Physics (Internal)

Heterogeneous Connectivity in E/I Network Allows Entrainment to a Wide Range of Frequencies

This event is part of the Preliminary Oral Exam.

Examining Committee: Nancy Kopell, Michelle McCarthy, Rama Bansil, Anushya Chandran

Abstract: Networks of excitatory (E) and inhibitory (I) neurons can produce rhythms and react to input with spectral content in multiple ways. Homogeneous PING networks are a type of E/I networks that produce robust gamma rhythms without noise. They can be entrained only to a narrow range of frequencies of a subthreshold periodic input. Here we investigate the response of such networks to subthreshold periodic input in the presence of noise and heterogeneity. Our focus is on understanding which features of the target network and the input allow for a large frequency range of entrainment by the periodic input. The target network consists of single compartment neurons modeled with the spiking currents (Na+, K+, leak) and with Hodgkin-Huxley-type conductances. The network has a tonic external input that generates a robust natural gamma rhythm at about 40-50 Hz. White noise is added to the E population. Heterogeneity is added in the AMPA synaptic conductance from E to I. We find that noise helps entrain a population of E cells: Without noise, the E population exhibits a resonance effect only when the input frequency is within 5Hz around the natural frequency. When noise is added the E population can be entrained to a much wider frequency range. However, in a PING network with feed-back inhibition from the I population, noise is not enough. With both noise to E population and heterogeneity in AMPA synaptic conductance from E to I added, we show that, for a wide range of input frequencies (~ 8 to 10-fold increase), the network can be entrained. We identify a novel mechanism for entrainment that involves frequency-dependent subgroups of I cells that are phase-locked to the input frequencies; those phase-locked I cells in turn strengthen the entrainment.