Face(book) perception: Is the own-race advantage due to perceptual learning?

Abstract

Many research studies, typically employing old-new recognition tasks, have shown that faces of one's own race are often easier to recognize than faces of other races. Many theorists assume a perceptual-learning basis for this phenomenon; because one normally has more exposure to own-race faces, perceptual processes learn to extract information that optimally discriminates members of that race from each other, but these processes may be sub-optimal at discriminating shape differences among faces of other races. In recent years, however, a number of proposals have emerged that suggest a locus in social categorization for the ORA; specifically, that we normally individuate in-group members (such as own-race individuals) but not out-group (other-race) individuals. This distinction in processing is thought to be strategic, because encouraging people to individuate other-race faces has been found to eliminate the ORA. To differentiate between these theoretical alternatives, we used a new task. Faces of specific individuals (with their permission) were downloaded from Facebook. Rather than perform old-new recognition, participants (Hong Kong Chinese and White Australians) learned to name the person from five different photos of their face. Initially, names were learned for individual faces, and then had to be generalized to new photos of the same individuals. Participants only moved to the next block if they could correctly name all the photos in the set. Participants learned eight individuals in each of four blocks, where individuals in one block were the same sex and race. Crucially, participants were strongly motivated to learn to individuate each face, because the time they spent in each phase increased with each error they made. We found a strong ORA (fewer trials to reach criterion for own-race than other-race faces). These results show that there remains a strong perceptual-learning basis for the ORA.