Heavy tails in the brain

In my talk I will discuss the relation between two seemingly unrelated measures in the brain that exhibit heavy tails: neuronal avalanches, i.e. bursts of activity with power-law distributions of sizes and lifetimes, and synaptic weights that are believed to be distributed according to the log-normal distribution. Many current models of neuronal avalanches do not rely on heavy-tailed synaptic weight distributions, suggesting that heavy tails of these two quantities may not be related. However, our recent theoretical considerations indicate that this independence no longer holds if two biologically relevant constraints are introduced, i.e., that neurons (1) receive many incoming connections and (2) do not spike if the membrane potential is below some positive threshold, e.g., in the absence of inputs. Under these assumptions we have shown that heavy tails of synaptic weights are necessary to generate biologically plausible low activity levels and associated neuronal avalanches. Our results suggest that the observed distributions of synaptic weights may play important functional roles in the brain.