Moral Faces: How Spontaneous Ideological Inferences from Facial Cues Influence Moral Judgments

Abstract

This research demonstrates that individuals make moral judgments based on ideological inferences from facial cues. Using racially homogeneous male faces, four studies showed that individuals infer ideological beliefs—and thus group membership (in- versus out-group)—from a novel face and that the fit between the inferred social identity and the perceiver’s own social identity leads to judgments of moral superiority. Further evidence shows that a salient social identity moderates moral evaluation. These results reflect the automaticity of social categorization, which contributes to moral judgments of a person.

1. Introduction

Morality indicates what is the “right” and the “wrong” way to behave [1]. Acting in ways that are considered ‘moral’ by the group secures inclusion and causes one to obtain respect from others who are important to the self [2,3]. Identifying who is moral is essential in group living, and people spontaneously judge others’ morality, as it is a critical feature in survival (e.g., [4]). Morality is regarded as being at the heart of in-group identity, positive group esteem, and social action and is regarded as playing a central role in the quality of relations between groups [5,6].

Importantly, what people see as moral can shift, depending on norms and features of the group to which the person belongs; the socially shared idea of morality within a group anchors the person’s moral compass [7]. Despite this importance of group affiliation in understanding consumer morality, studies to date have paid scant attention to understanding the role of group affiliation in morality judgment. Previous research mostly approached morality as an individual phenomenon (for reviews, see [1,8]) and focused on individuals’ use of specific notions of right and wrong or how they agree and disagree about what constitutes being moral [9]. Consequently, groups have been given scant attention regarding morality, and our knowledge of the role of group affiliation in moral judgments is still lacking. The goal of this current research is to fill this gap.

This current research investigates the way in which individuals judge the morality of others based on inferred social identity from facial cues. This study specifically tests how individuals spontaneously infer other individuals’ social identity based on facial cues and how inferred group membership is subsequently used for making judgments of morality, such that inferred in-group membership leads to perceived superior morality, and incongruent group membership leads to perceived inferior morality. To this end, this study tests individuals’ spontaneous inferences of others’ social affiliations (in- versus out-group) based on facial features.

In prior research, social categories were explicitly manipulated by experimentally assigning group membership (e.g., own versus another team). However, most people’s affiliations are not visually detectable. When there are no explicit cues for categorization, one may use alternative cues such as perceptual similarity to a known person [10,11]. Furthermore, a person could have multiple social identities that can be categorized as multiple categories, e.g., simultaneous political or religious affiliations. These identities could be used as important alternative coalition cues. This current research proposes the possibility of spontaneous group inferences, demonstrating that when a novel face activates a mental representation of a known person, a related social identity can be used for categorization of the novel face. Subsequently, morality is judged based on group membership—whether it is in- or out-group.

This article proposes and demonstrates that people tend to judge others’ morality based on inferred social affiliation from facial cues. Four studies show that a person’s perceptual similarity to a known face activates a mental representation of social categories, influencing moral evaluations of the target person. Furthermore, the relationship between faces and moral judgments is moderated by the salience of social identity. When the target face is inferred as in-group (a social identity that is congruent to the evaluator’s own identity), the person is subject to more favorable moral evaluations, whereas if the inferred social affiliation is incongruent with that of the evaluator, the target person will be subject to less favorable moral evaluations. However, if the inferred social identity is not consistent with the salient social identity, the effect will be null.

2. Inferences of Social Affiliation on Moral Judgments

Prior researchers mostly considered morality as an intra-individual characteristic and counted on references to indicate what a moral person is like (for review, see [12]). Those researchers studied how individuals make inferences regarding their own or others’ morality, focusing on inferences of trustworthiness [13]. They found that people are so adept at inferring others’ trustworthiness that they do this very quickly and spontaneously in interactions, based on very little information (e.g., [14]). However, trustworthiness is not the only determinant of a person’s morality. Morality always implies the presence of a group that shares a particular view of what is moral [15,16,17] as the person’s beliefs about what is morally right or wrong are often socially shared ideas (e.g., cooperation, justice, caring).

