One of the breakthroughs in cognitive neuroscience is the discovery of an area of the brain that preferentially responds to human faces, known as the FFA [23, 24]. The FFA is thought to be responsible for extracting the physical information of different faces and is therefore considered the core brain area involved in facial recognition [25, 26, 27, 28, 29]. When individuals view faces of others that are either the same or different ethnicity, they tend to recognize faces of other ethnicities faster than those of the same ethnicity, and they are more accurate at recognizing faces of their own ethnicity than other ethnic targets which indicated that the recognition of same or other ethnicities may depend on different processes [30, 31]. To address this issue, the researchers conducted a classification task to examine whether there were differences in levels of activation within different brain regions when Chinese participants classified faces of the same ethnicity and other ethnicities. The study revealed that the right FFA (rFFA) exhibited higher activation during the recognition of same-ethnicity faces compared to other-ethnicity faces [32, 33, 34]. In addition, although participants recognize faces of other ethnicity faster, they can more accurately recognize faces of their own ethnicity than other ethnicity [35, 36]. This phenomenon is known as the other-ethnicity effect [37]. One explanation for this effect is that people conduct a more holistic analysis of faces of the same ethnicity. The rFFA was active when individuals performed an integrated analysis of their own ethnicity face, but not other ethnicities [34]. This differential response within the FFA to facial recognition of the same ethnicity and other ethnicities is more pronounced with increasing age [38]. These findings suggest that the rFFA plays a large role in the processing of faces of similar ethnicity than faces of other ethnicities.

Further investigation into the role of the FFA revealed its’ involvement in personalized processing of faces. Through the use of fMRI, researchers found that the specificity of FFA neural responses depended on the need for individualization. Specifically, when Korean participants performed tasks involving individualization, ethnicity categorization, and gender categorization, it was observed that the FFA exhibited activity specifically during individual face recognition and was inactive during the classification of common characteristics, such as ethnicity or gender [39]. This shows that the differential response of the FFA to faces of members of the same ethnicity and other ethnicity does not originate from its classification of faces of different group members, but from differentiated processing at the individual level [35, 40, 41]. Behavioral research has shown that recognition at the individual level and racial classification are two steps, and there is a trade-off between these two, which is an important sign of face-specific processing [42]. Another study involved having participants from China identify and classify faces of the same and other ethnicities. Specifically, they were presented with full-color face photos of 64 Caucasian and 64 Chinese young adults. The neural responses of the FFA and the occipital face area (OFA) were measured using fMRI. The study found that for faces of other ethnicities, both the FFA and OFA responded more to individual recognition than to classification, while for faces of the same ethnicity, the two brain areas were more sensitive to recognition and classification. The right superior temporal sulcus (STS) response was opposite to that of the FFA and OFA.

In addition, the individualized recognition process strengthened the functional connection between the FFA and the STS, while classification strengthened the functional connection between the OFA and the STS. The moderating effect of these two groups of functional connections was negatively correlated. That is, the recognition process attenuated the brain activity associated with the classification process, and vice versa. The results showed that in the core face-processing network, although the recognition and classification of faces of the same ethnicity and other ethnicities may activate the same neural network, the activation and functional connectivity of brain regions involved in different processing are affected by the type of face ethnicity and the situation [13, 43].

Human facial recognition is a dynamic process, and the connection between the core area and the extended area of the face system is key to facial recognition. Numerous white matter connections exist between the occipitotemporal-facial area, superior temporal sulcus, and the insula, suggesting that there may be direct anatomical connections from face-specific areas to frontal nodes which may be responsible for emotions associated with familiar faces, and the processing of the information provides the basis of the processing of emotional information associated to familiar faces. [44]. For example, an individual’s affective attitude affects the FFA. A positive attitude modulates blood oxygenation level-dependent signal (BOLD) in the right fusiform gyrus and the left inferior occipital gyrus when individuals process faces of the same and other ethnicities. When confronted with out-group members, individuals’ positive attitudes and experiences directly modulate the neural processing of out-group members [45, 46].

