Making features relevant: learning faces and event-related potentials recording using an analytic procedure.

We present a procedure designed for the learning of faces (represented by realistic drawings) and a task for the subsequent recording of Event-Related Potentials (ERPs), with the aim of investigating the psychobiological mechanisms involved in the recognition of familiar (or known) faces. The learning procedure consisted in a series of six sessions in which subjects familiarized themselves with the faces through a forced-choice features task: after the study of a set of 40 faces, each of these was presented at random in an incomplete way (without the eyes/eyebrows fragment), together with two sets of eyes and eyebrows for the subject to decide which of them corresponded to each face. This procedure is thought to facilitate for subjects the acquisition of the information related to the structural description of the faces (that is, without associated verbal/semantic information) using preferentially analytic processing strategies. The parameter d' of the Theory of Signal Detection allowed us to evaluate in the course of the learning sessions the degree of familiarization achieved with the studied faces. Subsequently, an ERPs' recording session was carried out, during which subjects executed a face-feature matching task. In this task, similar in structure to that carried out in the learning sessions, the subject had to decide whether the automatic completion of the faces that was made was correct or incorrect. ERPs' effects related to mismatching features were obtained, which indicate the existence of specific cerebral mechanisms involved in the recognition of faces.