Neural mechanisms of cortico-cortical interaction in texture boundary detection: a modeling approach.

Texture information is an elementary feature utilized by the human visual system to automatically, or pre-attentively, segment the visual scene. The neural substrate underlying human texture processing as well as the basic computational mechanisms remains largely unknown up to now. We propose a neural model of texture processing which integrates the data obtained by a variety of methods into a common computational framework. It consists of a hierarchy of bi-directionally linked visual areas each containing topographical maps of mutually interconnected cells. It builds upon the two key hypotheses that (i). texture segmentation is based on boundary detection and that (ii). texture border detection is mainly a function of higher visual cortical areas such as V4. This model, while attempting to explain the processing of textures, is embedded in a more general neural model architecture of the infero-temporal pathway of form processing.The model allows to link human performance in texture segmentation with model cell activation patterns, in turn permitting to trace back fundamental psychophysical results on texture processing to their putative neural origins. Most importantly, it enables us to identify and evaluate the functional role of feedback connections between cortical areas in the context of texture processing, namely the suppression of ambiguous cell activities leading to a sharply localized detection of texture boundaries. One of the likely neural origins of modulatory effects on V1 cell activation levels, as observed in electrophysiological studies using single- and multi-unit recordings, can be resolved.