In summary, the FFA is mainly involved in the differential recognition process of faces of the same ethnicity and faces of other ethnicities. People are better at identifying the faces of members of their own ethnicity. This is mainly reflected in the fact that compared with individuals of different ethnicities, people can perform better-integrated analysis and reidentification processes for an individual face of their own ethnicity. At the same time, the FFA is dynamically connected to other brain regions during face processing, and therefore, intergroup face recognition could be affected by personal attitudes and prior experience.

The literature suggests that the amygdala strongly influences the function of the fusiform gyrus during face perception and that this influence is determined by an individual’s experience and the salience of face stimuli [47]. In healthy participants, increased activation of the amygdala has been associated with the other-ethnicity effect. For instance, when Caucasian adults are presented with faces of African Americans, there is an increase in amygdala activity. [48]. Clinical studies have also found enhanced amygdala responses to facial stimuli (four different gender faces) in individuals with asperger syndrome (AS) following oxytocin treatment [49]. Additionally, increased bilateral amygdala responses have been observed in Asian and Black/African American participants compared to Caucasian participants when exposed to Caucasian face stimuli, both in healthy participants and participants with major depressive disorder [50]. The amygdala also plays a crucial role in the recognition of emotion. In a classical facial emotion perception task, the right amygdala is thought to be involved in the autonomic arousal associated with facial emotion generation, while the left amygdala may be involved in decoding or evaluating facial expressions during early perceptual emotion processing [51]. Furthermore, as a component of facial information, age may influence the neural networks connected to the amygdala during the processing of faces with different ages (accompanied by emotional expressions) in participants of different age groups [52]. In face and house contrast experiments, strong bilateral activation of the amygdala and lateral activation of the fusiform gyrus have been observed, indicating that the role of the amygdala may include processing stimuli of neutral emotional faces in relation to social factors [53]. Moreover, increased bilateral activation of the amygdala is associated with delayed memory performance, therefore amygdala activity also seems to be related to facial memory [54].

The ACC is an important brain region involved in social cognition [67]. When people engage in social activities, the first impression is facial recognition. By testing individuals’ perception of disgusting faces, it was found that the insula-ACC neural circuit was largely involved in racially biased perception of disgust [68]. Disgust perception suggests that individuals are emotionally biased towards members of the “in-group” and biased against those of the out-group [68]. When participants made choices about their own team, brain regions such as the ACC, inferior frontal gyrus, and dorsolateral prefrontal cortex (DLPFC) were commonly recruited. These brain regions are known for their roles in cognitive control and social perception [69]. Research has shown that reversal learning of threatening or safe stereotypes through behavioral experiments involving fear conditioning can alter individuals’ prejudice against other group members, and corresponding activity in ACC [70]. This suggests that perceived bias can be reshaped by the social environment [71].

During social interaction, individuals extract multiple pieces of social information, such as ethnicity, gender, and behavioral habits, that influence their specific emotional responses. It was found that when Caucasian Americans see African Americans faces, it can trigger a negative emotional experience [72]. This is an example of intergroup stereotypes. Additionally, the study found that the ACC showed increased activity when social information contradicted the emotional information it carried, indicating that the ACC is sensitive to social information and stereotype inconsistency. The response of the ACC is plastic, and this plasticity could be diminished by increasing the exposure of the participant to another other social groups [14]. Furthermore, the dorsal ACC was activated when participants felt excluded from their racial group [73]. In addition, when individuals face potential threats from others of the same or different ethnicities, the brain responds differently [74]. The ACC showed increased activity when participants were faced with same-ethnicity members who did not pose potential threats [75]. In summary, in the process of social classification identification, the ACC is involved in the stereotyping of specific social groups, particularly when the stereotype of the group conflicts with the image of a single individual. This is consistent with the conflict detection function of the ACC [76].

