Characteristic Analysis Of Retinal Neural Circuit With Spike-timing Dependent Plasticity

Retina is the first station for the collection and processing of visual information,with a distinct hierarchical structure in physiological anatomy.The information transmission pathways contained in retinal neural circuit is extremely complex,but it can be simply understood as parallel transmission of visual information through a large number of parallel pathways,combined with horizontal adjustments to the processing of information.Visual information are mainly converted,primary processed and collected by the structural and connective characteristics of retina.Modeling the retinal neural circuit based on its structure and connection helps to simulate and analyze the process of signal conversion and information processing.STDP(Spike-Timing Dependent Plasticity)is a kind of synaptic plasticity learning rules that are closely related to the development of the brain and its learning and memory functions.The innovative application of STDP rules in the retinal neural circuit model will help to explore the characteristics of it in the processing,learning and memorizing for graphic information.We have constructed a retinal neural circuit model that includes cones,horizontal cells,bipolar cells,AII amacrine cells,ganglion cells and lateral geniculate nuclei,as well as ON pathways and OFF pathways,based on the physiological anatomic structure of retina)STDP rules have been introduced into the synaptic connections between the ganglion cells layer and the lateral geniculate nuclei layer.By adding single pattern stimulation and alternating pattern stimulation,we have studied the information transfer characteristics of retinal neural circuit through comparing the potentials and frequencies of both ganglion cells and lateral geniculate nuclei,and the changes in synaptic weights between of them,and then obtained several conclusions.First,the retinal neural circuit model can effectively convert the light intensity information received by the cone cells into the timing information of the ganglion cells,and show the information structure characteristics of retina)Second,the application of STDP rules makes the lateral geniculate nuclei layer receive the corresponding stimulation pattern and learn and memorize this pattern,and there are learning independence between ON and OFF pathways.Third,the learning mechanism of STDP rules is to reflect the graphical pattern in the spatial distribution of synaptic weights.Finally,for the alternate graphical stimuli,STDP rules can make the two patterns that appeared alternatively to be superimposed on the spatial distribution of synaptic weights,and the learning effect of overlapping parts is more significant.