Abstract
A robust finding in the face recognition literature is that people are better at recognizing faces from their own race than faces from another race. According to the perceptual expertise account, the other-race effect is due to differences in categorization of own- versus other-race faces rather than differences in perceptual exposure to them. As objects of expertise, own-race faces are categorized at the subordinate level of the individual (e.g., Joe, Tom, Bob) whereas other-race faces are categorized at the basic level of race (e.g., African American, Asian, Hispanic). If own-race recognition is a type of perceptual expertise, it is an open question as to whether people can be trained to become proficient in other-race face recognition. Following a previous expert training protocol (Tanaka, Curran & Sheinberg, 2005), Caucasian participants received two weeks of training where they learned to individuate African American (or Hispanic) faces at the subordinate level and classify Hispanic (or African American) faces at the basic level. During training, visual exposure was controlled such that participants received an equal number of training trials to African American and Hispanic faces. The main finding was that post-training recognition performance improved for novel faces from the subordinate level racial group, but no improvement was found for faces from the basic level racial group. These results indicate that perceptual categorization rather than perceptual exposure is critical for promoting transfer effects in other-race face recognition. In a subsequent experiment, we used the Bubbles technique (Gosselin & Schyns, 2001, Schyns, Bonnar, & Gosselin, 2002) to identify race-specific features that support other-race face recognition. The goal of these experiments is to identify how the demands of basic and subordinate level categorization can cause changes in perceptual sensitivity to the features that distinguish faces from different racial groups and faces within the same racial group.