Excitation-inhibition Balance And Information Transmission In Neural Networks Underlined By Inhibitory Synaptic Plasticity

The excitation-inhibition balance(E-I balance)of cortical neurons plays a crucial role in maintaining the normal electrical activity and achieving the normal physiological functions of the brain neural network.On the other hand,the occurrences of neurological diseases such as epilepsy and Parkinson's disease are closely related to excitation-inhibition imbalance of the nervous system.Information transmission between different brain regions is the basis for the cooperation of various brain regions to complete various advanced cognitive functions.Therefore,the study on mechanisms of the E-I balance and the information transmission of neural networks is of great significance for understanding the mechanism of neural electrical activity,advanced cognitive functions and neural diseases in brain.Most studies on E-I balance mainly focus on the E-I balance of homogeneous neural networks.However,neural anatomical researches revealed that the structure of the neural network is heterogeneous,which exerts an important effect on E-I balance of neural network.Therefore,the E-I balance of heterogeneous neural networks still remain to be addressed.The performance of information transmission is dependent on the network activity states affected by the E-I balance,thus the E-I balance has an important influence on the information transmission in neural networks.Related researches mainly exploited the influence of inter-layer feedback connections on the information transmission of feed-forward neural networks.However,it is still unclear how the balance of intra-layer excitatory and inhibitory feedback connections exerts an effect on the information transmission performance of the feed-forward neural networks.Therefore,this present study constructs homogeneous and heterogeneous feedback neural networks models and feed-forward neural network model respectively,and aims to probe on the mechanism of E-I balance of heterogeneous neural networks with inhibitory synaptic plasticity,as well as the regulation mechanism of information transmission of the feed-forward neural network underlined by the intra-layer balance of excitatory and inhibitory feedback.The contents are as follows:(1)To study the mechanism of E-I balance of heterogeneous neural network.Firstly,we construct homogeneous and heterogeneous neural networks models,to study the balance of excitatory and inhibitory synaptic currents of target neurons in the two types networks with feedback connections.Balance measures,including synaptic currents,neuronal membrane potentials,spike rasters and firing rate charts,are employed to compare the state of E-I balance between the homogeneous and heterogeneous neural networks.Secondly,we incorporate inhibitory synaptic plasticity into the heterogeneous neural network and the E-I balance degree is evaluated by the sum of synaptic currents,the ratio of synaptic currents,the scatter plot of synaptic currents,membrane potentials,spike raster,firing rate chart,respectively.Thirdly,we analyze the robustness of E-I balance in heterogeneous networks in the face of external disturbances and parameter perturbations,respectively.The results show that the excitatory and inhibitory synaptic currents of neurons in the homogeneous neural network can achieve E-I balance only with feedback connections,while the structural heterogeneity of the heterogeneous neural network disrupts the E-I balance.Further studies reveal that,by introducing inhibitory synaptic plasticity into the heterogeneous neural network,the balance of excitatory and inhibitory synaptic currents of all target neurons can be achieved.Further studies show that the E-I balance has a strong robustness to external disturbances and parameter perturbations.In addition,we discuss the effect of the frequency of Poisson input on the E-I balance of homogeneous neural network.The results show that,the large-frequency Poisson input breaks the E-I balance of the homogeneous neural network,and the E-I balance of the homogeneous neural network with the inhibitory synaptic plasticity can be re-established.(2)To study the mechanism of information transmission in the feed-forward neural network with intra-layer E-I balance.The feedback neural network is used as subnetwork for each layer to construct a feed-forward neural network model.The states of excitation and inhibition of intra-layer neural networks are changed by adjusting the excitatory and inhibitory synapse strength of each intra-layer neural networks respectively,and then we analyze the performance of information transmission of the feed-forward neural network under different intra-layer states of excitation and inhibition.Firstly,we evaluate the performance of synchronous spiking propagation in the feed-forward neural network under different intra-layer states of excitation and inhibition with the total number of spikes of pulse packets in each layers,the standard deviation of spike times of the pulse packets,survival rate and signal to noise ratio,respectively.Secondly,we analyze the performance of firing rate propagation of the feed-forward neural network under different intra-layer states of excitation and inhibition with the correlation coefficient between the average firing rate of neural population in each layer and the external current.The results show that the intra-layer networks have obviously different electrical activity states under different intra-layer states of excitation and inhibition,and the intra-layer excitation and inhibition states can regulate effect on the synchronous spiking transmission and the firing rate transmission in the feed-forward neural network.With strong intra-layer excitation,the feed-forward neural network performs an unstable synchronous spiking transmission and the performance of firing rate transmission is poor;on the other hand,with strong intra-layer inhibition,the feed-forward neural network performs an failed synchronous spiking transmission and the performance of firing rate transmission is also poor;therefore,when the intra-layer neural network operated at the balanced state,the feed-forward neural network achieves optimal information transmission performance.The present study shows that inhibitory synaptic plasticity is an effective mechanism to establish and maintain E-I balance in neural networks;furthermore,the balanced state in each layer allows the feed-forward neural network achieve optimal information transmission performance.The results provide insight into the E-I balance mechanism of heterogeneous neural networks and the regulation mechanism of neural information transmission,and further shed light on understanding the mechanisms of neural electrical activity,brain function and diseases in brain,respectively.