How the brain orchestrates the dozens of muscles required for even simple movement has long been one of neuroscience's most debated questions. New electrophysiological evidence from intraoperative cortical stimulation now provides the most direct human proof yet that the primary motor cortex itself encodes coordinated muscle groupings — so-called synergies — rather than commanding individual muscles in isolation. This has meaningful implications for rehabilitation science, neuroprosthetics, and our understanding of motor recovery after stroke or injury.
The study leveraged a rare clinical opportunity: patients undergoing awake craniotomy for glioma resection, during which direct electrical stimulation of the primary motor cortex was applied to map eloquent tissue. By recording electromyographic responses across multiple muscle groups simultaneously, the researchers identified discrete neuromotor modules — reproducible patterns of co-activated muscles — that aligned with theorized synergy structures. Crucially, these modules were elicited directly from cortical sites, bypassing peripheral or spinal contributions, establishing a cortical rather than subcortical or spinal origin for at least a subset of human muscle synergies.
This finding enters a long-standing debate. Influential work in animal models — particularly in frogs and non-human primates — had suggested synergies might emerge primarily from spinal interneuron circuits, with the cortex simply activating pre-wired spinal modules. The present data complicate that picture for humans, indicating the cortex may itself house or at minimum strongly co-determine these coordinative units. For rehabilitation and brain-machine interface design, this is consequential: it suggests motor cortex decoding strategies that target synergy-level signals — rather than individual muscle commands — may more faithfully represent the brain's native output vocabulary. The study's key limitations are inherent to the design: small patient cohort, pathological brain tissue, and stimulation parameters that do not replicate natural neural firing. Replication in non-lesioned populations and integration with high-density cortical recording techniques will be essential next steps.