Real neural networks in movement control

Biological control systems use neurons and muscles. Most use thousands of neurons and dozens of muscles and we despair of understanding them in any reductionist sense. There do exist, however, a few neural circuits that we "more-or-less" understand. One is the vestibulo-ocular reflex that stabilizes our eyes in space by compensating for head movements. Investigation soon reveals the existence of a neural network that integrates velocity commands (coded in neural firing rate) to produce eye-position commands. Thus we converge on the problem of how a group of neurons can connect themselves up to perform integration in the sense of Newtonian calculus. This paper reviews current research on this specific, if rare, example of how the brain processes signals.