Modern coordinates for the motor homunculus

Anyone who has compared Penfield’s ultimate motor homunculus (Penfield & Rasmussen, 1950) with some of the original data on which this summary cartoon was based (Penfield & Boldrey, 1937) will be left wondering: Was the ribbon of discrete zones of separate body-part representations an accurate insight envisioned through noisy data? Or was it an easy-to-grasp and easy-to-remember oversimplification of the actual underlying somatotopic organization? After all, Penfield’s database combined many different patients, the stimulators were crude by modern standards, and stimulated locations were marked by hand on a standard outline of the rather variable cortical gyri. Bringing modern techniques to bear, Roux and colleagues have compiled 14 years of meticulous intraoperative observations to re-examine the human primary motor cortex (M1) in the current issue of the Journal of Physiology (Roux et al. 2020).

Like Penfield’s studies, Roux and colleagues have necessarily combined observations from multiple patients, both because time for investigations during human neurosurgery is limited and because surface electrical stimulation of M1 can evoke seizures. Applying technical advances not available to Penfield, however, this modern study used constant current pulses delivered between bipolar electrodes, the stimulated points were tracked with a modern neuro-navigation system, and the tracked locations were morphed to the standard coordinates of the Montreal Neurological Institute. Penfield and colleagues evoked movements from many points anterior or posterior to the precentral gyrus per se, whereas Roux and colleagues evoked movements almost exclusively from the precentral gyrus. Penfield and colleagues evoked somatosensory percepts at a number of loci on the precentral gyrus, whereas Roux and colleagues did not. The modern data thus are focused more sharply.

Do these 21st century data now demonstrate a sequence of discrete, separate zones representing the shoulder, then elbow, then wrist, then hand? No, they do not. Instead, the data show what the authors term ‘relative somatotopy;’ namely, a gradual progression in which shoulder movements are represented medially and hand movements laterally along the central sulcus, each overlapping considerably with elbow and wrist movement representations in between (their Figs 2 and 4). Furthermore, although in 3 of 51 patients individuated movements of the digits tended to be evoked more laterally than en masse movements, the classic medial-to-lateral progression from little finger to thumb is absent in the modern data (their Figs 1 and 3). Gradual progression with overlap is even more evident in the representation of the face, tongue and larynx. In both the representation of the upper extremity and in that of the head, the modern data show gradual somatotopic progression with overlap of different body parts, not sequences of distinct, well-demarcated zones.

Another interesting point emerges from these modern data. Though no clear somatotopic sequence of the five fingers appeared, flexion of the fingers was evoked far more often than extension (flexion 63 times, extension 8). This observation contradicts clinical experience, in which lesions of M1 or the corticospinal tract weaken extension of the fingers more than flexion. It may be that lesions impair both flexion and extension in proportion to their representation, but the clinical weakness appears more profound in the extrinsic finger extensors because they have a much smaller muscle mass than the flexors. Alternatively, finger extensor muscles might be represented more heavily in the portion of human M1 that lies buried in the anterior bank of the central sulcus, relatively inaccessible to electrical stimulation at the surface. A third factor might be that representation of finger flexions is more ‘magnified’ than that of extensions because almost all functions of the hand in grasping objects and using tools – from hammers to keyboards – require strong and/or precise finger flexion, with extension being used simply to release or withdraw the fingers.

The scientific progress achieved by Roux and colleagues in part reflects technological advances. The next step in electro-physiological investigation of human M1 may come from applying intracortical microstimulation (ICMS). Rather than delivering milliampere pulses of current through electrodes touching the cerebral surface, ICMS delivers microampere pulses through microelectrodes that penetrate into the cerebral grey matter. ICMS thus achieves more focal stimulation but requires a more invasive procedure. One might expect that more focal stimulation would reveal more clear-cut somatotopic organization. But in animal studies, where mapping in individual subjects can proceed systematically over days to weeks, the same gradual progression with extensive overlap is, if anything, even more evident (Park et al. 2001).

Currently, microelectrode arrays are being implanted in certain human patients for development of brain–computer interfaces (BCIs). These implanted arrays offer the opportunity to stimulate through dozens of microelectrodes arranged in a known grid, making repeated observations over days to months. In the human primary somatosensory cortex the percepts evoked by ICMS are under study (Flesher et al. 2016). But as yet, although microelectrode arrays have been implanted in human M1, ICMS has not been applied there. Why not?

In non-human primates ICMS of M1 does not seem to evoke seizures (personal observation), so this may be less of a concern than with surface stimulation. For ethical reasons, however, patients receiving implanted microelectrode arrays to date have had a neurological disease – spinal cord injury (SCI), amyotrophic lateral sclerosis (ALS), locked-in syndrome, etc. – damaging the pathways between M1 and muscles, and impairing or precluding observation of evoked muscle contractions or movements. But as implanting microelectrode arrays in humans becomes more widely accepted, ICMS in M1 of selected patients with incomplete lesions will be informative. And beyond extending our scientific understanding of M1, ICMS eventually might provide new therapeutic modalities in which BCIs stimulate M1 to facilitate movement in certain patients with incomplete SCI, ongoing ALS, or hemi-paresis. Achieving such progress requires dealing realistically with the kind of M1 organization demonstrated by Roux and colleagues.