Searching for face-specific long latency ERPs: a topographic study of effects associated with mismatching features.

In a previous study [E. Olivares, M.A. Bobes, E. Aubert, M. Valdés-Sosa, Associative ERPs effects with memories of artificial faces, Cogn. Brain Res. 2 (1994) 39-48] we reported the presence of a negativity associated with mismatching features when subjects carried out a face-feature matching task whilst their evoked potentials were recorded. Since the stimuli used were learned faces (realistic drawings), for which the subjects possessed no semantic information or associated verbal labels, the mismatch negativity obtained was considered a face-specific N400. In this work we present a new experiment to study the topographic distribution of these mismatch effects. As in the above-mentioned study, in each trial the subjects observed previously an incomplete (without the eyes/eyebrows fragment) familiar face, which served as a structural context for the face recognition. The face was then completed by grafting either matching (learned) features or mismatching features (from another face). In line with neuropsychological studies on prosopagnosia and electrophysiological findings in humans and non-human primates, we found as one of the most relevant items of data that the most-posterior (principally, left occipital) cortices appear to be a region in which are located the possible neural generators of the negativity associated with the detection of incongruencies in the structure of familiar faces. We also reported a late positivity, distributed in more anterior regions, which follows the mismatch negativity. This complex N-P is interpreted as reflecting a dual process of retrieval and integration of information in memory.