others in distress is particularly important because it motivates behavior aimed at alleviating those others' suffering, and this, in turn, promotes social cohesion and resource-sharing among members of society [5,6].
Unfortunately, witnessing a person in distress does not inevitably evoke feelings of empathy, nor does it always result in prosocial helping behavior. Even though we may encounter many potential empathy-eliciting scenarios in our everyday lives, we respond with empathy to only a fraction of them [7,8]. In fact, recent evidence suggests that empathic failures are not always characterized by attenuated empathy or indifference, but quite often by counterempathic responses, like Schadenfreude and Glückschmerz, which may facilitate hostility [9,10]. Empathic reactions are therefore not automatic, but rather, the degree to which we respond empathically are modulated by multiple interlocking factors, which science is only beginning to unravel. For example, growing evidence suggests that empathic responding is influenced significantly by personal features of the empathizer (e.g., gender, trait empathy, childhood trauma), by interpersonal factors (e.g., perceived fairness, social stigma), by cultural factors (e.g., interdependence vs. independence, preference for social hierarchy), and importantly, by the social group membership of the person in distress (e.g., race, political affiliation, sports team identification) [11][12][13][14][15][16][17][18]. Although empathy for one's own social group is particularly important and holds several long-term advantages [19], the flipside of this phenomenon, i.e., diminished empathy for the outgroup, poses profound challenges for our modern human world where a multitude of groups, ethnicities, and cultures compete for the same resources [20].
This chapter explores a fundamentally important, albeit less attractive, feature of empathy, namely, its breakdown in response to the pain of outgroup others. In particular, it asks the following questions: Where and how does the empathic process break down? Which factors exacerbate this empathic failure? And is it possible to ameliorate, or even reverse, these effects? The chapter frames the discussion by contextualizing empathy as consisting of three interacting component processes, namely, experience sharing, perspective taking, and empathic concern. It then goes on to examine research describing the effects of intergroup bias on each of these component processes. Next, it explores the factors, both at the level of the group and at the level of the individual, which may contribute to empathic breakdown in intergroup contexts. Finally, it considers strategies that may have potential in mitigating intergroup empathy bias.

Empathy component processes
Empathy is a complex psychological phenomenon. Whereas early conceptualizations of empathy have typically stressed either its cognitive or affective aspects [21], more recent research considers empathy a multidimensional construct that may be parsed into three dissociable but interacting neurocognitive components: emotional sharing, perspective taking, and empathic concern [22,23].
Emotional sharing constitutes the affective component of empathy and is also commonly referred to as emotion contagion or affective resonance. It refers to the automatic capacity to become aroused by another's emotions and relies on subcortical emotion circuits [7]. For example, seeing an angered individual may lead the empathic observer to vicariously experience similar feelings.
Perspective taking constitutes the cognitive component of empathy and refers to the capacity to view a situation from another's point of view, or to put oneself in their shoes, as it were. Doing this allows one to better recognize and understand another person's affective experience. It therefore partly overlaps with theory of mind-like processing and relies on the mentalizing network [24,25]. Because perspective taking begins with the perception of affective signals from another individual, low-level processes involved in face perception and emotion recognition are integral to understanding another individual's internal state [26].
Finally, empathic concern constitutes the motivational component of empathy and refers to other-oriented feelings of care and compassion when perceiving an individual in distress. The experience of empathic concern may stimulate prosocial helping or caring behaviors in the observer [5,27]. Recent neuroimaging studies have highlighted the role of the medial prefrontal cortex (MPFC) [13] and septal area [25] in empathy-driven prosocial motivation.
Each empathic component is sensitive to both bottom-up automatic affective processes and to top-down perceiver-controlled cognitive processes, which feed into each other [7]. For example, too much affective arousal may result in personal distress within the observer, promoting a self-focused escape response rather than other-oriented empathic response [28]. In turn, executive functions serve to consciously regulate the self-focused distress response, so that an awareness of the other's perspective and concern for their circumstances becomes possible [29]. These interconnected mechanisms function together to produce an overall empathic response that combines automatic experience sharing as well as intentional feelings of concern.

