SIMILARITY DRIVEN SELF-FACE CLASSIFICATION : TESTING THE ROBUST REPRESENTATION HYPOTHESIS

Abstract

This study provides converging evidence for the five properties of robust representations of self faces using a template-target matching task. When observers were asked to decide if a morph was derived from an actual face, performance was better for morphs that contained the self face. This processing advantage persisted even as similarity of the morph to the actual face decreased. According to the attractor field model, a pronounced advantage for self also suggests that, the atypicality of a face may be increased through extensive visual experience, leading to the enlargement of the attractor field of the robustly represented face. The finding suggest that the robust representations for a highly overlearned face (a) mediate rapid asymptotic visual processing, (b) require extensive visual experience to develop, (b) contain some abstract or view-invariant information, (c) facilitate visual and decisional processes across tasks and contexts, and (d) demand less attentional resources.