OBJECTIVE: To investigate the clinical and anatomical correlates of supernumerary phantom limb.
BACKGROUND: Supernumerary phantom limb have been associated with right cerebral lesion, anosognosia and neglect.
DESIGN/METHODS: Using a cognitive model of motor control and an fMRI motor paradigm, we investigated a case of persisting kinesthesic phantom illusion of a supernumerary limb after sub-cortical infarction.
RESULTS: M., a 70-year-old left handed woman, developed right hemiplegia with sensory I03S, hemianopia, and neglect with denial of arm ownership after anterior choroidal artery infarction (no cortical diaschisis on perfusion CT). 7 months later, hemianopia, severe hemiparesis and sensory loss persisted. M. described a supernumerary right arm which had appeared a few weeks after stroke. It was a"normal" limb, attached to her paretic arm at elbow level. The illusion was triggered by any intentional attempt to move the paretic limb, or by voluntary movement of the healthy arm with bimanual tasks. Passive movements did not elicit the illusion. Looking at the paretic arm or applying a tactile stimulation did not cancel the illusion. Comparison between fMRI signals performed during virtual movement of the phantom hand vs. imaginary movement of the paretic hand showed increased activation in thalamus and caudate nucleus in the first condition.
CONCLUSIONS: Our findings suggest that supernumerary phantom limb may develop after a deep brain lesion sparing cortex. The striking point was the kinesthetic character of the phenomenon and its link with the sense of will. According to motor control models, action initiation awareness depends on a representation of the predicted consequences of action execution. This 'premovement brain signal of action'(efference copy), relies on the integrity of controllers which provide the motor commands to achieve a desired movement and the predictors which estimate the sensory consequences of the action. In M., despite paralysis, motor commands could still be issued, leading to prediction of movement. Concordance between intended and predicted movement could not be contradicted by sensory feedback, leading to the illusion of movement and supernumerary limb. In principle, awareness of the discrepancy between predicted and actual movement should have led to predictors updating and loss of the illusion. However, this was probably prevented by failure to analyze errors in sensory entries, due to interruption of thalamic afferences which is supported by the fact that the illusion was not cancelleded by sight and touch. The fMRI activation of the basal ganglia-thalamus-cortex loop during virtual movements of the phantom may suggest that the thalamocortical network operated in abnormal closed-loop way due to interruption of sensory inputs, while integrity of predictors allowed to generate the internal phantom representation of limb position.