Most paralysis technology works in one direction: read the brain, move a limb. A Nature Medicine paper describes a system that goes both ways in a person with complete tetraplegia. It links an intracortical brain-computer interface with patterned stimulation of the spinal cord and cortex. The system uses recurrent artificial neural networks and reinforcement learning for fine grasp control, plus patterned spinal cord stimulation and activity-informed intracortical microstimulation.
A Nature Neuroscience research highlight explains the design. Decoded movement intentions from motor cortex drive muscle stimulation, force sensors on the hand feed touch back through microstimulation of the somatosensory cortex, and a therapeutic arm delivers spinal stimulation paired with sensory input to promote plasticity. The participant could self-feed and grasp fragile objects. The striking part is that gains in elbow flexion strength and wrist sensation persisted even when stimulation was no longer being delivered. That suggests the system may be retraining circuits, not just substituting for them.