Aside from personality traits, group information is known to be automatically inferred from human faces. With only a glance, people automatically refer to and encode social category information and instantly categorize the face on multiple dimensions such as race, gender, and age [18,19]. This automatic social categorization influences the way the person is mentally represented. Indeed, recent neuroimaging research has consistently demonstrated that simple social categorization generates favorable effects toward in-group faces. For example, experimentally manipulated in-group faces preferentially stimulate the amygdala and the fusiform face area (FFA) more than out-group faces [20,21], with this even overriding racial bias (Van Bavel & Cunningham, 2009). These findings show that any salient social identities can drive social perception and evaluation [22,23].

A substantial body of research shows that perceptual similarity to a known face leads individuals to evaluate a novel face in a similar fashion as the known face. For example, morphed faces with 50% of the characteristics of a participant’s romantic partner were rated more favorably [24]. When a novel face was manipulated to look similar to a known person, assessment of the novel face assimilated to the evaluation of the known face [25,26]. For example, Kraus and Chen (2010) [27] found that an unknown face that has facial features perceptually similar to those of a familiar person is evaluated similarly to the known person. Furthermore, people expect unknown individuals to behave similarly to known individuals when their faces more closely resemble the known face [28,29]. As such, when evaluating novel faces, a related mental representation is activated and generates representation that leads to consistent inferences [30,31]. This exemplar-based model of social judgment implies that representations of specific individuals influence the judgments of similar persons through activation of a related mental representation [32]. As such, perceptual similarity activates associated mental representations of a known face and influences the social categorization of the face.

Once categorization occurs, individuals tend to show in-group favoritism and negative out-group attitudes [33,34,35]. Stephan, Renfro and Davis (2008) [36] suggested that an out-group imposes symbolic threats that are based on stereotypes of that out-group that suggest an opposition to the in-group’s cherished values and beliefs. Thus, symbolic threats are often moral in content. Cottrell and Neuberg (2005) [37] also suggested a general model of threat in prejudice that focuses on moral threats. They suggested that out-groups can be seen as threatening the in-group’s values in a way that might morally contaminate the in-group. This threat promotes the maintenance of the in-group’s moral values by distancing them from the out-group. In support of this idea, recent studies have demonstrated that prejudice against social groups was derived from perceived similarities and dissimilarities in political ideologies [38]. Different groups of people support different ideologies because they have distinct motivational goals, and various ideologies demonstrate unique patterns of moral concerns. Based on this evidence, it is surmised that when the target person’s social affiliation is inferred to be with the in-group (congruent with the evaluator’s affiliation), the person will be judged as having moral superiority. On the other hand, when the target person is inferred as part of an out-group, the person is more likely to be judged as inferior in morality.

H1. 

Individuals infer the social identity(ies) of novel faces based on perceptual similarity to known faces.

Once social categorization occurs automatically, one’s processing of it will differ [19]. An in-group bias in memory influences early perceptual processing such that in-group faces are processed more in-depth and with superior memory retention [39]. Additionally, in-group faces are processed in a brain area linked to self-referential processing [40]. This means that conservatives are more motivated to understand the mentalities of other conservatives rather than liberals, and vice versa. Thus, it is probable that people are more likely to correctly identify in-group ideological values from faces that resemble those of the in-group compared to other countenances, because personal ideological memories are more readily accessible. Therefore, minimum implicit perceptual cues are sufficient to activate mental representations of in-group categorized faces.

When the identity is salient, individuals make sense of the world using an identity-consistent mindset [41]. This leads to heightened sensitivity to identity-relevant stimuli, shape encoding, and retrieval processes [42]. This also leads to more positive evaluations of identity-consistent stimuli and motivates people to act in ways consistent with the social identity [41,43,44]). This occurs while individuals avoid objects and behaviors that appear to be inconsistent with the perceived social identity [45]. Thus, it is predicted that evaluation of moral superiority related to the novel face is dependent on which identity is salient at the time of evaluation among the perceiver’s multiple social identities.

H2. 

The inferred social identity mediates the relationship between facial familiarity and moral evaluation.

H3. 

Perceivers will make more favorable moral evaluations of target faces that they perceive as sharing a social identity (i.e., as an ingroup member) than of target faces perceived as out-group members.

