Nonlinear Network Models of the Oculomotor Integrator

Abstract

The neural integrator of the oculomotor system is modeled as a network of spiking, conductance-based model neurons. The static function of the integrator, holding the eyes still when the head is fixed in space, is the focus of the modeling. The synaptic weight matrix, which is of outer product form, is tuned by minimizing the mean squared drift velocity of the eyes over a range of eye positions, leading to an approximate line attractor dynamics. The conductance-based model is reduced to a rate-based one to simplify the tuning procedure.