Figure 1

Stochastic hopping between metastable states improves accuracy of choice. (A) Results of 100 simulations with two mutually inhibiting populations, each of 100 neurons and fixed noise. Current is applied to each group, with a bias to favor one outcome. With low applied current, both populations would remain inactive (undecided state) in the absence of noise, so a change in state corresponds to stochastic hopping (B). With greater applied current the inactive (undecided) state is unstable so a deterministic drive (C) causes one population to become active.