People form impressions of others from their faces, inferring character traits (e.g., friendly) along two broad, influential dimensions: Warmth and Competence. Although these two dimensions are presumed to be independent, research has yet to examine the generalizability of this model to cross-group impressions, despite extant evidence that Warmth and Competence are not independent for outgroup targets. This thesis explores this possibility by testing models of person perception for own-group and other-group perceptions, implementing confirmatory factor analysis in a structural equation modeling framework, and analyzing the underlying trait space using representational similarity analysis. I fit 402,473 ratings of 873 unique faces from 5,040 participants on 14 trait impressions to own-group and other-group models, exploring whether perceptions across race and gender are more unidimensional. Results indicate that current models of face perception fit poorly and are not universal as presumed: the space of trait impressions varies depending on targets’ race and gender.
Keywords: person perception, impression formation, face perception, intergroup processes, social cognition