These hypotheses were tested in four studies. Study 1A provides initial evidence in support of the hypothesis that ideological beliefs can be inferred from a novel face based on its perceptual similarity to a known person. Study 1B further provides evidence of automatic social categorization (group affiliation) based on inferred ideology. This study shows that a match between inferred social identity from novel faces and the perceiver’s own identity (in-group) leads to an evaluation of moral superiority. Study 2 further supports the proposed mechanism by showing that when explicit ideological beliefs do not match with the inferred social identity from the face, an evaluation of moral superiority does not occur. Finally, Study 3 shows the moderating role of different types of salient social identities. When salient social category appears to be incongruent with the inferred ideology, inferred group affiliation does not result in an evaluation of moral superiority. These four studies consistently support the premise that moral judgments can be made spontaneously by inferred social affiliation through perceptual similarity to a known person.

3. Study 1

Studies 1A and 1B tested the basic relationship between perceptual similarity, social affiliation based on perceived ideological beliefs, and moral evaluation. To this end, the study utilized a composite face of a well-known presidential candidate from the conservative party, Donald Trump. Study 1A tested whether perceptual similarity influences perception of the social identity of novel faces, and Study 1B further tested whether inferred social identity influences moral evaluation.

3.1. Methods

Participants participated in the form of an online panel (103 in Study 1A and 186 in Study 1B) and took part for payment. Study 1A had a 2 (face: Trump 35% versus control 35%) × 2 (perceived ideology: liberal versus conservative) mixed design. Study 1B had a 2 (face: Trump 35% versus control 35%) × 2 (political affiliation: liberal versus conservative) × 2 (perceived ideological beliefs: liberal versus conservative) mixed design, with the latter being within subjects. Two face stimuli were created using Abrosoft FantaMorph software by morphing 65% of an unfamiliar stock model face with 35% of either Donald Trump’s face or 35% of another unfamiliar stock model face as a control face (see Figure 1). A pretest (n = 147) confirmed that the two composite faces were perceived similarly in dimensions such as attractiveness, intelligence, and age (all ts < 1).

In study 1A, participants were randomly assigned to one of two conditions (Trump 35% versus control 35%). After being presented with one of two composite faces, all participants were instructed to form an impression about the pictured person before measuring the perceived social identity of the person. Instead of directly asking the group membership of the pictured person, participants were asked to rate perceived ideological beliefs of the person on 12 items (4 liberal and 8 conservative beliefs; Haidt and Graham, 2007; Graham et al. 2012) on 7-point scales ranging from 1 (not at all) to 7 (extremely). Examples of traditionally conservative beliefs are “men and women have different roles” and “immigrants are a danger to society” (α = 0.84), while examples of liberal beliefs are “everyone needs to be treated fairly” and “justice is the most important requirement for a society” (α = 0.81).

For Study 1B, after randomly being presented with one of two composite faces, participants were first asked to rate the strength of the pictured person’s morality using a 3-item measure (e.g., morally superior, ethical, self-righteous; α = 0.91) on a 7-point scale (1 = not at all, 7 = extremely). They then rated perceived ideological beliefs using the items in Study 1A. Finally, participants’ political attitudes were collected (1 = very liberal, 7 = very conservative) along with demographic information. This score was dummy-coded into liberal versus conservative political attitudes. Participants underwent debriefing after providing their ratings. They were first asked, in an open-ended format, if the face reminded them of anyone they knew and were then asked to report what they thought the purpose of this survey might be. No participants recognized Trump from the Trump composite face or reported the purpose of this study correctly.

3.2. Results

Study 1A: Perceived Ideological Beliefs. A 2 face (between) × 2 perceived identity (within) mixed ANOVA revealed a significant interaction (F (1, 103) = 11.93, p < 0.001) in that participants rated Trump’s composite face as supporting conservative ideological values more (M = 4.26) than the control face (M = 3.96, F (1, 103) = 6.08, p < 0.05, d = 0.34) and as supporting the egalitarian ideology less (M = 4.28) than the control face (M = 4.69, F (1, 103) = 10.87, p < 0.001, d = 0.45). No differences were found between the evaluations of attractiveness, age, or intelligence (all ts < 1).

