Entrainment in up and down states of neural populations: non-smooth and stochastic models (Q728541)

In this paper, the authors study a single recurrently connected excitatory population of neurons with negative feedback in the form of a slow adaptive current. A rate description is used. For a range of parameter values the system shows slow oscillations between ``up'' (active) and ``down'' (quiescent) states. The authors calculate the system's phase sensitivity function and use this to determine the effects of small deterministic and stochastic perturbations on the oscillations. They find that perturbations of the adaptation variable have much larger effects on oscillation phase than do perturbations of the activity variable. Introducing noise to the model decreases the oscillation period and balances the durations of the up and down states. The authors also study two uncoupled populations which receive noise, either common, or partially common and partially independent, and find that this synchronises the populations. The paper concludes with a demonstration that the effects found in the rate models also occur in equivalent spiking neuron models.