Coarse-grained models of cortical circuits

Biologically realistic models of cortical circuits are useful for elucidating mechanisms and generating testable predictions, but can be challenging to build, tune, and analyze due to the large numbers of neurons and their complex interactions. Reduced, or coarse-grained, models are more tractable, but there is a nontrivial trade-off between tractability and biological fidelity / interpretability. In this talk, I will describe a coarse-graining strategy inspired by ideas from nonequilibrium statistical mechanics. The aim is to balance biological realism and computational efficiency. I will illustrate how this strategy applies to the primate primary visual cortex. This is joint work with Zhou-Cheng Xiao and Lai-Sang Young.