神经元集群自持续放电活动的层次网络仿真模型与分析方法

Hierarchical Neural Networks Stimulation Model of Neurons Cluster Self-sustained Firing Activity and It's Analysis Methods

神经元集群的自持续放电活动是大脑内广泛存在的现象,其被证实在大脑的工作记忆与目标导向等行为中有重要体现。作者以非线性的整合发放(integrate-and-firing,IF)神经元模型为网络节点,构建了具有小世界特征的层次网络仿真模型,以研究自持续活动中神经元发放的一些特性。在合适的模型参数下,层次网络能产生自持续放电活动,其整体发放频率在撤掉外部刺激之后的20 s内比较稳定,而层次内部发放频率的高低与层次顺序无关。整体发放频率关于突触连接数量与短路径密度都呈现出先正关系增长再达到饱和的趋势,同时,规模越大的神经元网络的整体发放频率对短路径密度更为敏感。研究结果对揭示大脑神经元功能性核团之间的相互作用机制具有重要意义。

Self-sustained firing activity of neurons cluster is a widespread phenomenon in the brain, and is confirmed that it has important manifestation in working memory and goal-directed behavior and so on. In this paper, with the nonlinear Integrate-and-fire （IF） neuron model as network node, the authors built the hierarchical modular networks of spiking neurons, combined with the characteristics of small-world network, to study some properties of neurons firing during self-sustained activity. Under the appropriate model parameters, the hierarchical network could produce self-sustained firing activity, as its overall firing rate was stable in the previous 20 s after removal of external stimuli, while the firing rate of inner hierarchical network was irrelevant of its order. The overall firing rates with respect to the number of synaptic connections and shortcut density showed a trend of growing during the early period and then getting saturated. At the same time, the bigger the scale of neural network was, the overall firing rate was more sensitive to shortcut density. These results are of great significance to reveal the interaction mechanism between functional nucleuses in the brain.