Structural encoding and identification in face processing: erp evidence for separate mechanisms.

The present study had two aims. The first aim was to explore the possible top-down effect of face-recognition and/or face-identification processes on the formation of structural representation of faces, as indexed by the N170 ERP component. The second aim was to examine possible ERP manifestations of face identification processes as an initial step for assessing their time course and functional neuroanatomy. Identical N170 potentials were elicited by famous and unfamiliar faces in Experiment 1, when both were irrelevant to the task, suggesting that face familiarity does not affect structural encoding processes. Small but significant differences were observed, however, during later-occurring epochs of the ERPs. In Experiment 2 the participants were instructed to count occasionally occurring portraits of famous politicians while rejecting faces of famous people who were not politicians and faces of unfamiliar people. Although an attempt to identify each face was required, no differences were found in the N170 elicited by faces of unfamiliar people and faces of familiar non-politicians. Famous faces, however, elicited a negative potential that was significantly larger than that elicited by unfamiliar faces between about 250 and 500msec from stimulus onset. This negative component was tentatively identified as an N400 analogue elicited by faces. Both the absence of an effect of familiarity on the N170 and the familiarity face-N400 effect were replicated in Experiment 3, in which the participants made speeded button-press responses in each trial, distinguishing among faces of politicians and faces of famous and unfamiliar non-politicians. In addition, ERP components later than the N400 were found to be associated with the speed of the response but not with face familiarity. We concluded that (1) although reflected by the N170, the structural encoding mechanism is not influenced by the face recognition and identification processes, and (2) the negative component modulated by face familiarity is associated with the semantic activity involved in the identification of familiar faces.