Abstract
Human beings are experts at face perception, but this ability is greatly improved for familiar as compared to unfamiliar faces, as shown in individual face matching tasks across viewpoint for instance (Young et al., 1986). Here we seek to clarify whether this effect reflects a change of perceptual representation for familiar faces, and more generally when it takes place during face processing. We used a face identity adaptation paradigm in event-related potentials (ERPs) (Jacques, d'Arripe & Rossion, 2007). In each trial, two faces, either the same or different identities, were presented sequentially with a short ISI (∼250 ms). The adapting face appeared for ∼ 2800 ms and the test face appeared for 200 ms. The adapting face was presented in a frontal view, and the test face was presented in a three-quarter view. In half of the trials, the faces were personally familiar (same classroom as the participant). Twelve participants performed an identity matching task between the adapting and test faces, performing better for personally familiar faces (96% vs. 89%; pp=0.035), confirming previous results of an early identity adaptation effect (Jacques et al., 2007). In addition, a triple interaction between familiarity, repetition, and hemisphere (p=0.016) revealed an adaptation effect for familiar faces in the left hemisphere and for unfamiliar faces in the right hemisphere. A stronger effect of identity adaptation for familiar faces, independent of hemisphere, emerged only after 250 ms following stimulus onset (p=0.04). Overall, these data suggest that personally familiar and unfamiliar individual faces can be distinguished as early as 170 ms in the occipito-temporal cortex, and that these two types of faces are processed differently by the two cerebral hemispheres at that latency.