DO VIEWPOINT-DEPENDENT MECHANISMS GENERALIZE ACROSS MEMBERS OF A CLASS

Evidence for viewpoint-specific image-based object representations hav e been collected almost entirely using exemplar-specific recognition t asks. Recent results, however, implicate image-based processes in more categorical tasks, for instance when objects contain qualitatively di fferent 3D parts. Although such discriminations approximate class-leve l recognition, they do not establish whether image-based representatio ns can support generalization across members of an object class. This issue is critical to any theory of recognition, in that one hallmark o f human visual competence is the ability to recognize unfamiliar insta nces of a familiar class. The present study addresses this question by testing whether viewpoint-specific representations for some members o f a class facilitate the recognition of other members of that class. E xperiment 1 demonstrates that familiarity with several members of a cl ass of novel 3D objects generalizes in a viewpoint-dependent manner to cohort objects from the same class. Experiment 2 demonstrates that th is generalization is based on the degree of familiarity and the degree of geometrical distinctiveness for particular viewpoints. Experiment 3 demonstrates that this generalization is restricted to visually-simi lar objects rather than all objects learned in a given context. These results support the hypothesis that image-based representations are vi ewpoint dependent, but that these representations generalize across me mbers of perceptually-defined classes. More generally, these results p rovide evidence for a new approach to image-based recognition in which object classes are represented as clusters of visually-similar viewpo int-specific representations. (C) 1998 Elsevier Science B.V. All right s reserved.