The Computational CEO of the Brain

In the first lecture, I will review developments in computational neuroscience as it pertains to cognitive processes such as working memory (the brain’s ability to internally maintain and manipulate information in the absence of external stimulation) and decision-making (the deliberate process of making a choice among several options under uncertainty). I will especially discuss biologically-realistic neural circuit modeling that is based on two pillars of neurobiology: anatomy and neurophysiology. Computational work in close interplay with experiments of behaving animals has shed insight onto the mystery of the prefrontal cortex, often called “the CEO of the brain”. Mathematically, our work revealed a novel type of attractor networks that does not merely behave as switches between steady states but is characterized by a duality: slow transients underlying graded accumulation of information over time in the form of quasi-linear ramping neural activity, and winner-take-all competition giving rise to a categorical choice by virtue of convergence to an attractor state.