Bodies were perceived as more threatening as they had added musculature & portliness, & less threatening with more emaciation, but threatening faces exerted the most influence when paired with non-threatening bodies

McElvaney TJ, Osman M, Mareschal I (2021) Perceiving threat in others: The role of body morphology. PLoS ONE 16(4): e0249782, Apr 8 2021. https://doi.org/10.1371/journal.pone.0249782

Abstract: People make judgments of others based on appearance, and these inferences can affect social interactions. Although the importance of facial appearance in these judgments is well established, the impact of the body morphology remains unclear. Specifically, it is unknown whether experimentally varied body morphology has an impact on perception of threat in others. In two preregistered experiments (N = 250), participants made judgments of perceived threat of body stimuli of varying morphology, both in the absence (Experiment 1) and presence (Experiment 2) of facial information. Bodies were perceived as more threatening as they increased in mass with added musculature and portliness, and less threatening as they increased in emaciation. The impact of musculature endured even in the presence of faces, although faces contributed more to the overall threat judgment. The relative contributions of the faces and bodies seemed to be driven by discordance, such that threatening faces exerted the most influence when paired with non-threatening bodies, and vice versa. This suggests that the faces and bodies were not perceived as entirely independent and separate components. Overall, these findings suggest that body morphology plays an important role in perceived threat and may bias real-world judgments.

4. Discussion

In two preregistered studies, we found evidence supporting the hypothesis that systematic changes in body morphology can significantly influence how threatening a person appears. Judgment of threat was primarily driven by facial information, with the odds of perceiving a person as more threatening increasing nearly threefold with each unit increase in perceived facial threat. However, larger bodies also tended to be seen as more threatening than smaller bodies, both in the absence and presence of facial information. Indeed, the odds of perceiving a person as more threatening increased more than one and a half-fold with each unit increase in perceived body threat. While the association between body size and perceived negative traits is not novel, this represents the first study, to our knowledge, to demonstrate that perceived threat can shift significantly with systematic changes in body morphology. Using this methodology, we were able to directly measure the effects of body morphology on perceived threat. In Experiment 1, bodies were perceived as more threatening the larger they became, most notably with increased musculature. This finding was replicated in Experiment 2.

Our findings are consistent with Palmer-Hague, Twele & Fuller [30], who found that perceived threat in facial stimuli was significantly predicted by BMI. They are also somewhat in line with Hu et al. [25], who found that more muscular builds tend to be seen as more dominant. More generally, these results dovetail with the growing literature on the capacity of appearance to significantly affect character trait inferences, while also adding to the sizeable obesity stigma literature. It appears that larger people may be perceived as more threatening. This would make sense from an ecological theory perspective, with size potentially serving as an inferred cue of strength. This increased perceived threat may contribute to biases against larger people [28, 29]. For example, in the realm of courtroom decision-making, it has been shown that defendants who appear untrustworthy are more likely to fall victim to harsher sentencing [6]. It is conceivable that people who appear threatening may also be more severely judged.

The study also contributes to the literature on the joint processing of bodies and faces. In line with the emotion recognition literature, we found that two stimuli sharing the same facial information can be perceived as significantly different depending on body information. This common interaction of face and body information in both this study and previous work on emotion recognition is perhaps unsurprising given the link between emotions and trait perception. The perception of emotional expressions has been shown to fuel, and can directly contribute to, overgeneralisations about other people’s trait characteristics [45–47]. Indeed, work by Montepare & Dobish [48] showed that actors posed with angry emotional expressions were perceived to be high in trait dominance and low in trait affiliation, while actors posing with surprise and happiness were seen as high in both trait dominance and affiliation.

However, the current study diverges from findings in emotion perception in the nature of the observed interaction of the face and body information. In contrast with work on combined emotional faces and bodies stimuli [15, 43], the contribution of the body here was maximised when paired with faces of low threat signal, rather than ambiguous threat signal. This could be attributable to the more transient nature of emotions in comparison to more stable character traits. Emotions are short and distinct feelings, which tend to have a specific cause [17], while character traits tend to be consistent over many years [18]. Similarly, the perceived emotion of a face can be rather malleable, and highly dependent on contextual and body cues [13]. Hence, contextual cues may be of particular importance when the facial cue is ambiguous. However, a face that signals a “neutral” level of a character trait such as threat may not be ambiguous or uninterpretable. Rather, it may be signalling a “medium” level of threat, an amount that can be processed and interpreted.

