Short-term Synaptic Plasticity Regulates The Bistability And Memory Effect In Neural Networks

Working memory is a unique memory ability of human and animal,which is crucial to higher cognitive functions such as reasoning,understanding and learning.During the working memory task,the prefrontal cortex(PFC)of the brain shows persistent activity and intermittent different spiking rate state.The bistability of neural networks may provide the functional basis for this persistent activity.Studies have shown that shortterm synaptic plasticity can be used as a mechanism for working memory and as a nonlinear factor in dynamics,resulting in the bistability of neural networks activity.However,the mechanism of short-term synaptic plasticity regulating the bistability and memory effects of neural networks is still not very clear.In this thesis,we address this question by combining both the theoretical derivation of spiking rate model and computational modelling of working memory network.The main obtained results are briefly summarized as follows:Firstly,the bistability phenomenon under the regulation of short-term synaptic plasticity is studied by theoretical derivation.Here we argue that in working memory the PFC operates in a regime of balanced excitation and inhibition,and the short-term synaptic plasticity is introduced into the balance state equation.By analyzing the form of the solution and phase diagram of the excitatory neuron population spiking rate equation,the dynamic state of the neural network is discriminated.It has been proved that when the system parameters are in the bistable region of the phase diagram,the short-term synaptic plasticity can induce the bistable switching phenomenon of the spiking rate model under appropriate external stimuli.Secondly,a working memory network model based on short-term synaptic plasticity is further constructed,and the bistability and working memory effects of the network are systematically studied.Considering that the excitatory projections between the excitatory nerve groups have short-term synaptic plasticity,it is found that appropriate external stimuli can induce the network to work in the bistable region,and the network activity is in the excitation-inhibition balance state in this region,and the excitatory and inhibitory currents of neurons are linear.This provides a basis for exploring the memory effects involved in working memory tasks.Thirdly,the working memory network shows different encoding effects under different external stimuli.The spiking activity of the memory location increased with the stimulus duration and intensity.When the stimulus exceeded a certain threshold,the working memory network showed a persistent memory effect,which could be erased by another appropriate reverse stimulus.The weak periodic signal stimuli added by the baseline and persistent activity state show that the system produces resonance phenomena when the signal-to-noise ratio is at its maximumOur research suggests that short-term synaptic plasticity may be a nonlinear mechanism for generating the bistability in neural networks.This bistability regulated by short-term synaptic plasticity provides a persistent memory effect for the brain to perform working memory tasks.In addition,the working memory networks have different encoding effects under different external stimuli.These results well explain some of the existing experimental observations of working memory tasks and help us to further understand the complex dynamics of the nervous system,which is of great scientific significance.