Object Constancy in Face Processing: Intermediate Representations and Object Forms

Publisher Website
Google Scholar
Abstract
Bruce (1988) has argued that face recognition mechanisms may utilise a description of the 3-dimensional surface of the face as well 2-dimensional pictorial information and that shading provides sufficient information to the extraction of the surface-based description. The present paper considers what kind of intermediate representation of surface shape might best serve the description of the face, in the context of the problem of object constancy under scalings and rotations of the object, and argues for a description based on measurements of local surface curvature. Failures of object constancy are attributed to the need to establish object forms through learning, in order to describe allowable global configurations of the face across and within individuals.

This publication has 21 references indexed in Scilit: