Abstract
Human beings are natural experts at processing faces, with the exception of other-race faces. Despite numerous studies it is still unclear whether the so-called other-race effect results from changes of visual information processing depending on the race of input faces. We first investigated this question with hybrid faces (Schyns & Oliva, 1999) that combined a face of a specific race (Asian or Caucasian) at a coarse spatial scale with the face of the opposite race at a fine spatial scale. A condition combining faces of the same-race at the different spatial scales was included for controlling response biases. When both races were simultaneously presented in the hybrids, Caucasian and Asian observers were biased to perceive their own race faces, independently of scale, revealing a tuning for same-race face information. To understand which facial information observers use to classify (as ‘Asian’ or ‘Caucasian’) and recognize (face identify) same- and other-race faces, we used Bubbles, a response classification technique that sampled facial information across the faces (Gosselin & Schyns, 2001). Surprisingly, participants used across tasks more information to classify and recognize same- (eyes, nose and mouth) than other-race faces (e.g., eyes and nose for the classification task). Crucially, however, they exploited facial information more efficiently for same- compared to other-race faces. Altogether our results provide new insights on the other-race effect and define how the face system, calibrated with experience for processing same-race faces, flexibly optimizes its selection in the use of facial information.
RC is supported by a post-doctoral fellowship provided by the Swiss National Science Foundation.