P5: Mechanistic Multi-Region Brain Models

点击翻译按钮获取中文摘要

Project Summary/Abstract: Project 5, Mechanistic Multi-Region Brain Models Elucidating the specific computational roles of different brain areas and how they work together to solve complex evidence-accumulation and decision-making problems is a key goal of our U19 program and of the BRAIN initiative. This project will take advantage of the unique multi-region experimental datasets from Projects 1-4 to construct a set of mechanistic models of how multiple brain regions work together to perform our accumulation-of-evidence based decision-making task. Aim 1 focuses on the role of the basal ganglia, often associated with the gating or selection of actions, within our cognitive decision-making task. Building on the experimental data in Project 4, Multi-Region Interactions, we will construct models hypothesizing how the two core pathways traversing the basal ganglia, the direct and indirect pathways, may serve to gate evidence or position information to accumulator circuits in the neocortex. In turn, we will build models of how accumulated evidence from the neocortex is used to drive the transition from evidence-seeking to choice-selective actions in the basal ganglia. Aim 2 focuses on the role of the entorhinal cortex and hippocampus, and their interactions with sensory and frontal regions of neocortex, in generating the joint cognitive map of animal position and accumulated evidence observed in our neural recordings of Project 2, Geometry of Neural Dynamics and Representations. The model will provide a powerful theory, grounded in biologically plausible mechanisms, for how cognitive maps are formed and will form predictions for neuronal manifold structures and effects of causal manipulations in our entorhinal, hippocampal, and neocortical recording experiments. Aim 3 focuses on the role of the cerebellum, and its interactions with neocortical accumulation-of-evidence circuits, in decision-making. Drawing inspiration from classic motor cerebellar experiments suggesting how the cerebellum may mediate the production of smooth, well-coordinated motor actions, we propose a new theory of the cognitive cerebellum as stabilizing noisy neural trajectories to produce smooth cognitive actions. This Aim will be informed by, and in turn form predictions for, the Cerebellar aim of Project 4, Multi-Region Interactions. Aim 4 will combine the regional models of Aims 1 to 3 with a further model of multiple interacting neocortical regions to produce a single, large scale model of accumulation-of-evidenced based decision-making. The model will be informed by data from all projects and will enable us to dissect the roles of individual regions and their interactions in the performance of the many variants of our decision-making task. Taken together, we expect that these modeling efforts, deeply integrated with experiments in the other four Projects, will substantially advance three priority areas of the BRAIN Initiative: the brain in action, demonstrating causality, and identifying fundamental principles.