Intergroup empathy bias
Abundant evidence from social psychology suggests that individuals instinctively categorize themselves into salient social groups with which they identify and with whom they feel a sense of belonging [30,31]. Unfortunately, this categorization also maximizes differences between groups, leading people to more readily distrust, fear, and discriminate against outgroup members and to instinctively favor or side with members of social groups with which they identify [32].
Importantly, this intergroup bias has been shown to impact emotional responding toward others when it matters most-when they are in pain. Whereas greater empathy toward one's ingroup makes sense from an evolutionary perspective [20], empathic failures toward outgroup members may lie at the heart of most intergroup conflict situations, including political violence, xenophobia, and genocide [33]. Interestingly, outgroup empathy failures do not seem to depend on a person's trait empathic concern [34]. That is, even the most deeply empathic person can mute their empathic response toward a perceived enemy under the right circumstances-a phenomenon that has been referred to as the mind's "empathy gap" [35]. Although the mechanism that underlies this empathy gap remains unclear, studies investigating intergroup empathy have demonstrated that outgroup membership status can compromise all levels of empathic responding (i.e., affective, cognitive, and motivational), as well as helping behavior [18,[36][37][38].
An important aspect of intergroup bias is the fact that it appears to depend heavily on the perceiver's social motivation: various studies have shown that self-categorization broadly along some ingroup/outgroup distinction is flexible and that re-categorization with an arbitrarily defined group may be sufficient to override automatic response biases [39]. In these studies, participants are typically assigned to novel groups using a minimal group paradigm [40] and then subjected to a variety of tasks assessing perceptual, affective, and behavioral ingroup biases [41][42][43][44]. Recently, researchers have specifically advocated the advantages of assigning people to novel groups, rather than focusing on specific social or historic groups, to advance our understanding of the processes that guide intergroup behavior across multiple contexts and levels of analysis [45].
When it comes to empathic responding more specifically, arbitrary group categorization by way of a minimal group manipulation can also facilitate intergroup biases, particularly when the groups are in competition [9]. For example, previous research found that similar group membership between a helper and target (regardless of whether the group was real or artificially determined) strengthened the role of empathy and helping [46]. Similarly, mere categorization of participants into non-relevant social groups appears sufficient to facilitate an ingroup bias in empathy for physical pain [47].
Not many studies have explored the relationship between race and minimal group biases in empathic responding, however, and results from these studies are inconsistent. For example, in a recent study, participants showed clear minimal group biases (unaffected by race) on both an explicit group identification and implicit affective priming task, whereas neural imaging responses were indicative of significantly greater empathic arousal in response to own-race compared to other-race individuals [48]. The authors concluded that racial categorization may be a stronger modulator of the ingroup bias in empathic neural processing than general social group categorization (i.e., by assigning participants into random teams). One possible reason for this finding is that humans may (automatically) detect and encode race as a by-product of an adaptation to identify fellow group members [49,50]. Given that findings to date have been inconclusive, however, more research will be necessary to tease apart these effects (see also [51]).

Group membership and emotional sharing
Emotional sharing is important in the context of empathy, because it plays a fundamental role in generating the motivation to care for and help another individual in distress [52]. Despite a general notion that affective arousal is automatic, empirical evidence suggests that many variables, including a priori attitudes and culturally learned associations, affect its induction in the observer (see e.g. [53,54]). Notably, reduced affective resonance when viewing an outgroup member in pain may be associated with fewer physiological signals from the observer's body to help interpret the other individual's emotional state and stimulate prosocial action [26].
Various studies exploring affective resonance in response to in-and outgroup members in physical pain found dampened autonomic arousal in response to outgroup members' pain [17,37]. Interestingly, in these studies, greater levels of racial prejudice on implicit measures of prejudice were associated with greater lack of empathic arousal toward outgroup members. Reduced emotional sharing in response to outgroup members is not exclusive to scenarios of physical pain; however, it also extends to scenarios of emotional pain. For example, Gutsell and Inzlicht [55] recorded electroencephalographic (EEG) alpha oscillations when participants observed ingroup and outgroup members expressing sadness. They found that, whereas participants showed similar activation patterns when experiencing sadness themselves and when observing ingroup members feeling sad, participants did not show these same vicarious activation patterns when observing outgroup members feeling sad. Participants thus appeared to experience reduced emotional sharing in response to outgroup members' sadness, and this became more pronounced the more prejudiced they were.
Several neuroimaging studies have shown that intergroup empathy bias may also manifest as increased hemodynamic activation in the anterior cingulate cortex (ACC) and anterior insula-areas thought to underlie the subjective representation of affective distress [18,56,57]. Notably, in a recent study, we detected significantly enhanced amygdala activity in response to own-race compared to other-race individuals in perceived physical pain [58]. The amygdala is typically activated during events high in emotional salience or novelty and may direct attention to motivationally relevant stimuli [59]. We thus argued that heightened amygdala activation toward same-race individuals in pain reflects approachrelated motivation and attention in line with task demands, which urged participants to empathize with individuals in distress [39].

