Perception of Emergent Leaders’ Faces and Evolution of Social Cheating: Cross-Cultural Experiments

The aim of the present study was to investigate the role of facial appearance as a source of information about individual qualities, and its’ relation to the phenomenon of cheating in the context of human leadership. The problem of free-riding has been the most comprehensively studied within the framework of cooperative behaviour (Ensminger & Henrich, 2014; Fehr & Gächter, 2000; Heckathorn, 1989; Ledyard, 1994; Marlowe, 2009; Yang et al., 2018). From an evolutionary perspective, both cheating and free-riding constitute similar behavioural constructs, which are ultimately aimed at achieving individual rather than social benefits. However, the phenomenon of cheating implies not just “consumer” behaviour, but rather braking social norms and own promises. Although it is usually assumed that leaders should act in the interests of a group and facilitate coordination between group members for achieving common good (Van Vugt & von Rueden, 2020), a number of empirical studies suggest that cheating may also occur in the context of human leadership, where it is closely related to specific personality features of emergent leaders (Bereczkei, 2018; O'Reilly & Doerr, 2020; Rostovtseva et al., 2022). For instance, individuals who score high on Macheviallianism are perceived by others as good and charismatic leaders (Deluga, 2001; Jones & Paulhus, 2009), they can occupy positions of top managers, especially in young companies, whereas in a long-term perspective such leaders have a detrimental effect on the careers and well-being of their wards (Bereczkei, 2018; Volmer et al., 2016).

The present study builds on the previous experiment, in which emergent leadership in the context of male group cooperation was investigated (Rostovtseva et al., 2022). In the previous study, based on the face-to-face iterated Public Goods Game (iPGG) (conducted in initially leaderless groups of four male strangers), three types of behaviour related to emergent leadership were revealed: 60% of the male participants were classified as non-leaders, 30% expressed potential for prosocial leadership, and 10% appeared to be leaders-cheaters (for details see Methods). Independent analysis revealed that each of these behavioural types had a set of distinguishing personality, communicative, and cooperative features. Both prosocial leaders and leaders-cheaters were equally followed by other group members (on average in 88% of cases), indicating successful communicative strategies of leaders-cheaters, which allowed them to evoke trust in others despite actual deceptive intentions (Rostovtseva et al., 2022). This experiment was conducted in a specific population – among young Buryats of Southern Siberia (traditionally nomadic pastoralists of Mongolian origin). Although the obtained results demonstrated that some communicative features of leaders-cheaters were perceived by others as indicators of trustworthiness (see also Okubo et al., 2012, 2013, 2017, 2018), it was not clear whether such individual features represent only behavioural components, or they may also be imbedded into static individual morphology. To make this point clear, we have conducted this follow-up study.

The present study was designed to extend our previous findings. Here we address four new questions: (1) do neutral faces of individuals defined as non-leaders, prosocial leaders, and leaders-cheaters have common morphometric features (per type), which can be perceived by others in a consistent way?; (2) if so, which physical and behavioral traits are attributed to the faces of individuals belonging to each of these behavioural types?; (3) are there sex and population differences in facial perception of non-leaders, prosocial leaders, and leaders-cheaters?; (4) is there an impact of social environment on perception of faces of different population (racial) origin?

To answer the new research questions, we have created three composite portraits of Buryat men with different propensities for leadership and leadership styles based on individual photographs obtained in the prior experiment. The study was restricted only to male portraits since leadership qualities in the previous study were measured only in men, and generally leadership is of greater concern to men compared to women, which is deeply rooted in human phylogeny (Smith et al., 2020; Van Vugt & von Rueden, 2020; von Rueden et al., 2018). Moreover, according to the findings by other authors, behavioural traits are better recognized in male, but not female faces (Tognetti et al., 2013). The three composite portraits were then presented to independent raters (Buryats and Russians of both sexes) using an online form. Russian raters were originating from areas with different levels of urbanization and familiarity with Mongolian appearance in everyday life. The composite portraits were scored on a number of physical and behavioural characteristics (masculine, physically strong, healthy, attractive, trustworthy, generous, leader, competitive, and dominant), without providing participants with information about behaviour of individuals constituting each of the stimulus portraits. Intuitively, it may be expected that those Russian raters who were more exposed to perception of Mongolian appearance in everyday life, living side by side with Buryats (Russians from Buryatia), will be more accurate in judging personal traits based on facial appearance (for instance, such tendencies have been previously described for accuracy of emotion recognition: Elfenbein & Ambady, 2002).

