Analysis and Simulation of Cerebellar Circuitry

Abstract: The cerebellum, a fist-sized structure of the brain, plays a crucial part in the execution and coordination of motor control tasks and cognitive activities. It is also remarkably able to adapt itself to new tasks and circumstances whenever errors in motor output are made. Over the years, valuable insight on cerebellar functionality has been gained through the application of concepts traditionally associated with control theory. In this thesis, neural spike train data recorded from cerebellar neurons at rest under in vivo conditions were examined. The goal is to give an initial characterization of the spike firing properties of such neurons, using statistical point process theory. The findings indicate that the distributions of inter-spike intervals are skewed, and often could be approximated with either a gamma distribution or a lognormal distribution. Furthermore, it is found that several spike trains could be reasonably treated as renewal processes, while others require more complex description of their inter-spike interval dependency. Secondly, a Matlab model was developed to simulate a structure of cerebellar contribution to motor control of the vestibulo-ocular reflex. The cerebellar output is feedforwarded to a linear model of the oculomotor plant. A comparison between head velocity and eye velocity serving as a teaching signal to the cerebellum enables it to alter its functionality to better compensate for head movements. A simulation of a cerebellar circuitry output reveals that feedforwarding of the cerebellar output to the oculomotor plant through learning is able to improve the control performance, and offers a plausible explanation for motor control of the vestibulo-ocular reflex.