Brain Representations of Dexterous Hand Control: Investigating the Functional Organization of Individuated Finger Movements and Somatosensory Integration
Abstract
Using our hands to manipulate objects in our daily life requires both dexterous movements and the integration of somatosensory information across fingers. Although the primary motor (M1) and somatosensory cortices (S1) are critical for these two complementary roles, it is unclear how neural populations in these regions functionally represent these processes. This thesis examined the functional organization of brain representations (the representational geometry) in M1 and S1 for dexterous hand control and somatosensory processing. To that end, representational geometries were estimated from fine-grained brain activity patterns measured with functional MRI (fMRI). Since fMRI measures a blood-based proxy of neural activity, any non-linearities in the coupling between neural activity and the fMRI signal could distort the representational geometries. Chapter 2 therefore evaluated the stability of representational geometries. Human participants made individuated finger presses at varying pressing speeds, such that overall activity was modulated across a broad range. Representational geometries were relatively stable across pressing speeds in M1 and S1, validating the use of this analysis framework with fMRI data. Chapter 3 then explored how M1 is organized for dexterous hand control. In agreement with previous research, representations of each finger were quite distinct. However, representations of the same finger moving in different directions were very similar. Insight into this observation was gained by comparing the fMRI results to neural spiking data recorded in monkeys trained to perform an identical task. By leveraging the complementary perspectives offered by fMRI and spiking, a new organization of M1 for finger control was proposed. Chapter 4 then examined how somatosensory inputs from multiple fingers are integrated in S1. The full nature of this integration is unknown. Here, human participants experienced simulation of all possible single- and multi-finger combinations. Representational model analyses revealed that unique non-linear interactions between finger sensory inputs occur throughout S1, with stronger (and more spatially distant) interactions occurring in posterior S1. Altogether, these results provide new insight into how M1 and S1 are functionally organized to serve the motoric and sensory processes of the hand, and more broadly demonstrate how fMRI can be used to make inferences about the underlying functional organization of brain representations.