神经回路的动力学现象及脑决策任务仿真

发布时间：2018-06-02 20:15

本文选题：模型 + 神经回路　；参考：《华南理工大学》2016年博士论文

【摘要】：大脑的学习,记忆和认知等功能是由相互作用的神经元构成的神经回路来完成.因此,神经回路的理论研究对理解神经系统的功能有着至关重要的作用.本论文对局部生物神经回路,动物海兔的尾部收缩反射回路和额叶视区决策回路进行了动力学分析以及仿真模拟,具体工作如下第一章是绪论,主要介绍本论文的研究背景,神经回路的动力学现象和模型仿真的研究框架,并提出本文的主要课题和研究内容.第二章介绍模拟神经回路所需的研究基础和研究背景.这些内容主要包括神经元的解剖结构和数学模型,离子通道及其数学模型,神经元的兴奋性,神经元间的突触连接以及它们的数学模型,突触可塑性和神经回路的拓扑结构以及一些经典的神经回路模型和神经回路研究的最新进展.神经元模型包括Hodgkin-Huxley模型,Connor-Stevens模型,Morris-Lecar模型和呼吸神经元模型四个基于电导的模型以及整合发放模型,还有Rall电缆模型及其离散化后得到的多房室模型.论文用到的突触类型有电突触和化学突触.第三章研究生物局部神经回路,主要讨论耦合呼吸神经元的相位同步.论文首先用电突触将两个呼吸神经元连接起来,建立了一个耦合呼吸神经元模型.通过对比单个神经元模型和耦合神经元模型的门控变量的极大值分岔图,发现耦合神经元模型表现出更复杂的发放变化模式.特别是改变耦合强度和膜电容后,耦合系统出现不同的相位同步状态以及它们之间的转迁.更进一步,在耦合强度和膜电容的二维平面上,我们给出了同步状态的分布情况,并总结了同步状态转迁的规律.第四章研究动物神经回路,主要通过对神经回路的仿真研究海兔尾部收缩反射现象.先得到感觉神经元的波形特征,然后通过改变刺激强度和受刺激感觉神经元的个数得出反射回路的典型规律.此外,反射神经回路与肌纤维模型连接后得到新的模型,并成功模拟了神经信号与肌纤维间的兴奋-收缩耦联.最后,论文还探讨了突触可塑性对停止刺激后反射回路-肌纤维连接模型的长持续反应时间的影响.这些模型结果揭示海兔尾部收缩反射的特性.第五章研究脑决策的动力学机制,基于一个改进的额叶视区模型,对三类脑决策任务进行仿真研究.首先,在一个额叶视区模型中加入了神经元群体的方向偏好性,规则控制模块和基于奖励的可塑性突触,使其变为一个基于学习的额叶视区模型.修改后的模型经过训练后成功模拟了三个脑认知决策任务：反向眼跳任务,no-go任务和关联任务.训练后的模型表现出一些与猴子实验一致的性质,如反向眼跳的反应时间分布的延迟,停止信号取消反射性眼跳的机制和半极大选择性延迟的变化.此外,对训练后的模型进行切换任务训练,即在没有提示的情况下,让回路模型在两个任务中进行切换并重新训练,结果表明模型经过一个重新学习的过程可以在正向眼跳任务和反向眼跳任务间以及不同的提示-眼跳关联间切换.这些额叶视区回路的模型结果揭示脑决策任务的内在决策过程和认知规律,为理解大脑的认知功能提供理论基础.
[Abstract]:The learning, memory, and cognition of the brain are accomplished by a neural circuit composed of interacting neurons. Therefore, the theoretical study of the neural circuits plays a vital role in understanding the function of the nervous system. This paper is a part of the local neural circuit, the tail contraction reflex loop of the animal sea hare and the decision loop of the frontal lobes. The first chapter is the introduction, which mainly introduces the research background of this paper, the dynamic phenomena of neural circuits and the framework of model simulation, and puts forward the main topics and research contents of this paper. The second chapter introduces the research foundation and research background of the simulated divine loop. The main contents include the anatomical structure and mathematical models of neurons, ion channels and their mathematical models, excitatory properties of neurons, synaptic connections among neurons and their mathematical models, synaptic plasticity and topological structures of neural circuits, and the latest advances in some classical neural circuits and neural circuits. The Hodgkin-Huxley model, the Connor-Stevens model, the Morris-Lecar model and the respiratory neuron model are four based on the electrical conductivity model and the integrated distribution model, and the Rall cable model and the multiple atrioventricular model after the discretization. The synaptic types are used as the electrical synapse and the chemical synapse. The third chapter studies the biological local nerve. In this paper, the phase synchronization of coupled respiratory neurons is mainly discussed. Firstly, two respiratory neurons are connected by electric synapses, and a coupled respiratory neuron model is established. By comparing the maximum bifurcation diagram of the single neuron model and the gated variable of the coupled neuron model, it is found that the Coupled Neuron Model is more complex. In particular, after changing the coupling strength and membrane capacitance, the coupling systems have different phase synchronization states and their transition between them. Further, on the two dimensional plane of the coupling strength and membrane capacitance, we give the distribution of the synchronous state and the law of the transition of the synchronous state. The fourth chapter studies the movement. The neural circuit is used to study the contractile reflex in the tail of the rabbit by simulation of the neural circuit. The characteristics of the waveform of the sensory neurons are obtained first, and then the typical law of the reflex loop is obtained by changing the intensity of stimulation and the number of stimulated sensory neurons. In addition, a new model is obtained after the reflex neural circuit is connected with the muscle fiber model. Finally, the effect of synaptic plasticity on the long sustained response time of the muscle fiber connection model after the stop stimulation is also discussed. The results reveal the specificity of the contractile reflex in the tail of the rabbit. The fifth chapter studies the dynamic mechanism of brain decision-making. In a modified frontal visual area model, three types of brain decision-making tasks are simulated. First, the orientation preference of the neuron group is added to a frontal visual area model, the rule control module and the reward based plasticity synapse become a learning based frontal visual area model. After training, the modified model is trained. Three brain cognitive decision-making tasks were successfully simulated: reverse saccade task, no-go task, and associated task. The model after training showed some properties consistent with monkey experiment, such as the delay of reaction time distribution of reverse saccade, the mechanism of stopping the reflex saccade and the change of semi maximum selective delay. The model is trained for switching tasks, that is, to switch and retrain the loop model in two tasks without prompting. The results show that the model can switch between the forward saccade task and the reverse saccade task and the different hint - saccade association through a relearning process. The results of the model reveal the internal decision-making process and cognitive rules of the brain decision-making task, and provide a theoretical basis for understanding the cognitive function of the brain.
【学位授予单位】：华南理工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：B842

【相似文献】