Abstract
The other-race effect (ORE), the finding that same-race (SR) faces are better remembered and discriminated than faces of other races (OR), has been consistently demonstrated in a variety of tasks and has been associated with better holistic processing of OR faces compared to SR faces. The mechanisms underlying the ORE are currently debated: expertise models claim that repeated exposure and individuation of SR faces account for the ORE, while social-cognitive models argue that individuation resources are allocated to in-group members, but not out-group members. Previous studies demonstrate that it is possible to reduce the ORE by explicitly instructing participants to individuate OR faces or by administering intensive (multi-session) individuation training with OR faces. The possibility of reducing the ORE by training with SR faces, however, has not been explored, and would be a key test of the assumption of the expertise model: that expertise gained with SR faces would not generalize to OR faces. The current study investigated whether training with SR faces can generalize to improvements in processing of OR faces, thus reducing the ORE. Specifically, Caucasian participants performed a 10-day training program that aimed to focus attention on configural and holistic aspects of SR faces. Participants were assessed pre- and post-training on processing abilities of SR and OR (Korean) faces using the Part-Whole Task. Participants showed an overall improvement on OR faces, driven by increased accuracy on OR whole trials. For several participants, holistic processing of OR faces matched that of SR faces. These findings are inconsistent with the assumption of the expertise model, and suggest individuation skills can transfer across races. The current training may accomplish this by creating an automatic bias to encode configural and holistic aspects of all faces, overcoming the social-cognitive bias to process OR faces in a more categorical and less configural manner.