Abstract
Within face recognition we can distinguish two main tasks, categorization and identification. Based on studies of normal face recognition development and of brain-damaged patients we have argued that two qualitatively different notions of face configuration correspond to these two tasks. The categorization-related notion of configuration is implemented in a domain specific, fast, experience independent and neuro-anatomically distributed face detection system. The notion of configuration that underlies identification shares properties of the object recognition system. In normal adult behavior the two are difficult to disentangle because of the partial overlap between the time course of the two processes. Here we report behavioral and electrophysiological studies of normal viewers and face-disordered patients that focus on the time course of these two components. In the behavioral Experiments we presented a delayed masking task and measured identification performance as a function of the kind of backward masking (masks made of whole faces or of face parts) presented at different latencies following the first stimulus. Comparable conditions were presented in the ERP experiments where the focus was the modulation of the N170 by the type of stimulus (schematic face, naturalistic face) and the type of prime (naturalistic face primed by schematic face and vice versa). A comparison of normal data with results from brain-damaged patients provides support for a dual route model with parallel categorisation and identification components.