Dissecting roles of inputs and recurrent connectivity in motor cortex

About This Grant

The overarching objective of the proposed research is to dissect how primary motor cortex (M1) flexibly controls different types of movements in the macaque monkey. In primates, M1 is a necessary structure for the generation of voluntary limb movements, like reaching towards your phone and then grasping it. M1 appears to operate under distinct computational modes in reaching and grasping. During reaching, neural population dynamics, that is the time-varying pattern of firing rates in populations of neurons in M1, are low-dimensional and with a specific structure (“rotational”). In contrast, neural population dynamics during grasping are high-dimensional and not rotational. The goal of this specific proposal is to understand how M1 can exhibit these different computational modes. My overall premise is that these diverse operating modes of M1 emerge due to relative contributions of two key components: autonomous recurrent dynamics within M1, and external inputs from other brain areas. Here, I will test my central hypothesis that M1 activity during reaching is controlled primarily by strong recurrent dynamics, whereas grasping is dominated by inputs in two Aims. In Aim 1, I will train monkeys to perform two different tasks: a delayed reaching task, and delayed grasping of various objects (using a turntable). I will use high-density electrophysiology (Neuropixels) to measure the neural population dynamics in M1 and two areas that provide input to M1 (dorsal premotor cortex (PMd) and primary somatosensory cortex (S1)) during these tasks. I will then use residual dynamics analysis techniques to identify the dynamical landscape and inputs to M1 underlying reaching and grasping. The residual dynamics approach models M1 neural population activity as a dynamical system and uses trial-to-trial variability to distinguish between input-driven (e.g., rotations with minimal inputs) versus recurrence-driven (e.g., moving point attractors) systems. I will test the hypothesis that M1 dynamics during reaching are better explained by models with recurrent dynamics within M1, whereas those during grasping are better explained by models that leverage inputs from other areas. In Aim 2, I will use causal approaches to further test the hypothesis. I will express red-shifted opsins in inhibitory neuron populations of PMd and S1. I will then stimulate these inhibitory neurons one at a time to silence the area, while recording from M1 as the animals perform the reaching or grasping task. I will then assess the impact of this optogenetic silencing on (PMd/S1) on both behavior and M1 neural dynamics. My hypothesis is that silencing input areas will have stronger impact on both behavior and M1 dynamics during grasping than reaching. Collectively, the two aims will dissect the roles of inputs versus recurrence in M1 dynamics across multiple tasks and illuminate how M1 operates under dramatically different modes to flexibly control arm and hand movement behavior. Impact: The proposed work understands how M1 in concert with other clinically relevant brain areas flexibly controls two fundamental behaviors: reaching and grasping. Such an understanding will enable the development of better brain-computer interfaces and circuit-level interventions for stroke, cortical injury, and neurological disorders.