Individuals readily learn and keep track of the social group membership of people around them. Although this knowledge is often not visual, here we investigate the surprising possibility that it can nevertheless bias visual processing of faces. In Experiment 1, participants learned that novel faces belonged to students enrolled in either an engineering or humanities major. During this learning phase, participants viewed a name, major, and brief biographical description of each person, and then viewed the person’s face for 9s. During a subsequent evaluation phase, participants indicated to what extent they thought each face looked like a typical engineering or humanities major. Faces that were randomly assigned to both the engineering and humanities majors were rated as more typical of their respective groups, as compared to the ratings of a control group who did not complete the learning phase. Because this perceived typicality could result from demand characteristics, participants in Experiment 2 were explicitly told that the pairings of faces and majors were completely random. Despite this manipulation, faces were still rated as more typical of their respective groups. These results suggest that knowledge about social group membership biases face perception compulsorily. To build on this behavioral effect, Experiment 3 investigated the influence of social group information on neural representations of faces using fMRI. Participants first completed a similar learning phase to that described above, and then they viewed pairs of faces belonging to the same person, the same social group, or different social groups. Adaptation and multivariate pattern analysis are used to identify representations of individuals and social groups. Our results suggest that higher-order social knowledge can intrude in perceptual judgments, and they concur with the common intuition that people seem to ‘look’ like they belong in their social group.
Meeting abstract presented at VSS 2013