Presentations of Shape in Object Recognition and Long-Term Visual Memory
Abstract
A wide range of psychophysical experiments investigating the mechanisms and representations underlying human object recognition have been conducted. In particular, the focus of this research has been an approach in which object recognition is mediated by at least two systems, one based on an explicit qualitative encoding of viewpoint-invariant features and one based on a metrically specific encoding of shape. Within the literature, this dichotomy has been most often associated with measures of the effect of viewpoint on recognition performance. For the most part, the common assumption has been that viewpoint-dependent patterns of performance are the signature of one recognition mechanism, while viewpoint-invariant patterns of performance are in the signature of another recognition mechanism. Reinforcing this distinction, viewpoint-dependent mechanisms have been more broadly associated with metrically specific representations sensitive to a range of image-based properties, for example, size, handedness, color, or illumination, while viewpoint invariant mechanisms have been more broadly associated with coarsely-coded representations insensitive to image based properties. To this point, the majority of work on this project has focused only on the former in recognition tasks where perceivers must discriminate between visually similar objects (e.g., a within- category or subordinate-level judgment). During the past year we have continued this line of research, but have extended our approach to include recognition tasks using objects that are relatively dissimilar in that they may differentiated by a small number of quantitatively different parts (e.g., a between-category or entry-level judgment). Object representation, Object recognition, Visual cognition.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 05, 1994
- Accession Number
- ADA281336
Entities
People
- Michael J. Tarr
Organizations
- Yale University