Stochastic neural dynamics

How can neural networks reliably process information in spite of biological noise? Can the same neural assemblies exhibit different coding strategies? How do network structure and input drive combine to explain the adoption of a given coding strategy? More fundamentally, can a meaningful neural computation be distinguished from—perhaps irrelevant— spontaneous neural activity? In other words, do neural computations have a physical, observable signature?