Engineering Approaches to Understand Functional Connectivity in Neocortex

The mammalian neocortex is a crowning achievement of evolution. It is astronomically complex, with around 100 billion computational elements, each of which is staggeringly intricate by itself, and on the order of 10¹⁶ synaptic connections. In this talk, I plan to examine three questions related to neocortex.

First, what are the consequences of component miniaturization for neural computation? Second, how can we model neural computation on such a scale in a way that makes tractable predictions? Third, what does distributed neural computation “look like?” The bulk of the talk will focus on testing strong predictions from the relatively simple stabilized supralinear network (SSN) model of how neocortical networks behave in resting wakefulness, and how that behavior changes when the network is activated by sensory input or intentional movement.

Our data are collected from mouse somatosensory cortex, mainly under whole-cell patch clamp, but also using genetically encoded calcium indicators. Our results are mainly compatible with the SSN model.