Plastic Changes in Human Motor Cortical Output Induced by Random but not Closed-Loop Peripheral Stimulation: the Curse of Causality

Previous work showed that repetitive peripheral nerve stimulation can induce plastic changes in motor cortical output. Triggering electrical stimulation of central structures from natural activity can also generate plasticity. In this study, we tested whether triggering peripheral nerve stimulation from muscle activity would likewise induce changes in motor output. We developed a wearable electronic device capable of recording electromyogram (EMG) and delivering electrical stimulation under closed-loop control. This allowed paired stimuli to be delivered over longer periods than standard laboratory-based protocols. We tested this device in healthy human volunteers. Motor cortical output in relaxed thenar muscles was first assessed via the recruitment curve of responses to contralateral transcranial magnetic stimulation. The wearable device was then configured to record thenar EMG and stimulate the median nerve at the wrist (intensity around motor threshold, rate ~0.66 Hz). Subjects carried out normal daily activities for 4-7 h, before returning to the laboratory for repeated recruitment curve assessment. Four stimulation protocols were tested (9-14 subjects each): No Stim, no stimuli delivered; Activity, stimuli triggered by EMG activity above threshold; Saved, stimuli timed according to a previous Activity session in the same subject; Rest, stimuli given when EMG was silent. As expected, No Stim did not modify the recruitment curve. Activity and Rest conditions produced no significant effects across subjects, although there were changes in some individuals. Saved produced a significant and substantial increase, with average responses 2.14 times larger at 30% stimulator intensity above threshold. We argue that unavoidable delays in the closed loop feedback, due mainly to central and peripheral conduction times, mean that stimuli in the Activity paradigm arrived too late after cortical activation to generate consistent plastic changes. By contrast, stimuli delivered essentially at random during the Saved paradigm may have caused a generalized increase in cortical excitability akin to stochastic resonance, leading to plastic changes in corticospinal output. Our study demonstrates that non-invasive closed loop stimulation may be critically limited by conduction delays and the unavoidable constraint of causality.