Systems and computational tools for neuronal retinal models

Abstract

This paper presents some systems theoretical and computational tools which allow for qualitative and quantitative models to explain the rate of firing (outputs) of ganglion retinal cells to various stimuli, mostly in higher vertebrates. Section 1 presents the neurophysiological bases of the formal tools, in which the main point is that specialized computation by some ganglion cells is to be performed at the inner plexiform layer, via amacrines. Section 2, presents the minima prerequisites for a qualitative model of linear and non-linear behavior of ganglia for higher vertebrates. The points here are that signals reaching the inner plexiform layer are fast and retarded versions of the input image, and that non-linear local lateral interaction accounts for non-linearities. Section 3, is devoted to computational representations that will permit to go from qualitative to quantitative formal models. The computational tools are based on generalized Newton Filters.