Population rate coding in recurrent neuronal networks consisting of neurons with mixed excitatory–inhibitory synapses

Abstract

Neural coding is a key problem in neuroscience aimed to understand the information processing mechanism in brain. Among the classical theories of neural coding, population rate coding has been studied widely in many works. In computational studies, neurons are usually classified into excitatory or inhibitory ones. Excitatory neurons have excitatory output synapses, and inhibitory neurons have inhibitory output synapses. However, according to physiological observations, neurons potentially have both types of output synapses. Thus, in this paper, neuronal networks consisting of neurons with mixed excitatory–inhibitory synapses are constructed to investigate the population rate coding fidelity of neuronal systems. It is revealed that, under intermediate values of recurrent probability, inhibitory–excitatory strength ratio, and noise intensity, the performance of population rate coding could be improved by both excitatory synaptic strength and synaptic time constant. It is indicated that external stimuli can be encoded in the form of population firing rate by the studied neuronal networks very well. What is more exciting is that we find the neuronal networks considered in our work have higher coding efficiency than the traditional ones. Therefore, neurons with mixed excitatory–inhibitory synapses may be much more rational.