Network basis of action selection

J. David Gladstone Institutes, San Francisco CA

Investigators

Linked publications & trials

Abstract

? DESCRIPTION (provided by applicant): The anatomical substrates and cellular mechanisms underlying reward-dependent learning have been studied for decades, but the specific circuit and network interactions between the cortex, striatum, and midbrain that mediate action selection have not been systematically investigated. Here, we bring together three different investigators with specialized expertise in each of these three brain regions. As a team, the investigators are well positioned to perform cutting-edge cell-type-, region-, and projection-specific imaging and ontogenetic manipulations of neuronal ensembles during complex behavioral tasks designed to uncover the network-level representation of decision-related variables for action selection. The computational frameworks for analyzing these data are provided by reinforcement learning models. In Aim 1, the focus is on the striatum, which arguably lies at the center of reward-dependent learning. It is the point of intersection for sensorimotor and contextual information from cortex, and reward- and motivation-related information from the midbrain. Using microendoscopy with calcium imaging from genetically-specified neuronal subtypes in different striatal subregions during behavior, we will identify the key decision-related variables represented in the striatum during action selection tasks. In Aim 2, we will image from specific populations of dopamine neurons in the midbrain that project to different striatal subregions, in order to decipher their role in reward-related signaling. In Aim , we will image large ensembles of neurons in the motor cortex (M1 and M2) using two-photon microscopy, with a focus on neurons projecting to distinct cell types and subregions of striatum. All three aims will use the same behavioral tasks, and the same analysis techniques, in order to facilitate the integration of data from all three brain regions into a single coherent model for vertebrate action selection.