From "Cognitive-Type" Neural Circuits to Computational Psychiatry

The second lecture will focus on a unifying recurrent neural circuit model of working memory and decision-making, which led to the concept of "cognitive-type" neural circuits, in contrast to those dedicated to early sensory processing or movement generation. Several model predictions and their experimental tests will be discussed, and current debates in the field including the controversy over the dynamical nature of working memory representations will be addressed. Importantly, by incorporating reward-dependent synaptic plasticity, the model is shown to have explanatory power for a variety of learning value-based adaptive choice tasks, such as stochastic foraging, competitive games and probabilistic inference. I will also discuss contributions of subcortical structures to cognition in a spiking network model of a cortico-basal ganglia-thalamocortical loop for decision-making and inhibitory control of habitual responses. Because this brain system is critically implicated in mental illness, neural circuit modeling that bridges levels (from receptors and cells to circuits and behavior) serves as a potentially powerful platform for investigations of how this system goes awry in psychiatric disorders like Schizophrenia or addiction. I will provide examples that illustrate this nascent field called "Computational Psychiatry".