Cerebellar Learning Models Of Vestibulo-Ocular Reflex

With the emergence of research on brain science,the cerebellum has received world-wide interests from researchers in the fields of neuroscience,medical and computer science due to its important role in motor learning and behavior control.As a fast reflex,VestibuloOcular Reflex(VOR)induces an eyeball movement opposite to the direction of head movement to stabilize the imaging on the retina.Although the existing VOR learning models simulate the VOR circuit and its learning characteristics to some extent,they fail to consider the teaching function of climbing fiber in VOR learning.Therefore,it is of great importance to study the cerebellar learning models of VOR.This thesis makes an extensive study on the cerebellar learning models of VOR.The main contents are as follows.(1)It is found that complex spikes of Purkinje cells,which are transmitted by climbing fibers,encode error signals in the VOR circuit.However,how such error signals guide the cerebellar learning is still controversial.Therefore,this thesis studies the existing cerebellar learning models and proposes a plasticity learning model based on error feedback modulation.Following this,a staged non-uniform learning model is proposed by considering the plasticity rules of memory sites.Simulation results show the proposed model better coincides with physiological experimental results and demonstrates that proper interval training could promote the formation of long-term memory.(2)VOR learning involves two patterns,VOR-decrease and VOR-increase,and they are not accurately reversible.To investigate the learning characteristics of VOR-decrease,this thesis presents a VOR-decrease,learning model.By studying the contribution of synaptic plasticity in memory sites to adaptive learning,the proposed model analyzes the learning characteristics of mutant mice.Simulation results show that the proposed model not only simulates the physiological experimental results of wild mice but also illustrates the learning characteristics of mutant mice.In conclusion,this thesis focuses on the research on the cerebellar learning of VOR and proposes the error feedback modulation based plasticity learning model as well as the VORdecrease,learning model.It makes an effective analysis of the teaching function of climbing fibers in VOR learning and reduces the difference between the theoretical models and physiological experimental results.