Stochasticity, Coding, and Dynamics in Continuous-Parameter Working Memory

How does neural stochasticity influence the dynamics and function of neural circuits? This question is often difficult to answer, since both the sources of stochasticity and the functional constraints on the system are not sufficiently well characterized. Here I will focus on a fairly well defined task which is highly sensitive to neural noise: the retention of a continuous parameter in short-term memory. I will present several works that address related questions: is there a relationship between the neural coding of the stored parameter by spikes emitted within a memory network, and the fidelity at which the network can maintain a representation of the stored parameter over time? Does it matter whether the noise arises intrinsically within neurons and synapses, or due to the chaotic dynamics that characterize cortical networks in an asynchronous, balanced state? Is it possible to point to concrete neural circuits in which the above principles govern the dynamics of memory retention?