Neural correlates of temporal interference between faces

Theoretical models of face processing incorporate within-pool inhibition - a mechanism by which representations compete with one another to support recognition. However, direct evidence for these inhibitory mechanisms is sparse in studies of human face perception. Here, we studied immediate repetition priming of faces to investigate this issue, using event-related potentials (ERP). In Experiment 1, we presented two familiar faces (Prime 1 and Prime 2), in sequence, followed by a target. We observed clear priming from Prime 2 (when it showed the same identity as the target) but no effect from Prime 1. In Experiment 2 we manipulated the familiarity of Prime 2, and again found that any effects of Prime 1 were eliminated by subsequent presentation of Prime 2. Experiment 2 also established that the familiarity of Prime 2 did not affect this inhibition - suggesting that it did not occur at the level of familiar face representations. Experiment 3 did find Prime 1 repetition effects with longer stimulus onset asynchronies between primes (500 ms relative to 200 ms in Experiments 1 and 2), suggesting no effective interference from a subsequently presented face once the representation of Prime 1 had become fully activated. These findings show effects of inhibition within the face processing system, but not at the level of abstract representations of familiar people, as had been previously proposed. We discuss the implications of these findings for models of face recognition.