Investigating the structure of sensorimotor cortical computations

Abstract

Cortical computations underlie how humans interact with their environment through sensory information processing and movement execution, but our understanding of how they are implemented is limited. Computations are mediated by the structured transfer and transformation of information between specific neural populations with precise timing, which is mechanistically facilitated by different cell-types. To understand the structure underlying cortical computations we must understand the spatiotemporal structure, which can be used to make precise hypotheses about the mechanistic role of specific cell-types. This thesis aims to investigate the structure of cortical computations through a spatial, temporal, and transcriptomic lens. I investigated the spatiotemporal structure underlying motor control because the spatiotemporal structure that supports motor computations is not well established. I found that reach information was heterogeneously distributed across frontal motor cortex, yet the neural populations with the most task information also had neural activity patterns that evolved most similarly in time. Next, I assessed the spatiotemporal structure underlying how movement information transitions between two motor computations: planning and execution. I found spatially distinct, computation-specific neural populations that coexisted with a separate, spatially distributed population involved in both planning and execution. I also investigated the transcriptomic structure of cortical computations in visual cortex, where the spatiotemporal structure is well established. I tested and validated new optogenetic tools to specifically manipulate parvalbumin-expressing (PV+) neurons. Using these tools, I found that PV+ neurons modulated visual sensitivity. Collectively the results from this thesis shed light on cortical computations in areas without a well-understood spatiotemporal structure and begin revealing the role of transcriptomically defined cell types in areas where the spatiotemporal structure is established. Further work will be necessary to integrate novel optogenetic tools that target specific cell types with investigations into cortical computations to reveal how they are mechanistically implemented.