27-10-2012, 01:18 PM
A Unified Brain Architecture for Perception and Cognition With Applications to Information Processing Technology
ABSTRACT
This project develops neural and computational models of the brain mechanisms that underlie human perception and
cognition. It does this by mathematically characterizing and quantitatively simulating key brain mechanisms underlying
higher-order human information processing as carried out within the laminar structure of the cerebral cortex.Amodel Laminar
Computing paradigm promises to generate powerful information processing tools for dealing with large-scale problems.
Perceptual projects model how the laminar circuits of visual cortex are designed to group distributed visual information into
emergent object representations, to pay attention selectively to important information, and to learn their own optimal operating
parameters in different visual environments; Cognitive processing projects model how the laminar circuits of inferotemporal
and prefrontal cortex can rapidly classify, decide between, and predict noisy and potentially conflicting information in rapidly
changing environments. The resulting models can be applied to technological problems in which the ability to autonomously
visualize, learn, predict, and control information in rapidly changing environments is required. Testbed problems, including
geospatial mapping and medical database analysis, have been developed in the context of the AFOSR-sponsored CNS
Technology Laboratory.