Back

Subcortical control of reaching in humans

Contemori, S.; Loeb, G. E.; Corneil, B. D.; Wallis, G.; Carroll, T. J.

2025-11-26 neuroscience
10.1101/2025.11.26.690644 bioRxiv
Show abstract

Accurate visually guided reaching requires transformation of target-related photoreceptor responses into precisely coordinated activation of trunk and arm muscles. The cerebral cortex is widely believed to compute the requisite kinematic and musculoskeletal dynamics strategies in humans 1-3, even though vertebrates lacking a cerebral cortex achieve sophisticated visuomotor control 4-6, and brainstem circuits executing coordinated eye and head gaze shifts perform analogous sensorimotor computations in non-human primates 7. Here we used a visuomotor reaching task that yields extremely rapid, "express", target-directed muscle activations 8-10 to test whether a putative subcortical sensorimotor network can compute musculoskeletal dynamics to initiate reaching in humans. We found coordinated express visuomotor responses (EVRs) in task-relevant shoulder, elbow, and bi-articular muscles that reflected both starting posture and target direction in similar patterns to longer latency, presumably cortically mediated, muscle responses. When the task goal was to reach away from the stimulus (i.e. an "anti-reach"; 11) the EVR involved coordinated muscle activation to initiate the hand toward the stimulus location, opposite to the subsequent goal-directed response. The results suggest a unified theory of visuomotor control for reaching and gaze shifts, in which subcortical systems compute musculoskeletal dynamics based on sensory target information and cortically derived context. The results imply that the transformation from motor goals in extrapersonal space into musculoskeletal dynamics can be performed by neural circuitry in humans that does not involve the sensorimotor cortex.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.