Study On Stimulus-Induced Gamma Neuronal Oscillation And Its Cognitive Function

Gamma oscillations with high frequencies have been found in many brain regions,which are thought to be the intrinsic foundations of many cognitive functions such as the processing of sensory signals,learning and memory,feature binding,neural information coding and attention mechanism.The research on generation mechanism and cognitive function of Gamma oscillations is helpful for revealing the information processing mechanism of the brain's nervous systems,which is of great significance to the research of Brain-like Intelligence.It has been found that the stimulus signals with different properties such as image signals with different illumination contrast,mixed image-sound signals and stimulus signals with different sizes can induce Gamma oscillations with different frequencies in recent biology experiments,i.e.,the frequencies of Gamma oscillations are sensitive to change with the properties of the stimulus signals.However,the generation mechanism and the underlying information processing function of this kind of Gamma oscillation with its frequency dependent on stimulus(input)are still unknown.In order to reveal the underlying mechanism and the information processing function,this thesis focuses on two important issues by theoretical modeling and parallel simulation technique: the generation mechanisms of frequency-varied Gamma oscillations regulated by stimuli with various properties;the information processing function of frequency-varied Gamma oscillations regulated by stimuli.The main innovative work of this thesis is as follows:(1)In an excitability/inhibitory(E/I)neuronal network model regulated by simple stimuli,the mechanisms of Gamma oscillation induced by the unbalanced synaptic currents and the oscillation properties(oscillation frequency and oscillations power)regulated by the input difference of stimulus are revealed.Using simple input as the stimulus model,an E/I neuronal network model composed of Integrate-and-Fire neuron model and conductance-based synapse model is established.By simulations,Gamma oscillation is induced in the network and the phenomenon that the frequency and power of this Gamma oscillation change sensitively with the input difference between E-neurons and I-neurons is regenerated.Thereby,the generation mechanism of Gamma oscillation and the mechanism of it's properties sensitively dependent on the input difference of stimuli are revealed: Gamma oscillation is induced by the unbalance between positive and negative synaptic currents caused by the input difference between E-and I-neurons;and the larger the input difference is,the higher frequency and the stronger oscillation the network oscillation has.In addition,a simplified E/I network is used for mathematical analysis to further obtain a clear mathematical relationship between frequency and input difference,which provides a better way to understand the mechanism of stimuli-regulated Gamma oscillations.(2)In an E/I neuronal network model regulated by visual stimuli,the mechanisms of Gamma oscillation induced by illumination contrast and its oscillation properties regulated by illumination contrast are revealed.The receptive field model is introduced to imitate the response of real neurons in visual zones to external visual stimuli on the basis of the above research.An E/I neuronal network model with a grating visual stimulus based on Difference-of-Gaussians(DOG)receptive field model is constructed,which is closer to the real biological system.The grating stimulus with illumination contrast is mapped to E-and I-neurons with different receptive fields,so that E-and I-neurons have different external inputs resulting in the input difference.By simulations,Gamma oscillation is induced in this network and the phenomenon that the frequency and power of this Gamma oscillation sensitively change with the change of illumination contrast(input difference)is also regenerated.Thereby,the generation mechanism of Gamma oscillation and the mechanism of it's properties sensitively dependent on the illumination contrast are revealed: Gamma oscillation is induced by illumination contrast;and the higher the illumination contrast is,the higher frequency and the stronger oscillation the network oscillation has.The results further verify the biology experimental observations and provide a method for revealing the regulation mechanism of Gamma oscillations by visual stimulus properties.(3)The information processing function of Gamma oscillations regulated by stimuli is investigated by theoretical modeling.A new mechanism of the firing rate contrast enhancement is revealed: inhibitory synchronization in E/I neuronal networks that can cause Gamma oscillations acts as a global inhibition to enhance the firing rate contrast,which might facilitate information transmission in neuronal networks.Gamma synchronous oscillations with fixed frequencies are thought to be related to cognitive functions such as learning and memory,attention mechanism and feature binding.However,the information processing functions of frequency-varied Gamma oscillations dependent on stimulus properties may be different from the above functions,and need to be further revealed.Inspired by the well-known lateral inhibition in nervous systems which could increase the firing rate contrast(different external stimuli are encoded on the firing rates of different neurons;the higher firing rate contrast means that these different neurons have a better recognition effect on their corresponding external stimuli),a new mechanism of the firing rate contrast enhancement is revealed by inhibitory synchronization in E/I neuronal networks.The new mechanism is that synchronous inhibitory neurons act as a global inhibition to enhance the firing rate contrast of excitatory neurons globally in the networks,an hence improves information transmission of the networks.It is worthy of noting that this function of Gamma synchronous oscillations in E/I networks is not affected by variable frequencies.(4)In a large-scale complicated neuronal network with multi-layer columns,simulations are conducted by CUDA parallel technique and Gamma oscillations regulated by stimuli are explored.It is found that Gamma oscillations can also be regenerated and regulated by stimulus properties in large-scale neuronal network with complicated structure.According to the characteristics of actual biological nervous systems in the brain regions,a column structure with multiple network layers composed of multiple types of neurons is firstly constructed.Then a large-scale complicated neuronal network is established by using multiple columns.Based on CUDA parallel algorithm and a synapse optimization algorithm,a novel parallel algorithm is specially designed for simulation of this large-scale neural network with complicated structure.It is also found that Gamma oscillation induced by stimuli and the phenomenon that the frequency of this Gamma oscillation change with the stimuli difference in the large-scale complicated neuronal network with multi-layer columns.The research of this thesis provides a certain basis for revealing the generation mechanism of stimulus-induced Gamma oscillation and its underlying information processing functions.It is helpful for understanding of the relationship between brain cognitive functions and neuronal oscillations,and might have a great theoretical significance to the research on Brain-like Intelligence.