Structural study of the development of ocularity domains using a neural network model.

We present a model for the development of ocularity domains in the visual cortex of mammals during the embryonic stage. We model the thalamo-cortical pathway with a self-organising neural network with two source layers, each of them serving different retinae, and one target layer, where the connections end. The connectivity between the source layers and the target layer is driven by Hebbian learning. In both the source layers and the target layer we assume excitatory lateral signal diffusion between proximal neurons that causes them to be correlated. According to the developmental state being modelled, we do not consider either correlation or anti-correlation between the signals originated in neurons of different retinae. The basic assumptions made are proved to be sufficient to attain a distribution of connections arranged in ocularity domains. The dependence of the geometry of the ocularity domains on the parameters of the model is analysed and a correlation between the width of the signal diffusion and the extent of the domains is found. The generality of the assumptions made allows an easy translation of the model to explain the development of other elements of the sensory nervous system.