Abstract
It has been proposed that East Asian and Western Caucasian observers employ different eye-movement strategies during face identification, with Asians preferring to fixate towards the face centre (Blais et al., 2008, PLOS ONE). Here, we evaluated whether Asian and Caucasian observers differed in the location of task-critical first eye movements during a fast face-identification task. Using a foveated ideal observer (Peterson & Eckstein, 2012, PNAS), we assessed whether the pattern of behaviour can be explained by an interaction between the race-specific distribution of discriminatory information in faces and the varying spatial resolution across the visual field. Methods: Asian and Caucasian observers (16 each) participated in two experiments: one, identifying Asian faces among other Asian faces; another, identifying Caucasian faces among other Caucasian faces. In each condition, observers viewed and identified briefly presented faces (350 ms; 7.7° between eyes and mouth) using a 10-alternative forced-choice match-to-sample paradigm, while their eye movements were monitored. Results: No significant differences (p = .69) were found between Asian and Caucasian observers’ first saccades, with an average landing position 1.1° below the eyes regardless of the face race presented. A subsequent experiment that forced observers to fixate at different regions of the face confirmed the functional importance of fixating just below the eyes for either race of observers and stimuli: Fixations further from the preferred points led to decreased accuracy. A foveated ideal observer that integrates information in face images optimally but is constrained by the decrease in visual sensitivity as a function of retinal eccentricity predicted the same optimal regions of fixation for Asian and Caucasian faces. Conclusion: People of both races employ similar task-critical first eye movements. These first eye-movement strategies are optimal based on the distribution of discriminatory information within same-race faces and the foveated nature of visual processing.
Meeting abstract presented at VSS 2013