Group membership and perspective taking
At the level of emotion recognition, research suggests that observers are faster and more accurate at distinguishing own-race compared to other-races faces [60] and are better at identifying emotional expressions of racial ingroup compared to outgroup members [54,61]. Research has also found that people are less likely to attribute secondary emotions, which are uniquely human characteristics, to outgroup compared to ingroup members [38,62].
When it comes to perspective taking, people appear to be more likely, or more accurate, in taking the perspective of an ingroup member compared to an outgroup member. For example, when Asian and White participants viewed photographs of members of both racial groups in negative contexts (e.g., illness, grief, injury) and positive contexts (e.g., party, amusement, smiling), participant self-report data indicated greater perspective taking and empathy for own-race members than other-race members, particularly in the negative contexts [36]. Furthermore, several neuroimaging studies, including our own, demonstrated stronger hemodynamic activation in response to others' pain for racial ingroups versus outgroups in regions associated with mentalizing, including the medial prefrontal cortex (MPFC), temporoparietal junction (TPJ), and precuneus [12,58,63]. In our study, we concluded that, because these areas form part of networks implicated in self-referential processing, episodic memory retrieval, and thinking about other minds, heightened activity may allow for a richer representation of another's physical/psychological pain. By implication, reduced activity in these areas suggests impaired perspective taking when it comes to outgroup members.
Perspective taking is thought to enhance empathy by creating increased overlap between "self" and "other" cognitive representations, thereby reducing the self-other gap [64]. Merging another individual into one's self-concept thus results in a feeling of "oneness" and a sense of shared identity with the other person, which facilitates understanding someone else's emotions as if they were one's own. Unfortunately, people experience self-other merging much more easily with those whom they perceive as more similar to themselves, such as family members and close friends [65].

Group membership, empathic concern, and helping behavior
Various lines of evidence suggest that empathic concern and resulting helping behaviors are affected by intergroup biases. For example, Drwecki et al. [66] found that White participants reported greater empathic concern for White individuals in pain than for Black individuals in pain, and offered higher levels of analgesic treatment for ingroup compared to outgroup members. Also, in two neuroimaging studies, empathy-related neural activity when observing ingroup members' suffering relative to outgroup members' suffering predicted greater willingness to donate time and money [63], as well as greater costly helping [18] for ingroup members at a later stage.
The literature on helping behavior is complex, however, and a variety of factors may influence one's decision to offer help, including altruistic motivation, a sense of similarity with the outgroup, self-regulatory depletion, and competitive or status-related processes [26,67]. A quantitative meta-analysis of White individuals' helping directed toward Black individuals indicated no overall tendency to discriminate against racial outgroup members [68]. Instead, White individuals tended to help Black and White individuals equally, except when helping required considerable effort, time, or risk, which then resulted in an ingroup bias in helping. Broadly speaking, the literature suggests that when helping requires greater resources, individuals may cognitively justify not helping by basing their decision on reasons other than the (racial) outgroup of the person in need. Although helping ingroup members may thus be motivated largely by empathic concern, helping outgroup members may involve further systematic decision-making based on the costs and benefits of offering help [46].