The amygdala is vital for fear conditioning and fear memory retrieval [77]. It is also responsible for the detection and evaluation of potentially threatening stimuli [78, 79]. The primary role of the amygdala is coordinating cortical networks in response to emotional facial expressions [80]. When Caucasian Americans are confronted with unfamiliar African or Caucasian faces displaying negative emotions, the amygdala displays a heightened response [15]. However, although the activation of the amygdala in the perception of in-group members and out-group members is differentiated but, whether it is more active to faces of in-group members or out-group members remains controversial [81, 82]. In a study that examined 3T magnetic resonance imaging (MRI) scans of 20 participants of African descent and 22 participants of European descent while viewing the faces of African descent and European descent individuals, both groups demonstrated greater right amygdala activation, and the African ethnicity group had greater right amygdala activation for white faces [83]. In addition, a multi-voxel pattern analysis found that the implicit evaluation level of Japanese participants toward Korean participants could be predicted by the response of the left amygdala [84]. This indicated that the response of the amygdala may be related to more complex social and cultural background information. A study found that when participants were randomly assigned to a new group, they exhibited a range of perceptual, emotional, and behavioral in-group biases. When participants were assigned to a mixed-ethnicity team, the amygdala, fusiform gyrus, preorbital cortex, and dorsal striatum were more active when participants saw the new in-group member faces than when they saw the faces of new out-group members [85].

One explanation for the above findings is that cultural factors influence the neural basis of facial emotion perception [86]. The moderating effects of cultural factors on bilateral amygdala neural responses depend on in-group bias and collectivist values. Indeed, studies suggest that cultural factors can influence individuals’ perceptions of the facial emotions of in-group members [87]. Other studies have also shown that the amygdala is activated bilaterally by individuals’ perception of emotional expressions on faces of others of a different ethnicity, and this differential perception is influenced by cultural context [88]. Therefore, the amygdala responds to stereotypes or prejudices of out-group members, and the conditions under which the amygdala is activated depend on the subject’s social cognition regarding the target [89].

Another consideration is the differences in amygdala activity from the perspective of threats, such as the cultural context association between black males and potential threats involved in subjects’ identification of black males [90]. Using conditioning paradigms, researchers investigated neural activity underlying the development of aversive learning bias in European-decent participants when confronted with in-group (Caucasian) and out-group (Black) members. The study revealed that the amygdala and ACC played a crucial role in differentiating between Caucasian and Black faces during the acquisition and extinction of fear [91]. In addition, the role of the amygdala in group recognition also showed high plasticity. Amygdala sensitivity to ethnicity does not exist in childhood but emerges during adolescence [17]. Studies have shown that exposure to more members of out-of-ethnic groups during childhood is associated with reductions in amygdala responses to familiar black faces in adulthood [92]. This further demonstrates its strong relation to social factors.

Taken together, when engaging in intergroup identification, amygdala activation appears to be elicited by negative affect derived from the identification of social information. This differentiated emotional response is often related to social and cultural factors, such as stereotypes of members of other ethnic groups.

The PFC plays a key role in social cognition, with its’ various subregions having different functions. In the process of racial categorization, different faces of in-group and out-group members influence how individuals process emotions at the behavioral and neural level. Researchers found that the in-group member can exert a more powerful influence on social events (such as earthquake aftermath, attending a funeral, being in the hospital, etc.)-related emotional reactions than out-group members and during which the subregions of the PFC (e.g., ventromedial prefrontal cortex [VMPFC], dorsomedial prefrontal cortex [DMPFC], medial prefrontal cortex [MPFC]) were involved [16]. In children’s perception of different facial emotions (neutral, happy, angry), the MPFC exhibits a greater activation for happy faces [93]. In a facial recognition task using 3D digital face models, the dorsolateral prefrontal cortex (DLPFC) showed increased activity when emotional information conflicted with social stereotyping [14]. A multi-voxel pattern analysis revealed that the occipital neural networks might represent ethnicity in a fixed situation when making judgements about group membership. The early sensory perception of ethnicity was treated in a fixed manner, but the following-up integrated analysis was complex and depended on the decision-making context. In friendship judgements, the orbitofrontal cortex (OFC) preferentially encode ethnicity, while in personal trait judgements, the ventromedial prefrontal cortex (VMPFC) preferentially represents ethnicity [94].

