CRCNS: Multiregional computations for value-based decision-making

New York University, New York NY

Investigators

Abstract

A major goal of contemporary neuroscience is to understand how multiregional interactions instantiate computations for behavior and cognition. However, in complex cognitive tasks, there are many quantities and abstract relationships that must be computed, and it can be difficult to know which computations are specifically supported by the brain regions under study. This limits understanding of inter-area communication in support of behavior. We recently demonstrated a causal relationship between neural dynamics in the orbitofrontal cortex (OFC) of rats performing a temporal wagering task, and a precise behavioral computation - updating subjective beliefs about hidden states of the environment. We will leverage this finding to characterize how OFC interacts unidirectionally with the dorsal striatum (dStr) and bidirectionally with the basolateral amygdala (BLA) to support belief updating and context dependent decision making. We will perform simultaneous electrophysiological recordings from OFC and dStr or BLA and use novel statistical models of their joint population activity to characterize the distributed nature of their joint function. Our novel statistical modeling approach provides a description of the process by which these brain regions interact and communicate. Statistical data analysis will be complimented by multi-area recurrent neural network models (RNNs) that exhibit rat-like behavior in this task, and that will provide an opportunity for hypothesis generation and hypothesis testing in parallel with experiments. We will also causally manipulate OFC neurons that project to dStr and evaluate effects on behavior, and perform optogenetically-tagged recordings to characterize the responses of single neurons projecting to dStr. The results will provide a computational framework for determining how interactions between brain regions support computations for cognition. It will relate these dynamics to a precisely defined, core cognitive computation that we have shown causally requires the OFC: updating subjective beliefs about abstract, latent states of the environment based on outcomes. RELEVANCE (See instructions): Substance use disorders are characterized by aberrant value-based decision-making, and are thought to reflect dysfunction of cortico-basal ganglia circuitry. A better understanding of how these circuits support value-based decision-making could identify novel therapeutic targets for these disorders.