Abstract
Numerous studies support the critical role of horizontal spatial information in upright face identification. Recently, it was shown that the best face recognizers are more selectively tuned to horizontal information (Duncan et al., 2019; Pachai, Sekuler & Bennett, 2013), and the well-known face inversion effect induces a clear reduction in horizontal tuning for upright faces (Goffaux & Dakin, 2010). In the present study, we compared spatial orientation tuning for same-race and other-race faces in 10 White-Canadian participants. Participants completed a 5-AFC task in which they had to memorize a target face that appeared for 1000 ms. Participants completed 1,500 trials per race, and both races appeared in different counterbalanced blocks. A learning phase was completed prior to the experiment to assure that participants were able to identify both White-European (SR) and East-Asian (OR) faces with approximately the same accuracy. Selectivity to horizontal facial information was measured with the orientation bubbles method (Duncan et al., 2017), which randomly samples stimulus spatial orientation content on a trial basis. Classification vectors were produced separately for each race by calculating a weighted sum of orientation samples, using standardized accuracies as weights. Overall, successful recognition of both races was associated with horizontal orientation information. However, this link was stronger for SR faces, compared to OR faces (Zcrit = 2.101, p<0.05). These results reinforce the crucial role of horizontal facial contour in face recognition with same-race faces and also raise the possibility that these low-level visual information could account for the advantage of same-race faces to be better remembered than other-race faces (i.e. other-race effect; Meissner & Brigham, 2001).