Abstract
The other-race effect (better recognition of own- compared to other-race faces) has been framed as a problem with discriminating among other-race identities. Another impairment in recognizing other-race faces is the ability to recognize the same identity across a set of images that incorporate natural variability in appearance (e.g., changes in expression, lighting conditions, head orientation), known as within-person variability. We recently reported that within-person variability affects identity perception more for unfamiliar other-race faces than unfamiliar own-race faces (Zhou, Laurence & Mondloch, 2014). In the current study we examined how within-person variability affects identity perception in familiar other-race faces (i.e., whether participants would mistake two images of the same other-race person as belonging to different people even when viewing photographs of familiar identities). Chinese participants (n=100) were given 40 images of two identities (20 images/model) and asked to sort them into piles according to identity such that each pile had all images of the same person. The two identities belonged to one of four categories: familiar own-race, familiar other-race, unfamiliar own-race, or unfamiliar other-race. There was a significant interaction between familiarity and race of faces, p = .007. When faces were unfamiliar, participants sorted photos into significantly more piles (i.e., perceived more identities) for other-race faces (M = 11.56) than for own-race faces (M = 7.28), p = .006. This own-race advantage was eliminated when the identities were familiar (Mean piles = 2.12 and 2.24 for own- and other-race faces respectively). We are currently replicating this finding by testing Caucasian participants (n=60) with Caucasian and African American faces that are either familiar (NBA players) or unfamiliar (College basketball players). Our study adds new evidence of a fundamental difference between familiar versus unfamiliar face recognition; the other-race effect is limited to unfamiliar faces.
Meeting abstract presented at VSS 2015