| |
Abstract:
Object-centered theories of visual object recognition, such
as Recognition-by-Components, propose that object representations
are stored in memory as three-dimensional structural descriptions.
These structural descriptions are formed by breaking the retinal
image into simple components (geons) and their spatial
relationships. This study examines the pigeon,s ability to form a
structural description for simple, single-geon objects. Pigeons
were trained to discriminate among four geons in a four-alternative
forced-choice task. Pigeons were trained with five rotations of
each geon, spaced at 72 degree intervals. After the pigeons had
reached a high level of accuracy, one rotation of each geon was
assigned to a new discriminative response. The pigeons were then
tested with the four rotations that had not been trained to the new
response. If the pigeons had formed a generalized structural
description of each object, then they should make the new response
to the withheld views. But, the pigeons showed little tendency to
make the new response to the withheld stimuli, therefore giving no
support to the possibility of their having formed a generalized
structural descriptions for each of the four geons.
|