The physical and behavioural traits, used for facial judgments, were selected relying on the large body of previous findings. It has been recently shown that body muscularity and facial masculinity are positively related to male cooperativeness in the context of inter-group competition (Muñoz-Reyes et al., 2020). Physical strength and perceived masculinity are correlated with each other (Windhager et al., 2011). Previous findings by other authors suggest that leaders with masculine traits (e.g. low-pitched voices) are preferred over others since they are perceived as physically stronger (Klofstad et al., 2015). Preference for masculine, physically strong, and dominant leaders has been reported to be especially pronounced under condition of inter-group conflicts (Laustsen & Petersen, 2016, 2017; Little et al., 2007). It has been proposed earlier that humans choose leaders based on particular situation, favouring physical parameters under threatening and competitive conditions, and prosocial traits in cooperative tasks (Little, 2014). However, studies investigating perception of leader traits, by and large are based on stated preferences for leadership. In the current study, we address the question of whether individuals who actually emerged as leaders in male groups (under condition of cooperative interactions with no pressure of inter-group competition) were perceived by others as more masculine, physically strong, and dominant based on their static facial morphology.

The relation between masculinity and health is another question of interest. Facial and body masculinity represent exposure to male sex hormones (in particular, testosterone), which is costly to produce and metabolize (Alonso-Alvarez et al., 2007). A number of studies report positive association between male-specific traits and actual men's health (Boothroyd et al., 2013; Thornhill & Gangestad, 2006), however findings on perceived health and masculinity remain somewhat contradictive (Boothroyd et al., 2005; Henderson et al., 2016). Health was also studied with regard to prosociality and altruism, and it is well established that good health and happiness are linked to prosocial and altruistic behaviour (Post, 2007). However, it is still difficult to conclude what might be the cause and what the consequence.

Such traits as trustworthy, generous and competitive were chosen as measures of perceived prosociality. A number of studies have previously demonstrated that humans can deduce prosocial intentions of others based on silent videotapes (Fetchenhauer et al., 2010; Oda, Naganawa, et al., 2009; Oda, Yamagata, et al., 2009), and even static facial images (Little et al., 2013; Tognetti et al., 2013; Verplaetse et al., 2007), and that such abilities are cross-population. At the same time, cheating behaviour is very problematic for non-verbal recognition, which has been demonstrated in numerous works (Brown et al., 2003; Okubo et al., 2012, 2013, 2017, 2018; Verplaetse et al., 2007). Moreover, studies show that generally people tend to trust attractive-looking individuals even if they don’t actually behave trustworthily (Rezlescu et al., 2012; Shinada & Yamagishi, 2014; Takahashi et al., 2006; Wilson & Eckel, 2006).

Here we hypothesize that: (1) observers will be able to differentiate leadership potential and leader styles by certain physical and behavioral characteristics based only on static facial information; (2) leaders’ faces are expected to be perceived as more masculine, physically strong, dominant, and healthy, compared to non-leaders; given a high rate of following leaders’ suggestions in the game (88%), we also hypothesize that faces of both prosocial leaders and leaders-cheaters should be scored high on attractiveness, generosity, trustworthiness, and low on competitiveness; (3) we expect Buryat raters to be more accurate in judging individual traits (based on Buryat photographs), compared to Russian raters; and (4) we expect Russian raters from Buryatia to be more accurate in their judgements of Buryat portraits than Russians from areas where contacts with individuals of Mongolian origin are scarce. There was no clear hypothesis concerning the possible sex effects in perceiving the appearance of male individuals distinguished by the propensities for leadership and leadership styles. On the one hand, leadership is generally more closely associated with male activities (Smith et al., 2020; Van Vugt & von Rueden, 2020; von Rueden et al., 2018), which may cause better recognition of individual qualities in the male-male context, but on the other hand, women are generally known to be better at processing non-verbal signals and mind-reading (Stevens & Hamann, 2012; Thompson & Voyer, 2014), which could lead to the opposite effect.

The present study is the first to our knowledge to test whether neutral facial images may convey information about actual (not perceived) personal trustworthiness in the context of emergent leadership among males. Determining, which traits are attributed to leaders with cheating propensities will help to get an insight into complex phenomenon of successful social cheating.

The goal of the present study was to assess how actual prosocial/cheating behaviour in the context of emergent leadership among men (revealed earlier in experimental setting) is related to perception of their neutral faces by representatives of own and contrasting populations, as well as by raters of different sexes. Although many studies have been previously conducted investigating different aspects of human faces’ perception, very few were focusing on matching facial perception with actual prosocial behaviour of judged subjects (Tognetti et al., 2013; Verplaetse et al., 2007), and even those few were targeting the ability to distinguish between limited types of behaviour (prosocial/selfish) based on facial images. Here we applied the different approach, estimating which physical and behavioural traits are attributed to averaged faces of individuals with different types of leader/follower and prosocial/cheating behaviour. The present study involved raters of different population origin: Buryats (Mongolian people from Eastern Siberia, representatives of the population from the stimulus portraits) and Russians (people of Caucasian origin, representing contrasting population).

