Differences between the perception and the recognition of cross-race faces

The Cross-Race Effect (CRE) refers to the robust finding that individuals more accurately recognize faces of their own race than those of other races. According to the Face Space Model, this effect arises because cross-race faces are represented more densely-i.e., with less differentiation-in psychological space than same-race faces (Valentine & Endo, 1992). Although some studies have demonstrated that same-race faces occupy more dispersed positions in multidimensional space (e.g., Byatt & Rhodes, 2004), few have directly examined the link between recognition performance and underlying psychological face representations. The present experiments tested the Face Space account of the CRE by assessing recognition accuracy and perceived facial similarity across racial group. In Experiments 1A and 1B, Asian, Black, and White participants completed old/new recognition tasks for same- and cross-race faces. In Experiments 2A and 2B, separate groups of participants rated facial similarity for the same stimuli and were represented in the face space. We hypothesized (a) superior recognition for same-race faces and (b) that cross-race faces would be judged as more similar and thus more densely clustered in Face Space. Results partially supported these predictions. Consistent with the Face Space account, distinctive faces were overall better recognized than less distinctive faces and some conditions revealed recognition performance was aligned with perceived face clustering by race. However, other conditions showed a dissociation between memory and similarity where cross-race faces were more poorly recognized than same-race faces but were perceived as equally similar. These findings suggest that the CRE may not be fully accounted for by representational density in Face Space alone and highlight the potential dissociation between cognitive and perceptual processes of same- and cross-race faces.