Olivia S. Cheung^1,2 (), Rawan Abdelaal¹, Treedom Beiyin Zhang¹; ¹New York University Abu Dhabi, ²Center for Brain and Health, NYUAD Research Institute

At a glance, multiple aspects of information are processed from a face. Is invariant information (e.g., identity, race, gender) integrated more strongly than changeable information (e.g., expression), despite visual salience of the information? Here, integration of invariant and changeable facial information is revealed by how race, gender, or expression may influence identification. Participants searched for a target face (e.g., a happy Black female) among distractor faces either shared identical aspects of information (baseline), or differed in one, two, or all three aspects. Integration was revealed by faster search when there was a difference involving a particular aspect in the distractors, compared with the baseline (see Zhao & Hayward, 2013). Prior to the main experiment, we first minimized potential influences that were unrelated to face processing: 1) overall luminance was adjusted among all faces using SHINE (Wilenbockel et al., 2010); 2) relative visual salience of the three aspects of facial information was assessed using inverted faces. When participants (N=20) searched for inverted target faces, we found the strongest effect when the target and distractor faces differed in race, then expression, then gender, information. Moreover, differences in two aspects of facial information on inverted faces also reflected the combination of visual differences. However, unlike for inverted faces, search performance for upright faces did not appear to be merely affected by visual salience differences. Instead, while participants (N=20) were the fastest when target and distractor faces differed in race, search speed was comparable when the target and distractor faces differed in either gender or expression. Similar results were also observed when target and distractor faces differed in two aspects of facial information. These results suggest that for upright faces, invariant information such as race and gender are more integrated with identity processing than changeable information, despite visual salience of expression over gender differences.

Acknowledgements: The research was supported by a faculty grant (AD174) and a Tamkeen NYU Abu Dhabi Research Institute grant (CG012) to O.S.C., and a NYU Abu Dhabi capstone grant to R.A.