Factors that contribute to empathic breakdown in intergroup contexts
In recent years, at least two things have become increasingly apparent in intergroup research: social group membership is highly flexible and context-dependent, and not all outgroups elicit intergroup empathy bias equally. In trying to tease apart the complex array of factors that influence intergroup responding-at least at the level of the group-Cikara and van Bavel [45] have recently argued that two factors are critical: functional relations between groups (shared, competing, or independent goals) and relative group status (access to resources).
Empathic breakdown between members of rival or conflict groups is well documented [69][70][71]. People's relationships to others play a significant role in determining how they respond to their suffering: whereas a friend's misfortune typically elicits empathy, a foe's misfortune might be experienced as less distressing, or even as pleasurable [72,73]. Unfortunately, intergroup contexts tend to exacerbate people's motivation not to empathize or care about someone else's misfortune, in that groups may provoke significantly more competition and aggression than interpersonal interactions [10,74].
Explicit competition between groups has the effect of increasing the salience of social identity and generally strengthens the positive relationship between ingroup identification and intergroup bias and hostility [31,75]. For example, in a neuroimaging study of real-world sports rivalry between avid fans of the Red Sox and Yankees baseball teams, ingroup team failures were associated with increased activity in neural areas associated with the subjective experience of pain [76]. In contrast, outgroup team failures were associated with increased selfreported pleasure and activity in neural areas associated with reward processing. Moreover, and rather disturbingly, the more positive value (pleasure) participants attached to rival team failures, the more they were willing to aggress against a fan of the rival team. Similar results were also observed when novel groups were pitted against each other [9]. Of significance is that in the latter study, intergroup empathy bias between competing groups was robust beyond contexts that defined the groups themselves, and even when the competitive threat of the outgroup was removed (e.g., feedback that the outgroup has fallen behind).
A second factor central to intergroup dynamics concerns the question of resources: To what extent does a social group have the power to carry out their intentions? [45]. Groups higher up the social hierarchy have more status and greater access to resources, and thus greater potential threat value, whereas groups lower down the hierarchy are typically scorned and pose less of a threat [77]. Even without overt competition, differences in power and resources between groups have been shown to predict perceptions of competitiveness [78]. Importantly, historical asymmetries in power and status between groups affect intergroup empathic responding, as well as lower level perceptual processes that operate outside awareness [79].
In an interesting study assessing people's perceptions of Black and White pain, the authors detected a consistent racial bias in evaluations, such that Black people were consistently perceived to experience less pain than White people [80]. Crucially, this bias in pain perception (by both Black and White Americans, including nursing professionals) could not be attributed to racial prejudice, but instead appeared rooted in perceptions of status and the privilege or hardship it confers. Hence, the less privileged a target seemed, the less pain participants thought he/she would experience. In a similar vein, another study using facial electromyography showed that an individual's relative social status affects how other people respond to their misfortune: participants felt less bad and smiled more when negative events happened to high-status compared to low-status individuals [81].
Although group membership significantly impacts empathic responding, empathy is not solely influenced by external factors, such as the race or status of the person in distress. Individual differences of the perceiver may also moderate the extent to which there is an intergroup bias in empathic responding. That is, based on individual traits, some people might be more likely to show strong intergroup biases in empathic responding than others.
Empirical studies show that the strength of racial identification may contribute to intergroup bias in empathic responding [63,82]. Although social group membership defined according to race is a prominent aspect of interactions among individuals, the extent to which people identify with their own racial group varies from person to person [83]. Some individuals regard their racial identity as a crucial part of their self-concept, whereas others may not feel a strong belonging to their racial group. Strong racial identification makes it more likely that an individual will process the emotions of racial ingroup members in a self-referential manner, resulting in greater empathy toward own-race individuals. Interestingly, some research suggests that pervasive discrimination against members of disadvantaged groups is associated with increased ingroup identification, which, in turn, may alleviate some of the negative psychological consequences of societal dehumanization [84,85].
Another important factor that contributes to variation in intergroup relations is motivation to respond without prejudice [86]. Because overt racial discrimination is not socially acceptable, society generally favors individuals who act in non-prejudiced ways. Hence, individuals are motivated to alter their behavior to appear non-prejudiced. The strength of the motivations to respond without prejudice, and the extent to which these motivations influence behavior, however, vary between individuals [87]. Furthermore, people may be motivated primarily by sincere changes in their personal attitude (internally motivated; IMS), or they may be motivated primarily by external pressures to avoid judgment or punishment from others (externally motivated; EMS) [88].
Data from our research have shown that different motivations to appear non-prejudiced can modulate intergroup empathic responding. For example, in a functional magnetic resonance imaging (fMRI) study of intergroup empathy, higher EMS scores in White participants were associated with dampened neural empathic responses toward Black individuals in both physical and emotional distress [58]. And in another study, higher IMS scores in White participants were positively associated with prosocial helping behavior toward a Black individual in distress, whereas higher EMS scores were negatively associated with prosocial helping toward that individual [89].