Group bias is not only seen in racial recognition but also other types of social identification. Researchers found that individual’s cognition of yea-sayers (those with a lenient criterion) modulates several brain regions, including the caudate nucleus, DLPFC, and hippocampus. The caudate nucleus and DLPFC play a key role in personal cognition through repetitive feedback (receiving feedback multiple times about one’s actions or performance within a task, thereby aiding individuals in gradually adjusting and improving their behaviors, decisions, or cognitive biases) [95]. In addition, it was also found that right dorsolateral prefrontal cortex (rDLPFC) and right temporoparietal junction (rTPJ) neural synchronization among in-group members could be a candidate mechanism responsible for hostility between groups. Intragroup binding (in this study, “Intragroup binding” refers to the process of bringing individuals together through grouping) decreased rDLPFC activity and increased functional connectivity between the rDLPFC and rTPJ, while intragroup rDLPFC synchrony was positively correlated with inter-group hostility [96].

In summary, the PFC plays a key role in the process of ethnic classification, which could be influenced by the in-group members. And when identifying ethnicity within or outside the group, each subregion of the PFC assumes different roles.

Empathy enables humans to understand and share each other’s emotional experiences, creating emotional resonance between individuals and others, which is essential for successful social interactions [102, 103]. Observing others in pain activates corresponding pain pathways in the observer [11]. Research investigating the ACC and AI has shown that both brain regions play a crucial role in processing empathy [104]. In studies investigating pain empathy, when participants observed pain in the faces of chronic pain patients, the activation levels of the bilateral anterior insula, left anterior cingulate cortex, and left inferior parietal lobule in their brains were correlated with their estimations of the observed pain intensity [105].

When experiencing pain, altruists (a person who cares about others and helps them despite not gaining anything by doing this ) generate spontaneous empathic responses in the AI, anterior/middle cingulate cortex, and supplementary motor areas [106]. However, empathic perceptions of pain may be modulated by implicit racial bias [9]. One study asked subjects to rate the pain of in-group/out-group members in a minimal group paradigm experiment. They found that Chinese participants had stronger ACC responses when they rated the pain of Chinese participants than Western participants (Caucasian). The activity of the ACC was modulated by implicit racial biases while the right anterior insula (rAI) was modulated by an intragroup bias. Furthermore, it was found that the activity of the ACC was modulated by the activity of the rAI (in-group bias) under painful conditions, whereas the activity of the rAI was modulated by the activity of the ACC (racial bias) under painless conditioned stimuli [107]. When the activity of the AAC and AI regions were induced to respond to painful face stimuli, the empathic neural responses of the ACC were significantly reduced if subjects observed faces of other ethnicities that were in pain [108].

Other studies have found that individuals have greater activity in the ACC and the left anterior insula when observing members of the same ethnicity receiving painful stimulation [74]. In perception of personal pain, the participants’ perception of pain expressions by out-group members could also increase ACC-AI activity [109]. Subsequent studies have shown that increasing the subjects’ social interaction with members of other ethnicities significantly increases the response of the ACC to painful stimuli from members of other ethnicities [110]. This shows that the ACC is involved in the process of empathy processing, with a difference between in-group and out-group members, and it also shows that this socially biased ACC activity can be modulated through repeated exposure to out-group members.

Moreover, auditory perception can also induce empathy. The human voice is one of the main channels for social and emotional communication. Recent neuroimaging studies have shown that brain regions responding to the painful cry of others are similar to regions activated in the empathic processing of visual stimuli. Pain-related exclamations elicited increased activation in the superior and middle temporal gyri, left insula, secondary somatosensory cortices, thalamus, and right cerebellum, as well as reduced activity in the ACC. It should be noted that the emotional background used in the experiment may have contributed to this phenomenon [111].

