Disentangling motor execution from motor imagery with the phantom limb.

Amputees can move their phantom limb at will. These 'movements without movements' have generally been considered as motor imagery rather than motor execution, but amputees can in fact perform both executed and imagined movements with their phantom and they report distinct perceptions during each task. Behavioural evidence for this dual ability comes from the fact that executed movements are associated with stump muscle contractions whereas imagined movements are not, and that phantom executed movements are slower than intact hand executed movements whereas the speed of imagined movements is identical for both hands. Since neither execution nor imagination produces any visible movement, we hypothesized that the perceptual difference between these two motor tasks relies on the activation of distinct cerebral networks. Using functional magnetic resonance imaging and changes in functional connectivity (dynamic causal modelling), we examined the activity associated with imagined and executed movements of the intact and phantom hands of 14 upper-limb amputees. Distinct but partially overlapping cerebral networks were active during both executed and imagined phantom limb movements (both performed at the same speed). A region of interest analysis revealed a 'switch' between execution and imagination; during execution there was more activity in the primary somatosensory cortex, the primary motor cortex and the anterior lobe of the cerebellum, while during imagination there was more activity in the parietal and occipital lobes, and the posterior lobe of the cerebellum. In overlapping areas, task-related differences were detected in the location of activation peaks. The dynamic causal modelling analysis further confirmed the presence of a clear neurophysiological distinction between imagination and execution, as motor imagery and motor execution had opposite effects on the supplementary motor area-primary motor cortex network. This is the first imaging evidence that the neurophysiological network activated during phantom limb movements is similar to that of executed movements of intact limbs and differs from the phantom limb imagination network. The dual ability of amputees to execute and imagine movements of their phantom limb and the fact that these two tasks activate distinct cortical networks are important factors to consider when designing rehabilitation programmes for the treatment of phantom limb pain.