Datensatz

Zusammenfassung

Human psychophysics has shown that some perceptual learning effects are invariant with respect to stimulus manipulations, for example changes in object size.
Here, we examine invariances of visual object learning in the macaque monkey. Monkeys were first trained on regular sessions of a delayed-matching-to-sample task, in which they were briefly presented with a sample object, which was parametrically degraded with various amounts of visual noise. After a brief delay, an undegraded test object was presented, and monkeys had to release a lever if the test object matched the sample. Performance on this task decreased systematically as more and more noise was added to the stimuli. In addition, thresholds were significantly lower for identifying highly familiar objects in the presence of noise compared to novel objects, which the monkeys had not seen before. Thus, repeated experience with a particular set of objects allowed monkeys to identify them in the presence of greater amounts of noise compared to novel objects.
We used this experience-dependent difference in psychophysical performance to study learning invariances, by interspersing test sessions among the regular sessions described above. In test sessions, a systematic change was made to the objects. By comparing performance to regular session, we are able to assess whether visual learning is affected by a given stimulus manipulation. We have preliminary data (4 sessions) from one monkey with mirror-reversed objects, which indicates that mirror reversal leads to psychophysical performance intermediate between the trained views and novel objects. In addition, we are examining the effects of changes in size and visual field position.