Anisotropic Correlation Properties in the Spatial Structure of Cortical Orientation Maps

Abstract

Experimental and theoretical studies on visual information processing have evidenced the advantages of a delocalized representation of information through feature maps which mix topological contiguity with proximity in the feature space. The peculiar dispositions of cortical orientation-selective cells can be understood in terms of dimension-reducing mappings which translate neighborhood relations in the orientation-subspace in spatial neighborhood relations on the cortical surface [Durbin and Mitchison, 1990]. Thus, cortical maps are not only a repository of information about the features present in the image, but are the substrate for coordinated interactions among features. The basic computational principles of these maps can be investigated with different approaches (heuristic, theoretical information, computational). In this paper, we’ll provide information-theoretic insights about spatial arrangements of simple cell receptive fields in several biologically plausible orientation maps, by studying both autocorrelation and directional mutual information of cortical activity.