Role of slow temporal dynamics in the reliable activity of stochastically driven neurons

Abstract Neuronal networks maintain robust patterns of activity despite a backdrop of noise from various sources. Mutually inhibiting neurons is a standard network motif implicated in rhythm generation. In a 'building-block' network motif of two mutually inhibiting neurons capable of swapping from an active state to a quiescent state, we ask how different sources of stochasticity alter firing patterns. In this system, the alternating activity occurs via the combined action of a calcium-dependent potassium current, sAHP (slow afterhyperpolarization), and a fast GABAergic synapse. We show that one of the payoffs of extrinsic noise is extending the dynamical range of neuronal firing. Extrinsic noise also has the effect of increasing the switching frequency via a faster build-up of sAHP current. The switching frequency as a function of input current has a non-monotonic behavior. Interestingly, the noise tolerance of this system varies with the input current. It shows maximum robustness to noise at an input current that corresponds to the minimum switching frequency between the neurons. The slow decay time scale of sAHP conductance allows neurons to act as a low-pass filter, attenuates noise, and integrates over ion channel fluctuations. We also show that the slow inactivation time of the sAHP channel allows the neuron to act as an action potential counter. We propose that in this system, the slow decay time scale of sAHP conductance, an intrinsic property of the current, allows the network to maintain robust rhythmic activity critical for various functions despite noise.