A Model of Position-Invariant, Optic Flow Pattern-Selective Cells

Abstract

Two apparently inconsistent proposals for the functional architecture of MST cells exist, based either on a set of cardinal flow patterns [1] or a continuum of flow patterns [2]. A model is proposed by [I) that consists of excitatory and inhibitory subunits to support their architecture. Here, we used a neural network to show that a model of this type can be selective to the continuum of patterns [2]. We extended this model by introducing inhibitory connections between cells to account for the position-invariant characteristics of MST cells reported in [2]. References