Study 1B: Perceived Ideological Beliefs. A 2 face × 2 political affiliations × 2 perceived ideological belief (within) mixed ANOVA revealed a significant three-way interaction (F (1, 182) = 4.43, p < 0.05). Conservatives perceived the Trump composite face as having more conservative ideological values (M = 4.52, SD = 1.95) than that of the control face (M = 4.17, SD = 1.68, F (1, 182) = 3.39, p = 0.07, d = 0.2), but no discernable difference was found regarding perceived liberal ideological beliefs (M = 4.41, M = 4.67, p = 0.3). However, liberals did not rate the two faces differently regarding either conservative ideology (M_Trump35% = 4.58; M_control35% = 4.40, p = 0.4) or liberal ideology (M_Trump35% = 3.99; M_control35% = 4.23, p = 0.1).

Moral Evaluations. A 2 (face) × 2 (political identity) between-subject ANOVA on the moral evaluation score showed that conservatives rated the Trump composite face as morally stronger (M = 4.78, SD = 1.57) than the control face (M = 4.45, SD = 1.35, F (1, 182) = 4.51, p < 0.05, d = 0.22). However, liberals did not rate the faces differently (M_Trump35% = 4.56, M_control35% = 4.57, F < 1).

Moderated Mediation Analysis. To better evaluate the underlying mechanism of the observed effect, a moderated mediation analysis was conducted (Process model 8; see Figure 2). The analyses used 5000 bootstrap samples and revealed a significant moderated mediation effect (95% CI: −0.23 and −0.02), suggesting that the perceived social identity mediated the moral evaluations of the faces and that this effect was moderated by the participants’ personal social identities (political affiliation). That is, the perceived ideology influenced moral evaluations only when the composite face was evaluated by in-groups (conservatives; 95% CI: −0.59 and −0.08) and not when evaluated by out-groups (liberals; 95% CI: −0.23 and 0.22).

Studies 1A and 1B demonstrated that participants with a conservative political identity (versus liberal identity) were more likely to identify the Trump composite face (versus the control face) as being conservative presumably because Trump-associated memories are more readily accessible to them than for liberals; thus, minimum implicit perceptual cues are sufficient to activate mental representations of Trump. Based on the perceived social identity, the novel face is categorized as in-group versus out-group, and this social category further influences the evaluation of the face. These findings are consistent with existing literature showing that in-group faces have superior memory retention and the in-group bias in memory influences early perceptual processing such that in-group faces are processed more in-depth [39,40].

Although Studies 1A and 1B successfully supported predictions, there are still some important remaining questions. First, in Studies 1A and 1B, participants were forced to rate the presented faces in terms of their perceived beliefs. Although this score captures perceived social identity of the presented faces, it reflects participants’ conscious judgments of the presented person’s social identity rather than automatic inferences of it. The next study endeavored to remedy this issue. Instead of asking participants to rate the presented person’s beliefs, they were presented with explicit ideological statements supposedly made by the presented person before being asked to report the degree to which they believe in this person’s morality. That is, if the composite face activates mental representation of the known person (Trump in this case) when the explicit ideological statement matches with the inferred ideological value from the face (conservative beliefs), then enhanced processing fluency will lead them to more truthful evaluations (Lee and Aaker, 2004). On the other hand, if the perceptual similarity of the representation does not occur, then the explicit ideological statements that match the perceiver’s own ideological beliefs will be rated as morally superior regardless of the facial stimulus presented. The next study tested this prediction.

4. Study 2

The goal of Study 2 was to probe further evidence of perceptual similarity of a novel face activating mental representation of a known person. To this end, composite faces were presented with explicated ideological beliefs supposedly stated by the pictured person. The prediction was that if perceptual similarity indeed activates a mental representation of the known person, then a positive evaluation would be observed only when the explicit ideological statements were compatible with the activated mental representation. On the other hand, even if the pictured face explicitly stated the ideological beliefs, when the explicit statements were not compatible with the activated mental representation, then no positive moral evaluation would be observed.

