Neurorehabilitation BCI (stroke)

A neurorehabilitation BCI (most developed for chronic upper-limb stroke) uses the interface therapeutically rather than assistively: the patient attempts or imagines a movement, the decoder detects the corresponding motor-cortical activity — typically sensorimotor-rhythm event-related desynchronization over the lesioned hemisphere, or spikes/ECoG at higher resolution — and the system delivers contingent, tightly-timed feedback: a robotic orthosis that moves the hand, functional electrical stimulation of the muscle, or visual/proprioceptive feedback. The therapeutic hypothesis is Hebbian: pairing the volitional descending command with its matching afferent consequence, within a tight temporal window and over many repetitions, strengthens surviving corticospinal pathways and biases reorganization toward recovery. Unlike an assistive BCI, the goal is to become unnecessary — to leave durable gains after the device is removed.

Honesty about the evidence: several randomized controlled trials and meta-analyses (motor-imagery BCI paired with FES or a robotic orthosis) report impairment-scale gains that are statistically significant and exceed sham, but effect sizes are moderate, heterogeneous across studies, and the durable, real-world functional benefit — versus dose-matched conventional therapy — is still being established. It is a mechanism-motivated adjunct, not yet standard of care, and works best in patients who retain some corticospinal integrity to reinforce.