“…They have shown that such conditioning produces multisite changes at the level of the spinal cord that actually drive the observed differences in the H-reflex response, including a shift in motorneuron firing threshold (Carp and Wolpaw, 1994) and a change in the number of GABAergic terminals (Wang et al, 2006) . Importantly, successful operant conditioning of the spinal cord circuit itself requires a functional corticospinal tract and sensorimotor cortex as well as the cerebellum and inferior olive but no other major ascending or descending spinal pathways Wolpaw, 2002, 2005;Chen et al, , 2006aChen et al, , 2006bChen et al, , 2016Wolpaw and Chen, 2006) , indicating that cerebellar contributions via the sensorimotor cortex (as opposed to the rubrospinal tract) are critical for implementing the learning (Chen and Wolpaw, 2005;Wolpaw and Chen, 2006) . Given the differences between H-reflex operant conditioning, especially with respect to its development over weeks and months, an extremely long time-scale even relative to the slow learning in our paradigm, it is unclear whether the same mechanisms are in play for the type or learning we report here.…”