Effect of inhibitory firing patterns on information transmission in feedforward neural networks

Abstract

The feedforward neural network is a general structure of information transmission in the nervous system. It is widely used to simulate the transmission characteristics of neural information in the cortical neuronal networks. It is generally believed that cortical neurons conduct neural information through the firing rate. In this paper, a three-layered feedforward neural network is constructed based on the Izhikevich neuron model to explore the influence of inhibitory firing patterns on the information transmission of cortical networks. Numerical results show that, the number of inhibitory bursting neurons and inhibitory synaptic connection strength promote information transfer, but have different effects on the delivery of information between different layers. The results show that the most efficient transmission of information in the feedforward neural network can be achieved by adjusting the E/I balance, the number of inhibitiory bursting neurons and the input current.