Modeling visual cortex: Hidden anisotropies in an isotropic inhibitory connection scheme

Abstract

A detailed cortex model (15,000 cells) of the adult cat is presented and it is shown that a combination of unspecific inhibitory mechanisms together with input of aligned receptive fields from the LGN with little elongation (1:1.5) is able to reproduce cortical orientation selectivity and other features of cortical cell behavior. We introduce a novel isotropic intracortical connection scheme (“circular inhibition”) and demonstrate analytically that this mechanism results in two anisotropies: orientation tuning and a directional bias. Thus, our network shows that structurally unspecific isotropic connections can result in functionally specific behavior. Directional anisotropy introduced in this way could be the starting point for the development of the true direction specificity found in cortical cells.