A note on the responsiveness of neurons

Abstract

The time constant of biological neurons is typically around 5-20ms. This raises the question of how a system made of slow devices can generate and respond to faster signals. A potential solution is a population of uncorrelated neurons that, as a whole, can respond arbitrarily fast despite individual slow neurons. The response speed of a neuron or a population of neurons is often defined as how quickly neurons increase or decrease their firing rate given a change in the input, similar to a reaction time. Less emphasis is placed on the recovery time - the duration to transition and settle in a stationary state again in response to the change in the input. Here, we investigate the reaction time and the recovery time of leaky integrate-and-fire neurons given a change in, either the mean input drive, or input fluctuation level. In general, neurons react faster and fire quicker the larger the change in the input. Moreover, the closer the membrane potential is to the threshold, the faster the reaction. We observe that neurons react faster to increased input fluctuations than to stronger mean drive. Complementarily, the recovery time is systematically longer the larger the input change. Also, the closer the neurons' potentials are to the threshold, the longer it takes the neuron to recover. Most interestingly, the baseline firing of the neuron strongly determines whether recovery is faster when the change happens in the mean drive or the fluctuation level. This observation suggests that communication via the mean drive and the fluctuation level could target different aspects of the overall response - quick reaction time and the recovery speed - based on the initial baseline activity.