Our results revealed that raters of the both sexes and from both populations highly agreed at judging composite portraits of individuals with different behavioural backgrounds. In particular, it was manifested in judgements on physical traits (masculine, physically strong, healthy), and a number of behavioural traits (trustworthy, generous and competitive), meaning that the same kinds of composites received equally high or low judgements on these traits independently of sex or population origin of the raters (Figure 2). However, ratings on such traits as leadership, dominance, and attractiveness were very inconsistent, especially among Buryat raters. This suggests that judgements on these traits are highly dependent on personal characteristics of the raters (Little et al., 2011), which should be taken into account in future research. Russian raters in our study differed in the degree of familiarity with Mongolian appearance in everyday life. Contrary to what was expected, our results demonstrated that living in the mixed-racial social environment also did not affect judgements of the composite portraits: Russians from Buryatia (living side by side with Buryats), and Russians from multi-ethnic megalopolis (Moscow), and Russian provinces (with mostly Russian population) did not differ either in ratings of the Buryat portraits or in the consistency of judgements. These results together support the notion that humans perceive facial appearance of others in a similar way (Van’t Wout & Sanfey, 2008), and that perception of at least male faces is not strongly sex- or population-specific (Tognetti et al., 2013). The similarity of the mechanisms of facial perception in humans was previously proposed by many authors in relation to the recognition of facial expressions (Ekman et al., 1969; Ekman & Friesen, 1971, 1975; Sorenson, 1975). However, in the more recent literature addressing this issue, the Universality Thesis (as the claim that basic emotions are universally recognized by humans regardless of cultural and population backgrounds) has been widely debated and criticized (Matsumoto & Assar, 1992; Nelson & Russell, 2013). Recent studies revealed considerable cultural specificity, when remote (isolated) societies were tested (Crivelli et al., 2016; Gendron et al., 2014, 2020). These results questioned whether and to what extent specific aspects of facial perception differ across populations. One of possible arguments for high similarity of mechanisms of human perception of static faces across different populations may be based on the fact that certain behavioural features (e.g. cooperativeness or aggressiveness) (Carré & McCormick, 2008; Geniole et al., 2015; Rostovtseva et al., 2021b; Stirrat & Perrett, 2010, 2012; Wen & Zheng, 2020), as well as physical characteristics (e.g. physical strength) (Butovskaya et al., 2018) are known to be incorporated into individual facial morphology, and display similar patterns at cross-population level (although see: Butovskaya et al., 2021). This may partly explain why people react similarly to certain types of facial appearance (such reactions may be driven by their actual life experiences). However, this may also be partly a result of high exposure to the global international mass media (television, Internet, cinematograph), which introduces a large diversity of the humankind to the audience from different parts of the world. To clarify to what extent unified perception of neutral facial traits may be the consequence of the global information environment, studies in societies with limited access to international mass media, should be conducted.

The composite portraits of leaders generally were scored as more physically strong and more masculine than the non-leaders’ portrait, which was significant with no respect to age, sex, and population origin of the raters. This result demonstrates that preference for physically developed leaders may occur not only under condition of explicit intergroup competition, which is in line with findings by other authors (Little, 2014). Possibly, one of the reasons that individuals with high perceived physical strength and masculinity emerged as leaders during group male cooperative tasks lies in a specific cultural background of Buryats. In the recent past Buryats were traditionally nomadic pastoralists with highly developed male warfare practices, and male collective hunting (more details on their culture can be found elsewhere: Rostovtseva et al., 2020b). Such cultural traditions may still have echoes in their modern social behaviour. At the same time, the leaders-cheaters’ composite portrait was perceived as the most masculine by younger raters (< 28 years), whereas older ones tended to score the prosocial leaders’ composite higher on masculinity (Figure 4b). Visual inspection of the stimulus portraits (Figure 1) suggests that the main features distinguishing leaders-cheaters and prosocial leaders are the shape of the eyes, lips, jaw, and brows (with leaders-cheaters having larger and more rounded eyes, fuller lips, more rounded jaw, and more prominent brows). In previous research on facial sexual dimorphism of Buryats it was revealed that more rounded shape of the eyes, and more rounded jaw are among the main female sex-specific features characteristic of this population (Rostovtseva et al., 2021b). It has been previously demonstrated that feminized male faces are perceived as more attractive and honest (Perrett et al., 1998), which may partly explain perception of the leaders-cheaters’ composite as the most trustworthy. But still the question arises why younger raters tended to score their portrait as the most masculine, whereas it was characterized by more feminine facial appearance. Since stimulus portraits were based on the faces of young Buryats (17 −25 y.), we can conclude that the effect of perceiving the leaders-cheaters’ facial appearance as more masculine occurred only within peer environment. This result suggests some implicit mechanisms of peer facial perception (Rhodes & Anastasi, 2012), exploited by individuals with cheating intentions. However, yet we can only speculate on this issue, and more research on morphology and perception is needed to clarify it.

