Assessing the speed and ease of extracting group and person information from faces

The human face is a key source of social information. In particular, it communicates a target's personal identity and some of their group memberships. Different models of social perception posit distinct stages at which this group-level and person-level information is extracted from the face, with divergent downstream consequences for cognition and behavior. This paper presents four experiments that explore the time-course of extracting group and person information from faces. In Experiments 1 and 2, we explore the effect of chunked versus unchunked processing on the speed of extracting group versus person information, as well as the impact of familiarity in Experiment 2. In Experiment 3, we examine the effect of the availability of a diagnostic cue on these same judgments. In Experiment 4, we explore the effect of both group-level and person-level prototypicality of face exemplars. Across all four experiments, we find no evidence for the perceptual primacy of either group or person information. Instead, we find that chunked processing, featural processing based on a single diagnostic cue, familiarity, and the prototypicality of face exemplars all result in a processing speed advantage for both group-level and person-level judgments equivalently. These results have important implications for influential models of impression formation and can inform, and be integrated with, an understanding of the process of social categorization more broadly.