Abstract
Cervical spinal cord stimulation (SCS) improves arm function in people with chronic post-stroke hemiparesis, yet its mechanism remains unclear. The leading hypothesis is that SCS enhances motor drive to agonist muscles. To test this, we developed a proportional-derivative (PD) control model of agonist-antagonist interactions and estimated feedback control gains in 12 healthy and 5 stroke participants. In healthy individuals, gains followed a stereotypical structure that we showed was optimal for generating smooth, stable movements. In stroke participants, this structure was disrupted but shifted toward neurotypical (optimal) values with SCS. Contrary to the prevailing hypothesis, the most consistent effect of SCS was a reduction in stabilization (derivative) gains rather than an increase in motor drive (proportional) gains, suggesting engagement of inhibitory control mechanisms of antagonist (and agonist) muscles. SCS also tuned feedback gains in a frequency-dependent manner, indicating that controller gains could guide stimulation parameter selection. This model-based framework offers an objective means to quantify neuromodulation effects on motor control and identify mechanisms of functional recovery after stroke.