In addition, social and contextual factors can also modulate empathic neural responses to the pain of others [112]. Research has shown that empathic responses to others’ pain are mediated by a variety of situational and individual factors as well as by social hierarchies. The researchers found that painful stimulation applied to inferior-status (In this study, participants were ranked through recognition tests, and those who ranked lower were referred to as having an “inferior-status”) targets induced greater activation in the AI and anterior medial cingulate cortex (aMCC). Conversely, these empathic brain activations were significantly attenuated in response to superior-status targets’ pain. These results revealed that sharing painful emotions with others was influenced by the relative position of the individual in the social hierarchy and that the empathic neural responses were biased towards lower-status individuals compared to higher-status individuals [113].

In summary, the perception of pain in others is one measure of empathy. Observing the pain responses of others activates sensory and emotional areas of the brain which are believed to be neural markers of empathy. Individuals have higher empathic reactions to in-group members than out-group members, and this difference can be attenuated by increasing social interaction with out-group members. Whether within groups or between groups, empathy activates the brain regions of the ACC and AI, indicating a functional connection between the ACC and AI during empathy, where they mutually coordinate and regulate the empathic process.

Research in the field of cognitive neuroscience has identified the mirror neuron system (MNS), located in the posterior region of the inferior frontal gyrus (IFG) and the anterior region of the inferior parietal lobule (IPL), as a potentially crucial neural substrate for empathy. The generation and perception of actions activate the same neural circuits, which is defined as mirror neuron activity [114]. Individuals are highly influenced by the emotions of others, and the most typical example is the experience of empathy towards pain. When observing an injury on someone else, the cingulate cortex and AI are activated [115]. Conversely, the ACC and AI are also activated when experiencing empathy towards others [109]. Currently, there is limited research on inter-group empathy within the MNS. In a study investigating inter-group empathy, researchers tested 178 participants who were presented with emotional and neutral facial expressions of both in-group and out-group members. The results indicated that in the MNS, participants exhibited higher activation in the left insula when viewing in-group members compared to out-group members. This suggests greater neural resonance (mirroring) for in-group facial emotional expressions [116].

However, it should be noted that while there may be some overlap between the neural networks of the MNS and empathy, they cannot be considered identical. A meta-analysis of studies investigating the involvement of the MNS in empathy and imitation found that the superior parietal lobule, inferior parietal lobule (IPL), dorsal premotor cortex, and frontal lobe were activated during empathy and imitation. However, the inferior frontal gyrus (IFG), which is a major component of the MNS, did not show activation [117]. Other studies have indicated a moderate correlation between empathy and activity within the MNS [118]. In summary, the aforementioned literature may indicate from an alternative perspective that in human-centered research, the design of behavioral experiments has limitations and may not accurately reflect the correspondence between behavior and brain regions. Therefore, the role of the MNS in empathy requires further investigation. However, there is a significant scarcity of research on the involvement of the mirror neuron system in pro-social behavior. These gaps need to be filled by researchers.

In pro-social behavior, rewarding in-group members increases activation in the ventral striatum and left medial orbitofrontal cortex [122]. In a study including both Chinese and American participants, participants were asked to complete a pro-social decision-making task,voluntarily giving their own money to others without expecting compensation. Researchers found increased activation of the ventral striatum in all participants and this was associated with donation within the group. Participants with a strong sense of group identity and Chinese participants showed higher activity in brain regions related to self-control (ventrolateral prefrontal cortex and ACC) and cognition (dorsomedial prefrontal cortex and TPJ) when contributing outside the group [123].

In summary, empathy is the main driving force for individuals to engage in pro-social behavior. Compared with out-group members, ACC-related reward circuit is more active when individuals help in-group members.

In studies investigating the differentiation of pro-social behaviors between groups, there is a relative scarcity of reports regarding the AI. In studies examining within group dynamics, it was found that the rAI plays a significant role in promoting altruism and pro-social behavior [124]. Other studies have revealed that activation of empathy-related brain regions, such as the AI and MPFC, is linked to increased pro-social behavior towards “victims” (in this study, the term “victim” refers to individuals who have experienced social exclusion). These findings suggest that neural responses associated with empathy during social experiences may foster spontaneous pro-social treatment of individuals in need [125]. Furthermore, in a study of reciprocity, the insula and ACC were positively correlated with reciprocity behavior [126].