4.1. Methods

One hundred and thirty-five paid participants from the online panel (70 Republicans) participated in the study. To control for any possible cross-race effects, only white Americans were invited to this study. This study was a 2 (face) × 2 (political affiliation) × 2 (perceived ideological beliefs) mixed design, with the latter being within-subjects. After being randomly presented with one of the two composite faces used in study 1 (Trump 35% versus control 35%), participants were subsequently presented with a series of six statements made by the person pictured, expressing either a liberal ideology (e.g., “every human being has the right to receive universal healthcare”) or a conservative ideology (e.g., “I do not support same-sex marriage”). For each statement paired with the composite face, participants rated the moral superiority of the pictured person using 3 items used in study 1 (α_liberal = 0.93; α_conservative = 0.91). Participants were then asked to form an impression about the person and rate the face in terms of the perceived ideology according to the 12 items used in the study. Finally, participants reported their demographic information, including political attitude, before undergoing debriefing.

4.2. Results

Perceived Ideological Beliefs. A 2 (face) × 2 (political affiliation) × 2 (ideological statements: within) mixed ANOVA revealed a significant interaction between face and ideological statements (liberal versus conservative statements) on perceived ideological beliefs (F (1, 131) = 5.44, p < 0.05) and a three-way interaction between face, political affiliation, and ideological statements on perceived ideological beliefs (F (1, 131) = 3.96, p < 0.05). Conservatives rated the Trump composite face as being more supportive of conservative ideological beliefs (M = 3.67) compared to the control face (M = 3.28, F (1, 131) = 4.98, p < 0.05). Furthermore, conservatives rated the Trump composite face as less supportive of liberal ideological beliefs (M = 4.60) compared to the control face (M = 5.28, F (1, 131) = 6.59, p < 0.01). However, liberals did not rate the Trump composite face and the control face differently in terms of conservative ideological beliefs (M_Trump35% = 3.62, M_control35% = 3.48, F < 1) or liberal ideological beliefs (M_Trump35% = 4.99, M_control35% = 4.93, F < 1). These results confirm that perceptual similarity to a known face triggers related mental representation of a known person only for an in-group known face and not for the out-group known face.

Moral Evaluations. Although a 2 (face) × 2 (political affiliation) × 2 (ideological statements: within) mixed ANOVA revealed an insignificant three-way interaction (F (1, 131) = 0.13, p = 0.72), there was a significant two-way interaction between political affiliation and ideological statements (F (1, 131) = 21.27, p < 0.001) and a marginally significant interaction between face and ideological statements (F (1, 131) = 3.53, p = 0.06) on moral evaluations. In particular, when conservative ideological statements were presented as stated by the Trump composite face, and when their own ideological beliefs were explicitly presented as being stated by the pictured person, conservatives rated the Trump composite face as being morally superior (M = 4.56) compared to the same statements paired with the control face (M = 4.26, F (1, 131) = 3.32, p = 0.07). This is possibly because of the compatibility between explicit and implicit representation.

On the other hand, when liberal ideological statements were explicitly stated, liberals rated the Trump composite face as less moral (M_Trump35% = 4.06) compared to the control face (M_control35% = 4.44, F (1, 131) = 4.86, p < 0.05), possibly because the implicit mental representation of Trump does not match with the explicit ideological statements. These results indicate that explicitly presented ideological statements that are supposedly stated by the pictured person might activate the person’s identity (politician) and influence the subsequent evaluation, even for the out-group face for liberals.

Furthermore, when the explicitly presented ideological beliefs were that of the out-group, participants’ moral evaluation was not influenced by the face. In particular, when liberal ideological statements were presented as being made by the focal person, no discernable difference was found between the evaluations of morality for both faces (composite face and control face) by conservatives (M_control35% = 4.52 versus M_Trump35% = 4.66, F < 1). Similarly, liberals did not rate morality for the two faces differently when conservative ideological statements were presented as being made by the same faces (M_control35% = 4.63 versus M_Trump35% = 4.49, F < 1). These results support the prediction that perceptual similarity to a known person activates associated mental representation involving congruency between implicit and explicit ideological statements. This congruency determines evaluations of moral superiority.

