Our long-term objectives are to understand the ways in which visual form information is represented and stored in the primate brain. Electrophysiological and lesion data suggests that inferior temporal cortex (IT) is crucial to these functions.
The aim of this proposal is to contrast specific architectural components of computational storage models with information obtained from neural recording experiments in IT, as well as to understand basic principles of IT neural population behavior which transcend any particular model. The approach we take to investigate these issues includes, single unit recording, multi-unit recording, an awake behaving monkey preparation, various analytical techniques, and development of specific hypotheses based on computational models. The specific issues concern neural ensemble coding of information, neural assemblies and tests of specific computational model hypotheses. While there is suggestive evidence that information is stored in IT as a code across a population of neurons, there have been few attempts to measure these codes. Similarly, there is evidence that IT participates in visual information storage, yet there has been no demonstration that population codes form in correspondence with the ability of the animal to recognize visual patterns. While we have preliminary data concerning temporal coordination of neural activity in IT (i.e., neural assemblies), specific questions relevant to the mechanistic functions of feedback, which are reflected in temporal coordination, remain to be answered. We propose to test particular computational models by looking for evidence of """"""""competition"""""""" or pattern normalization, and for evidence of sequential search through representation codes. This project is part of a program aimed at the development of visual prosthetic devices, and machine vision. These two problems are potentially very closely related, in so far as an in-depth understanding of visual system mechanisms could rapidly advance artificial computational devices, while at the same time inherently guiding the design of these devices in a direction compatible with a biological interface.
