The neural correlates of own- and other-race face recognition and categorization: A fMRI study

Abstract

Faces from one’s own race and those from other race are processed in different ways. The most direct evidence supporting such hypothesis comes from two paradoxical cross-race effects. One is the own-race recognition advantage that refers to a more rapid and accurate individual recognition for the own-race faces than that for other-race faces. Another is the other-race categorization advantage, whereby when categorizing face by race, individuals categorize other-race faces fast and sometimes more accurately than own-race faces. However, no neural imaging study has concurrently examined the neural correlates underlying these two cross-race effects. Here, we used a 2 face race (Caucasian vs. Chinese) × 2 task (individual recognition vs. race categorization) within-subject factorial design to bridge this gap. We compared the activation between own-race faces and other-race faces as well as between the individual recognition task and the race categorization task. Results revealed a significant interaction effect of face race by task type within the fusiform face gyrus (FFA) as well as the occipital face area (OFA). Further, in each of them, the own-race faces elicited equal responses to the two tasks, whereas the other-race faces produced greater responses in the individual recognition task than in the race categorization task. Our findings suggest the response of the FFA and OFA to own-race faces is insensitive to the change of task demands, whereas their responses to other-race faces can be enhanced, perhaps by shifting from a default categorical level to the individual level of processing. Given the significant roles of the FFA and OFA play in the identification of individual faces, our finding is consistent with the hypothesis that the own-race faces can be by default individually recognized regardless the task demands, whereas the other-race face can be automatically classified by race unless an instruction of recognition is explicit.

Meeting abstract presented at VSS 2012