Moderated Mediation Analysis. A moderated mediation analysis using 5000 bootstrap samples was conducted [46]. It revealed a significant moderated mediation effect (95% CI: 0.11 and 0.36), suggesting that the perceived social identity mediated the effect of the perceptual similarity of faces on moral evaluations of the faces. This effect was moderated by the participants’ personal social identities (political affiliation). That is, the perceived conservative ideology influenced moral evaluations only when the composite face was evaluated by in-groups (conservatives; 95% CI: 0.01 and 0.25) and not when evaluated by out-groups (liberals; 95% CI: −0.14 and 0.004).

4.3. Discussion

The data from Study 2 revealed that conservatives’ assumptions of moral superiority for the Trump-morphed face appeared only when it was paired with conservative ideological statements, not when the same face was paired with liberal ideological statements. These results indicate that perceptual similarity indeed activates mental representations of the known face. Thus, conservative ideological beliefs were indeed inferred from the Trump-morphed face. This inferred ideology influences the social categorization and judgment of the face based on the social category. The findings from Study 2 confirms important evidence that perceptual similarity of a novel face to a known face activates a mental representation of the known face.

The one remaining question is whether this similarity-based inference actually leads to automatic social categorization. Social identity theory suggests that individuals possess a collection of discrete social identities that vary in their salience. Although a person’s global sense of self is the collection of multiple social identities, the most salient social identity at a given time leads to categorizing others as in-group versus out-group [10]. Thus, when salient social categorization is not congruent with the face being evaluated, it should not be categorized as in-group and, thus, is not positively evaluated.

During the presidential campaign and after being elected as the U.S. President, Donald Trump’s primary social identity was associated with the conservative party. However, many conservative people are skeptical about his Christian values, as Donald Trump’s stated views and beliefs do not align with traditional Christian thought [47]. Thus, the hypothesis was that if their Christian identity becomes salient, participants will not evaluate the Trump composite face as morally superior because it will not be categorized as in-group. However, if their Republican Party identity becomes salient, the same face will be evaluated as highly moral. Study 3 tested this possibility.

5. Study 3

The purpose of Study 3 is to test whether the found effect of perceptual similarity on moral evaluation is due to an automatic categorization of faces based on social identity. To this end, Study 3 manipulated the saliency of social identity with tasks pertaining to facial evaluation. In particular, participants’ social identity was activated either in terms of political party or religious beliefs. The prediction was that this type of salient social identity will moderate the effect such that conservative identity will moralize the Trump composite face, whereas Christian identity will lead to demoralization of the same face.

5.1. Methods

One hundred and seven white Americans participated online for payment. This study had a 2 (face) × 2 (social identity) between-subject design. Participants were randomly assigned to one of two conditions: political identity and religious identity. Participants in the political identity condition were presented with images of political parties, e.g., an elephant and a donkey fighting each other. Those in the religious identity activation condition were presented with images of a church, cross, and prayer (see Appendix A Figure A1). All participants were then asked to create a story based on the presented image. To ensure saliency of the Christian identity, participants were asked if they go to church at least once every four weeks. Upon completing the social identity activation task, all the participants were randomly presented with one of two facial stimuli: Trump 35% versus control 35% (Figure 3). They were then asked to evaluate the presented face on the same scale used in Study 1B.

5.2. Results

A regression model was estimated with moral rating of the faces as a dependent variable with face type (Trump 35% = 1, control 35% = 0), salient social identity (political identity = 1, religious identity = 0), and political affiliation (continuous) as predictors. The regression analysis revealed a significant three-way interaction between face, salient social identity, and political affiliation (β = 0.64, t = 2.91, p < 0.005). This study probed conditional effects of race on moral evaluation with the values of each moderator (political affiliation and social identity) and replicated the effects from Studies 1 and 2. Conservatives (1 SD above the mean) with a salient political identity rated the Trump composite face as being morally superior (β = 0.69, t = 1.81, p < 0.07). However, when the perceivers’ salient social identity did not match with that of the presented face, the person was not categorized as in-group. Consistent with this prediction, it was found that the Trump composite face was not rated as morally good when religious identity was activated. Specifically, conservatives (1 SD above the mean) with a salient Christian identity rated the Trump composite face as being morally inferior (β = −0.95, t = −2.94, p < 0.005). Similarly, participants with no party affiliation but with a salient Christian identity rated the Trump composite face as less moral (β = −0.52, t = −2.03, p < 0.05). These results supported the prediction that automatic social categorization by congruency between the salient social identity of perceivers and those being perceived leads to an evaluation of moral superiority.