These results suggest that, rather than simply summing the independent threat level of the face and body, the two are integrated into a single judgment, that tends to be more heavily driven by the face. In this way, the perception of the compounds diverged from the mere sum of their separately perceived properties. The relative contributions of face and body seem to be driven by discordance, with faces exercising their greatest influence when paired with discordant bodies, and vice versa. This may be attributable to a pop-out effect, in that faces that may not appear to “match” the accompanying body (and vice versa) may be more likely to capture attention, and thus more strongly drive the judgment of the overall compound [49]. Although not providing direct evidence for holistic processing per se, this significant interaction lends some support to the hypothesis forwarded by Aviezer, Trope & Todorov [10]; that people do not perceive others as separate body and face components. Rather, it seems likely they are perceived as elements of a greater, whole-person unit. In this case, the signals of threat from face and body are integrated such that their respective strengths are dependent on the nature of their paired signal. The holistic person-perception hypothesis has found rather consistent support, from the emotion/identity identification literature [11] to findings on gaze detection [12]. However, this study represents the first evidence for such complimentary face and body processing in the area of trait/character inference.

While we found strong evidence for our primary hypotheses, we found no effect of participant height or BMI on perceived threat. This could be attributed to the manner in which the stimuli were presented. Participants were presented with an image on screen, as opposed to judging a real-sized potential threat. In a more realistic environment, it may be that people judge potential threats in terms of the personal threat posed. In this sense, a large person may feel less threatened by a person of average build than would a small person. Here, the potential effect of relative size may have been nullified. In addition, we found that the relation between perceived attractiveness and perceived threat was somewhat inconsistent. Although more attractive faces were perceived as less threatening, males found more attractive muscular bodies to be more threatening in Experiment 1. This may be due to the male participants not finding the bodies attractive in a romantic sense, but rather in recognition of a typically attractive male form [50]. In this case, the more muscular bodies were perceived as more attractive, but did not detract from the signalled threat. However, as the current study did not record the sexual orientation of the participants, this interpretation is somewhat speculative. Future studies investigating perceived attractiveness and threat should record the sexual orientation of participants to elucidate more clearly the nature of this interaction.

In addition to our primary hypotheses, we also observed significant effects of age and education in our first experiment. Older participants tended to perceive less threat in the stimuli, which is in line with previous work [39]. Contrary to expectations, it was also noted that participants with third-level degrees tended to perceive more threat than those without a degree, which is contrary to previous indications that participants of lower educational status show more hostile reactivity to ambiguous social scenarios [40]. However, it has also been shown that those of lower social rank and education may be more adept at tracking hostility [40]. As our stimuli did not overtly indicate hostility, these participants may have thus ascribed lower threat ratings.

A number of limitations of the current study should be mentioned. First, our study was limited to body stimuli which consisted entirely of images of white males. In order to generalise these findings, it would be useful to replicate the study using both female and male stimuli. It would be particularly relevant to repeat this with stimuli of varying races given the documented bias of young black men being perceived as bigger and more physically threatening than white men [51, 52]. Furthermore, our stimuli were entirely CG. While this lent us a level of control over body morphology that would have been impossible with images of real people, it limits the ecological validity of our findings. Future studies could attempt to use photo-editing software to systematically vary the body morphology of images of real people. Finally, the stimuli presented in this study were relatively small, displayed on a computer screen. A study utilising virtual reality (VR) apparatus [53] to display life-sized human stimuli to participants, while manipulating facial information and body morphology, may tap into a more ecological measurement of perceived threat. Furthermore, a VR study could also manipulate the participants’ own virtual height, thus exploring the impact of discrepant size on perceptions of threat.