Mireille E. Broucke | Department of Electrical Engineering and Computer Engineering, University of Toronto
A Control-Theoretic Model of the Oculomotor System and the Cerebellum
One of the great open problems of neuroscience today is to obtain a computational model of the cerebellum, the part of the brain responsible for motor control. A mathematical model of the cerebellum would contribute fundamental knowledge to science (analogous to Kepler’s law in the motion of the planets), it would reduce the dependence on monkeys for experimental brain research, and it would propel a deeper understanding of neurological disorders involving the cerebellum such as ataxia, Huntington’s disease, and Parkinson’s disease.
Prevailing theories interpret the cerebellum as providing either a forward model or an inverse model of the part of the body being controlled (e.g. the eye, the arm, one's posture, one's gait, etc). Remarkably, the internal model principle of control theory has not been promoted as the driving principle of cerebellar function. In this talk I put forward a theory of cerebellar function based on the internal model principle. I argue that the internal model principle must be embodied at some location in the brain and that the architecture of the cerebellum is ideally situated in terms of its connectivity to other brain areas for this function.
The talk will begin by reviewing the neurological structure of the cerebellum, the components of the oculomotor system, and the system's slow eye movement behaviours: the vestibuloocular reflex, the optokinetic reflex, gazing holding, and smooth pursuit. Next, we present existing computational theories on cerebellar function based on forward and inverse models. With this background in place, we proceed to lay out a new control-theoretic model of the oculomotor system and the cerebellum. The model is validated by simulation, recovering behaviours from over 15 oculomotor experiments. We map the signals of the model to the oculomotor neural circuit, and we indicate where discrepancies still exist. Finally, we conclude with open problems in the area.
The talk is aimed at non-experts of neuroscience with an undergraduate-level knowledge of control theory.