6. General Discussion

Four studies collectively suggest that inferences of ideological beliefs can occur based on perceptual similarity to a known person, and a fit between the perceived social identity and the perceiver’s own identity leads to positive moral judgments. This research contributes to a burgeoning field of consumer research on morality. Most research on morality has mostly focused on how people make moral decisions (e.g., recycling). More recent consumer researchers have examined this topic in contexts of moral foundations related to consumers’ political orientation, i.e., how liberals and conservatives make different moral judgments. For example, researchers showed that persuasive appeal that is congruent to consumers’ political ideology increases sustainable consumer behaviors (e.g., [48]). Another study showed that when the moral foundations of a charity are aligned with the donor’s political identity, donations increase [49].

This current research contributes to this body of research by demonstrating the role of group affiliation on the moral judgments of persons. In particular, this study demonstrated that a person’s ideology—a collection of beliefs that determines what is considered the correct way of dealing with issues—can be inferred from facial cues and utilized for the judgment of morality.

This current research contributes to the literature on social categorization. Historically, social categorization research mostly focused on the role of primary category information (e.g., race, gender). Recently, many social neuroscience researchers have started to utilize the evolutionary perspectives of social categorization, suggesting that the brain is susceptible to any coalitional cues that signal social alliance and demonstrating that race might be simply a proxy for such cues. This idea has been supported by studies showing that assigned group memberships are sufficient to generate a social category effect by using mixed-race faces.

However, those studies mostly manipulated the category by explicitly assigning category membership, leaving unexplored the possibility that alternative facial information can be utilized as a proxy for coalitional cues. This current research tests one of these alternative cues, namely perceptual similarity to a known face. By using racially homogeneous male faces without explicitly assigned group membership, this current research demonstrates that social categorization is affected by a function of perceptual similarity to a known face.

This research also contributes to the body of literature on facial similarity or familiarity. A substantial body of research shows that individuals consider familiar faces in all possible positive aspects of the spectrum, such as being attractive, trustworthy, and suitable for an elected position [50,51,52,53]. This positive effect of familiarity on a person’s judgment suggests that familiar faces should be evaluated as morally good [54,55,56]. However, our data indicate that facial familiarity does not always lead to assumptions of moral superiority; rather, social affiliation fit predicts positive moral judgments. The previous literature on facial familiarity mostly considered the affective generalization effects of familiar faces in that unknown faces elicit the same affective evaluations as the known faces they resembled [24,57]. The current research contributes to this literature by demonstrating that not only affective evaluations (positive versus negative) but also a person’s representations (social affiliation) are likely to be activated based on facial similarity. Thus, even if a novel face looks familiar, it may not always be evaluated in a positive light. The positivity may be dependent on the currently salient social affiliation and mental representation that the novel face activates.

Prior research has shown that people may infer others’ objective political traits above chance level on the basis of their looks alone. These studies argue that political attitudes might be linked to detectable personality attributes such as competence and power suggestive of conservative behavior [58,59,60]. Thus, future research should explore whether individuals with conservative political attitudes are more likely to positively evaluate faces with features associated with competence and power.

Another future avenue of research is to investigate if and how race or gender influences similarity-based moral judgments. The current studies tested this effect using white male faces that are free from prejudicial evaluations. Future research should explore whether automatic prejudicial responses to faces of various races interact with the current findings and how the inhibition or suppression of initial evaluations and rebounded prejudice interact with the similarity-based moral judgments.

Finally, prior research demonstrated that certain contexts, such as resource scarcity, often shift visual perceptions of race such that out-group faces are perceived more stereotypically. For example, studies found that white people tend to perceive racially ambiguous faces as being more typically black under economic resource scarcity [61,62]. If this were the case, would resource scarcity also modulate social categorization based on implicit ideological cues? For example, under resource scarcity, might the found positive effect toward novel faces that are perceptually similar to the in-group not persist? What if the source of resource scarcity stems from inter-group competition? Would a greater number of in-group members be beneficial for winning a competition? In this case, would people broaden their perceptual bandwidth for in-group boundaries? These questions could pave a valuable path for future researchers.