Abstract
The human visual system is very fast and efficient at extracting socially relevant information from faces. Visual categorization studies employing foveated images have shown that human observers categorize other-race (OR) faces faster than same-race (SR) faces by race, producing an early perceptual bias termed the other-race categorization advantage (ORCA). Whether this perceptual advantage persists in the parafoveal visual field containing low spatial frequency, and which facial features are sampled during such categorization yet remain undetermined. To this aim, we recorded the eye movements of East Asian (EA) and Western Caucasian (WC) observers during a conventional foveal categorization by race of normalized WC and EA faces (FCRT, Experiment 1), a parafoveal saccadic reaction time (PSRT, Experiment 2) and a response time paradigm (PCRT, Experiment 3). EA and WC observers categorized OR faces faster than SR faces by race across all paradigms. However, this ORCA occurred at the rapid speed of just 300ms in the PSRT, almost twice as fast as in the manual response paradigms (PSRT < FCRT < PCRT: 300ms < 600ms < 750ms). Furthermore, fixation maps revealed a cultural perceptual bias in the FCRT, with WC observers sampling more the eye region and EA observers the central region of faces. Interestingly, such cultural contrast was abolished in the PSRT, with both groups deploying a single fixation to the middle of the eyes. Our PSRT data show that the speed of race categorization is boosted by visual field eccentricity, which eliminates unessential and time-consuming visual information processing. These findings also offer a novel explanation for eye-movement discrepancies reported on the cultural differences in face processing. Overall, our observations provide new evidence of race as a powerful rapid low-level visual signal transmitted by faces, which could relate to primitive functional mechanisms dedicated to the evolutionary-relevant social categorization of ingroup/outgroup members.
Meeting abstract presented at VSS 2018