The Laminar Architecture of Visual Cortex and Image Processing Technology

Abstract

The mammalian neocortex is organized into layers which include circuits that form functional columns in cortical maps. A major unsolved problem concerns how bottom-up, top-down, and horizontal interactions are organized within cortical layers to generate adaptive behaviors. This article summarizes a model, called the LAMINART model, of how these interactions help visual cortex to realize: (1) the binding process whereby cortex groups distributed data into coherent object representations; (2) the attentional process whereby cortex selectively processes important events; and (3) the developmental and learning processes whereby cortex stably grows and tunes its circuits to match environmental constraints. Such Laminar Computing completes perceptual groupings that realize the property of Analog Coherence, whereby winning groupings bind together their inducing features without losing their ability to represent analog values of these features. Laminar Computing also efficiently unifies the computational requirements of preattentive filtering and grouping with those of attentional selection. It hereby shows how Adaptive Resonance Theory (ART) principles may be realized within the laminar circuits of neocortex. Applications include boundary segmentation and surface filling-in algorithms for processing Synthetic Aperture Radar images.