Visual shape discrimination in goldfish, modelled with the neural circuitry of optic tectum and torus longitudinalis

Abstract

There is no satisfactory neurally-based theory as to how vertebrates that lack a neocortex discriminate even simple geometric shapes. In fishes, an intact optic tectum is necessary for such discriminations, but physiological studies of it have found nothing like the hierarchically arranged feature detecting neurons of mammalian visual cortex. Here, a neural model attempts a solution by basing shape discrimination upon the responses of only those elementary detectors (e.g. of size) that are within a focus of attention, formed by a winner-take-all arrangement of retinotopically mapped units representing tectal pyramidal cells. While this relatively primitive mechanism could recognize an object irrespective of position in space, it fails to distinguish patterns that differ only in their features’ spatial relationships. The model’s solution - imitating goldfish that naturally attend to the top of shapes - is to shift attention to the edges of a shape by spatially offsetting inputs to the pyramidal neurons, effected by the torus longitudinalis and its prolific synapses on pyramidal dendrites. The model’s shape discrimination was compared to an extensive behavioral study using shapes with points and projections. In one test series fish were sensitive to the relative number of points on the tops of shapes. In another, fish were trained to discriminate points on the sides. By using different offset connections and only one elementary feature detector for small dark spots, the model successfully emulated the two sets of goldfish data, as judged by significant correlations between model response and fish discrimination.