Positive attitudes towards women predict the extent to which trustworthy facial appearance is inferred from femininity in White faces
Under review
Collaborators: Neelamberi Klein, Erin Freiburger, Ryan Hutchings, Dhanya Elangovan, Jeanine Sempler, Kurt Hugenberg
Abstract
The inference of trustworthiness from faces has been attributed to the overgeneralization of adaptive cues. We tested an alternate model, that trait inferences from faces are the product of social attitudes, by investigating the effect of individual perceivers’ gender attitudes on the overlap between their representations of facial femininity and trustworthiness. In Study 1 (N = 478 and more than 286000 judgments), we built computational models of untrustworthy-to-trustworthy and masculine-to-feminine appearance from participants’ ratings of White synthetic faces. Participants with more positive attitudes towards women showed an overlap in their models of trustworthiness and femininity. In Study 2 (N = 178), a cluster analysis of data from Study 1 revealed a cluster that resembled the classic femininity-trust representational overlap but also an inverted cluster wherein trustworthiness overlaps with masculinity. The third cluster was middling. We projected the cluster-based models onto new faces and presented them to naïve participants to validate and characterize them. These clusters reflected differences in participants’ trust-gender overlaps but also gender attitudes.