Strategies to reduce intergroup empathy bias
In the introductory section of this chapter, we presented an evolutionary perspective as framework for understanding the social interdependence of human beings that is foundational in the development of empathy. In presenting strategies to reduce intergroup empathy bias, we draw on theoretical formulations from different branches of psychology regarding the development of empathy in intergroup contexts. Notably, various studies in social psychology have demonstrated the efficacy of cognitive strategies (e.g., increased attention to an individual's feelings vs. his/her group status), manipulating the intergroup relationship (e.g., cooperative vs. competitive), manipulating group membership (e.g., recategorizing or decategorizing individuals), blurring group boundaries (e.g., reducing perceptions of group entitativity), and effortful perspective taking, in reducing intergroup empathy bias [9,51,[90][91][92]. While success in reducing intergroup empathy bias along these lines is thus possible, results of such strategies appear to be highly context dependent [93]. Moreover, investigators have rarely evaluated the efficacy of the strategy they employed beyond the immediate study context or longitudinally [94].
Scholars influenced by the relational psychoanalytic discipline have often followed a different approach in restoring empathic bonds, observing that the need and proclivity for connection are central to human development [95]. Stolorow and Atwood, for example, have argued for the primacy of interconnectedness and advanced the theory of intersubjectivity-a development of self, understood in interaction with others [96]. Accordingly, connection with others is fundamental in the development of one's identity, and experience and subjectivity are shaped by these relationships with others. The intersubjective epistemological model provides an important conceptual guideline for understanding the deeper significance of processes of perspective taking that unfold in intergroup encounters, which may, in turn, lead to the strengthening of empathic bonds.
Specifically, the subtleties of the dynamic at play in dialogic intergroup encounters are shaped by the reciprocal influence and mutual awareness that develop in the intersubjective field-created in the coming together of people from two different groups representing two different historical perspectives. Thus, through a process of genuine listening to the other's story and pain in a facilitated, interactive process, the resonance that unfolds opens up the possibility for individual participants from each side to enter into the feeling state of the other. It is in this intersubjective engagement with the other's story that the emergence of shared empathy becomes possible.
We have referred to this unfolding process as "empathic repair" [69], a process of intersubjective repair that points to a deeper level of mutual recognition, one that occurs both intrapsychically and in the participants' external world through expressions of acknowledgment. This mutuality of a shared transformative moment is the fundamental moment of empathic repair and reciprocal recognition of the other's humanity that creates pathways to caring for the other as a fellow human being. In the following section, we describe apartheid South Africa as a concrete example of intergroup empathy failure and how dialogue processes initiated by the Truth and Reconciliation Commission opened up possibilities for empathic connection.

Apartheid South Africa as an example of intergroup empathic failure
Our interest in questions of empathic failures, and how empathic connections between groups might be enhanced, grew out of our work in the South African context with its history of intergenerational mistrust, hatred, and resentment born out of the violence of policies of apartheid. Both physical violence and the kind of violence that results from a lifetime of humiliation, passed down across generations of oppressed groups, create boundaries and deep divisions in relationships between self and others. This means that starting well before one's capacity even to make moral choices has been tested, one's sense of moral obligation toward others is rigidly channeled along lines of "us" versus "them," and the images of "them" depict a group that exists only as objectified others. Empathic failure operates under these conditions of a deep separation between racial groups as codified in apartheid laws.
As a strategy to find a sustainable way of dealing with these failures of empathy, and under Nelson Mandela's leadership, the Truth and Reconciliation Commission (TRC) was introduced in South Africa. What happened at the TRC may not be generalizable to all other post-conflict contexts. But what the work of the TRC has shown is that empathic connection between former adversaries can indeed be restored. An important condition for this to happen is the forging of dialogue and a vocabulary of compromise and tolerance, because the exercise of forming that vocabulary involves settling differences through the politics of contestation and compromise among people separated by laws based on intergroup hatred. South Africa's TRC, with its remarkable stories of forgiveness and healing, was a powerful illustration of how under certain conditions, instead of widening boundaries and deepening empathic failures, post-conflict dialogue processes can facilitate genuine connection in an intergroup context. Dialogue creates the possibility of setting the actions of "the other" in the broader framework of the political-ideological context that may have supported, and even directed, the hateful acts that excluded one from the moral and empathic obligations of "the other." Thus, the politics of abuse that were enshrined in the policies of an oppressive system such as apartheid could be acknowledged and confirmed in ways that opened up the possibility for the emergence of empathy between former adversaries.

Conclusions
Freud remarked that empathy "plays the largest part of our understanding of what is inherently foreign to our ego in other people" [97]. It is therefore not surprising that situations where empathy for another's distress is absent or reduced are very often also characterized by distrust, hatred, violence, discrimination, and even pleasure. The current chapter explored the complexities of intergroup empathic responding in an effort to gain a better understanding of the mechanisms that govern this process. First, we have shown that group membership affects all levels of empathic responding: experience sharing, perspective taking, and empathic concern. Second, we have pointed out the fluid nature of groups, in that both functional relations and differences in power and status may affect intergroup empathic responding at any given time. In addition, we have shown how various individual difference characteristics, notably racial identification and motivations to respond without prejudice, can influence intergroup empathic responding.
The intergroup landscape is not universally bleak, however, and empathic response differences across social categories are not inevitable. Drawing on our experiences in post-apartheid South Africa, we believe the intersubjective space that unfolds between former adversaries when coming together in facilitated dialogue with each other opens up rich possibilities for a shared empathy and mutual recognition of the other's humanity. These connections create new relational experiences that can help restore historical ruptures.