Cooperative neural networks underlying image description in visual cortex

Experimental data concerning different types of nonlinear spatial frequency filters in the cat visual cortex are presented. The results of the functional reorganization of the spatial frequency filtering properties of the receptive fields (RFs) are discussed. It is found that cooperative neural nets determine the rules of visual information processing due to the control mechanisms caused by attention (top-down control) and by cross-orientation and spatial frequency selective inhibition in visual cortical networks (bottom-up control). The visual cortex information processing is based on the spatial frequency properties of the RFs and depends on complex interactions within an intricate system of specific and nonspecific excitatory, inhibitory, and disinhibitory influences. The types and functions of the cortical filters in many respects are due to different organizations of the neuronal structures responsible for intracortical inhibition.<
>