Holger Wiese¹ (), Linda H. Lidborg¹, A. Mike Burton^2,3; ¹Durham University, UK, ²University of York, UK, ³Bond University, Australia

Research suggests that our resources for representing multiple faces simultaneously are severely restricted. Yet, in our daily life we are often confronted with more than one person simultaneously, a situation which usually does not seem to pose substantial problems to the face processing system. Here, we used event-related potentials (ERP) in combination with repetition priming to test under which conditions two faces can be represented simultaneously. All experiments presented two faces as primes and a single face as the target. When two different familiar face primes were presented, the repetition of either elicited ERP priming effects, consisting of more negative amplitudes at occipito-temporal channels with a maximum at approximately 400 ms after target onset. This effect was observed when the two primes were presented at equal distances relative to fixation (E1), and it neither depended on the predictability of prime location (E2), nor on the physiognomic similarity of the prime identities (E3). These findings suggest that two faces are represented simultaneously when processing resources can be divided between two equally salient faces. However, as shown in further experiments, if one prime is presented at a central location while the other prime is presented either left or right of fixation, this flanker stimulus does not elicit priming (E4). This absence of flanker priming is independent of whether the central stimulus is familiar or unfamiliar (E5). Accordingly, a face, appearing in a peripheral and unpredictable location, can be ignored. Finally, subsequent experiments did detect priming from a flanker face when the central stimulus was a written name (E6) or when the flanker was task-relevant (E7), ruling out reduced accuracy in peripheral vision as an explanation for absent flanker priming in E4/E5. Together, these findings demonstrate the flexibility of the face recognition system to prioritize the most salient information.

Acknowledgements: Funded by the Economic and Social Research Council, UK (ES/X002063/1)