Abstract
People are better at recognizing faces from one's own race than individuals from other races. Is this effect perceptual or judgmental? A judgmental effect could be produced by adopting a lower criterion for judging individuals as being the “same” if they are from another race. Does the presumed deficit in individuating faces extend to distinguishing the gender or expression of people from other races? We address these questions with a novel perceptual-matching paradigm that eliminated such biases. Subjects, half of whom were Asians (recently arrived to the U.S.) and half Caucasian, performed a two alternative forced-choice task in which one sample was identical to the target face. Half of the faces were derived from a sample of Asian faces; the other half were derived from a sample of Caucasian faces. Each race sample was composed of half male and half female faces that were were posed with different expressions. The distractor face differed from the matching face by linearly adding or subtracting a dimension that corresponded to gender, expression (happy vs. unhappy) or indentity. The three types of discrimination were presented in a latin square sequence within blocks of trials. The subjects were not informed about the underlying dimensions by which a distractor could differ from the matching face. The distance from the midpoint of the matching and distractor faces was varied to determine the energy (mean squared difference in pixel luminance between the correct and distractor faces) required for a subject to achieve a threshold of 79% correct. Subjects required much more energy to achieve threshold for individuation than either gender or expression, which had about equally low thresholds. This increase in threshold for individuation was approximately equal for faces whether they were of the same or different race from the subject. More data are required to determine if there is a reliable cost for performing this task with faces of different races.