The composite portrait of prosocial leaders was judged as the most healthy. This is in line with evidence proposing that prosociality and altruism are positively related to health (Post, 2007). Moreover, as a general trend, perceived health was also positively associated with scores on trustworthiness (Figure 2). However, associations of perceived health with other traits were very weak (this is the trait with the lowest correlations with other traits among all tested in this study) (Figure 2). Perceived health was not related to perceived masculinity, generosity, competitiveness, and leadership. This result corresponds to the previous findings that health judgements and masculinity are not directly associated (Henderson et al., 2016). However, perceived health was also not related to attractiveness, which contradicts earlier findings by other authors (Jones et al., 2001).

Considering human prosociality, which is known to be a cornerstone of the Homo sapiens species, large work has been conducted to investigate those mechanisms, which theoretically could lead to high density of prosocial individuals in human societies (see reviews: Apicella & Silk, 2019; Nowak, 2006). However, there are always a certain proportion of cheaters and free-riders in human populations. One of the reasons behind that are sophisticated mechanisms underlying choice of interaction partners, including recognition of social traits by perceived cues, but at the same time, a kind of prosocial mimicry or camouflage, which allow selfish agents to lure the victims for exploitation (Dawkins, 1976; Gambetta, 2005; Mokkonen & Lindstedt, 2016). In line with this theoretical background, the composite portrait of leaders-cheaters in our study was scored as the most trustworthy, most generous and least competitive compared to all other composites. These results indicate that leaders-cheaters had certain features of static facial morphology, which made them to be perceived by others as more prosocial. Although the respondents did not directly assess leaders-cheaters as better leaders, something in their appearance made people trust them more. This may partially explain why they could successfully free-ride on the goodwill of others. This effect was weak, but significant independently of sex, and population origin of the raters.

However, this study has certain limitations. The stimulus composite portraits were created based only on 7 individual photographs per type, which can be considered as relatively few. The number of individual photographs for averaging is limited by the minimal number of individuals per behavioural type, but due to frequency dependent selection (Avilés, 2002; Pruitt & Riechert, 2009) leaders-cheaters are expected to remain in great minority. Unfortunately, this limitation is difficult to overcome within a single experimental study, using design with face-to-face interactions and ethological analysis of behaviour. One of the possible solutions could be pre-screening individuals to gain larger sample sizes. Another limitation is that, as a certain compromise, the order of judged traits was not randomized in our study. Finally, one should remember that the aim of our study was to investigate emergent leadership in groups of individuals who were strangers to each other. This implies that reputation (which is known to be one of the most important aspects of leadership (Anderson et al., 2015; Henrich & Gil-White, 2001; Van Vugt & von Rueden, 2020; von Rueden & Van Vugt, 2015), was not involved in the experimental interactions. We assume that if reputation was involved, it would override the influence of all other factors (both at the level of leadership and cooperation). Our study targets the processes of social self-organization in initially leaderless groups, and the obtained results cannot be generalized to communities with already established social hierarchy. However, the results may apply to social processes in modern urban environments, where interactions among strangers are very common. More independent studies are required to test reproducibility of the obtained results.

To sum up, our results reveal that individual features of static facial morphology also contribute to the phenomenon of successful cheating in the context of male emergent leadership, along with personality and communicative traits (Okubo et al., 2012, 2013, 2017, 2018; Rostovtseva et al., 2022). Our study opens up new perspectives for further research, invoking investigation of particular facial structures, which distinguish individuals with different leadership qualities, and their specific contribution to facial perception.

We want to thank all our subjects for their interest and willingness to participate in the project. Special thanks to the East-Siberian State Institute of Culture (VSGIK) and the Faculty of Natural Sciences of the Tula State Lev Tolstoy Pedagogical University for collaboration. We also express our gratitude to the anonymous reviewer and associate editor Lisa Welling for proofreading the English writing.