Coherence resonance for time-averaged measures

TL;DR

This paper demonstrates that coherence resonance in neuron models can be characterized using time-averaged measures, showing their equivalence to ensemble-averaged measures in stationary states.

Contribution

It provides numerical evidence that time-averaged coherence measures are independent of noise samples and equivalent to ensemble averages in neuron models.

Findings

Time-averaged coherence measures are independent of noise sample paths.

In stationary states, neuron dynamics are uniquely determined by noise samples.

Coherence resonance applies to both ensemble- and time-averaged measures.

Abstract

Noise can induce time order in the dynamics of nonlinear dynamical systems. For example, coherence resonance occurs in various neuron models driven by a noise. In studies of coherence resonance, ensemble-averaged measures of the coherence are often used. In the present study, we examine coherence resonance for time-averaged measures. For the examination, we use a Hodgkin-Huxley neuron model driven by a constant current and a noise. We firstly show that for large times, the neuron is in a stationary state irrespective of initial conditions of the neuron. We then show numerical evidence that in the stationary state, a given noise sample path uniquely determines the dynamics of the neuron. We then present numerical evidence suggesting that time-averaged coherence measures of the dynamics is independent of noise sample paths and is equal to ensemble-averaged coherence measures. On the basis of this property, we show that coherence resonance is not only a phenomenon related to ensemble-averaged measures but also a phenomenon that holds for time-averaged measures.