In conclusion, the ACC and AI contain crucial networks for empathy and play a role in driving pro-social behavior. However, there is currently limited research on the differentiation of pro-social behaviors between groups in relation to these two brain regions. The differences in empathy between in-group and out-group members may explain the variations in pro-social behavior exhibited by individuals towards these different groups. This also partly accounts for the strong correlation between the brain regions involved in empathy and pro-social behavioral regulation. Research should focus on this aspect in future studies, as it holds significant importance in gaining a deeper understanding of the relationship between empathy and pro-social behavior. Furthermore, as mentioned in the section on empathy, the mirror neuron system is involved in the regulation of empathy.

Neural processing associated with empathy is strongly influenced by social identity. Human emotion plays an important role in decision-making [127]. Individuals feel more empathy for in-group members because the shared neural circuits that generate and perceive facial expressions, including the temporal pole region, the amygdala, left insula, left inferior frontal gyrus, inferior and middle temporal gyrus, right hippocampus, and para-hippocampus, showed higher activity when participants view other in-group members [128]. Moreover, recent research implicates that the amygdala is involved in costly altruism (donating a kidney to a stranger) [129]. In other studies, participants were asked to donate real money or hypothetical money. Real donation behavior is associated with increased activation of the amygdala, caudate nucleus, and rACC, during which the amygdala shows higher activity than hypothetical donation [130].

In addition, in the presence of vulnerable in-group members or other pain signal stimuli, coordinated transmission of information occurs from the amygdala to subcortical regions (including the striatum, bed nucleus of stria terminalis, hypothalamus, and periaqueductal grey), through which the amygdala can rapidly encode nonverbal pain signals that support the development of caring behaviors [131, 132, 133]. There is also research showing activity and functional connectivity of the amygdala and periaqueductal grey in response to care-evoking situations by examining rare altruists (such as organ donation). Furthermore, altruists were found to have increased sensitivity to pain and increased amygdala activity [134]. In summary, the amygdala seems to respond to modulate the feeling of empathy and, therefore, modulate altruistic behavior through connects to other brain regions.

Activation of the PFC influences pro-social behavioral choices, and in real moral decision-making, different subregions of the PFC determine whether subjects make selfish or pro-social moral choices [135]. In subregions of the PFC, pro-social behavior preferentially recruits the VMPFC and DLPFC. The results of functional connectivity analyses have suggested that the VMPFC and DLPFC may be involved in the assessment of pro-social behaviors and determining whether to engage in related behaviors, respectively [136]. Researchers employed an unsupervised dictator game to test subjects’ pro-social tendencies, using continuous theta burst stimulation (cTBS) to interfere with the DLPFC and dorsomedial prefrontal cortex (DMPFC) function of two prefrontal regions, and the results suggest a predominant tendency in humans to be pro-social when cognitive control is diminished [137].

The effects of the PFC on pro-social behavior could be modified. Compassion training can increase altruistic behavior associated with altered activation in brain regions involved in social cognition and emotion regulation, including the sub-parietal cortex, the DLPFC, and the connectivity of the DLPFC to the nucleus accumbens. Therefore, empathy can be cultivated through training, and strong altruistic behavior may result from the alteration of neurological regulation of understanding the suffering of others, executive and emotional control, and reward processes [138]. Interestingly, brain activity in the dorsal ACC and bilateral insula in response to pain was significantly reduced after individuals engaged in altruistic behavior. The decrease in pain-induced activation of the right insula was mediated by neural activity in the VMPFC. These results suggest that helping others could reduce the perception of unpleasantness [139].

In conclusion, the PFC activity affects decision-making in pro-social behavior. Through manipulation of the PFC, pro-social behavior can be increased or decreased. However, there are relatively few studies on the role of the PFC in pro-social behavior among groups. It seems that the plasticity of PFC responses for in-groups may also apply to members of out-groups. The neural mechanism underlying the difference in pro-social behavior among groups needs to be further explored.