Signal-to-noise ratio gain of a noisy neuron that transmits subthreshold periodic spike trains

Abstract

The transmission properties of an integrate-and-fire neuron model that transmits coherent subthreshold spike trains in a shot noise environment are investigated by numerical simulation. For very weak coherent couplings, it is shown that the input-output signal-to-noise ratio (SNR) gain is easier to exceed unity; while for stronger coherent couplings it is difficult to observe the SNR gain larger than unity at the optimal noise intensity. These observations are different from those acquired in the case of continuous noise. Our analysis further suggests that the larger SNR gain in the very weak coherent coupling case should be due to the noise-induced resonance. It is also shown that there is more possibility of the SNR gain above unity for slower periodic spike trains transmitted by the model. The results may be useful in understanding the performance of real noisy neurons acting as signal-processing elements.