Frequency Spectrum of Coupled Stochastic Neurons with Refractoriness

Biological neurons are noisy and integrate their inputs over an extended period of time. On the other hand, the nervous system is able to transmit temporally-changing signals reliably. We address this problem with a homogeneous, fully connected model network. Neurons in our model have an absolute refractory period and fire with a stochastic rate that depends on the input they receive. An analytical solution of the noise spectrum of a population of such neurons is derived, which is in good agreement with simulations. We find that strong inhibitory couplings can considerably reduce the noise level in a certain frequency band. This allows the system to transmit signals with very low noise at frequencies close to the single-neuron mean firing rate.