M1, but also from PMv, PMd, SMA, CMAd, CMAr, and CMAc. Corticospinal axons from the NPMAs terminate in the intermediate zone and the ventral horn of the spinal cord, in a distribution that is quantitatively less but qualitatively similar to axonal projections from M1 (for review see Dum and Strick, 2002).
Further evidence supporting a parallel, distributed model of motor control and not a sequential, hierarchical model comes from reaction time and neuronal firing time data. Given that conduction times from M1 to spinal motoneurons and then to hand muscles require only 20-30 ms, human reaction times, approximately 200 ms for visual cues, and 100 ms for somatosensory cues (Cordo and Flanders, 1989), allow sufficient time for information sharing within and between multiple cortical and subcortical areas prior to a motor response.
Likewise in non-human primates, task-related Ml neurons and NPMA neurons increase their firing approximately 50-200 ms prior to the onset of movement (Tanji et al., 1988; Shima et al., 1991), allowing for the exchange of information between multiple cortical and subcortical areas related to the planning and execution of the intended movement. Although the average onset of increased task-related neuronal firing in the NPMAs slightly precedes the average onset of firing in M1, the distributions of onset times are broadly overlapping, making it unlikely that a motor command would progress in a sequential, hierarchical manner.
Understanding the roles played by these NPMAs in voluntary motor control is a focus of ongoing research. Though this work cannot be reviewed thoroughly here, current research suggests that different NPMAs may have overlapping but partially separate functions. A good example is the difference between the PMv and the SMA during visually-cued versus self-paced movements. Passingham and his coworkers have shown that bilateral lesions of the premotor cortex (which include PMv) produce profound deficits in a monkey's ability to choose which movement to make based on visual cues, but these same lesions do not impair the ability to generate self-paced, internally-cued movements (Passingham, 1987). Conversely, bilateral SMA lesions produce a deficit in internally-cued movements but not in visually-cued choices. Though one might conclude that visually-cued movements are controlled only by the PMv, and internally-cued movements only by the SMA, physiologic recordings show some overlap of function. Mushiake et al. (1991) trained monkeys to press three buttons in sequence. At first, a new sequence was cued to the monkey by illuminating the buttons one at a time on each trial. Once the monkey had practiced the sequence enough, he could continue performing it for several trials without having the buttons illuminated, guided by remembered internal cues. Though the PMv and SMA both contained neurons that were active during both visually-cued and internally-cued movements, PMv neurons on average were more active during visually-cued movements, whereas SMA neurons on average were more active during internally-cued movements. So although the PMv may play a greater role in visually-cued movements, and the SMA in internally-cued movements, these are quantitative differences and there is some overlap of function between PMv and SMA.
Although the majority of studies investigating NPMAs have been done with animals, there is now a considerable amount of evidence indicating that humans have multiple NPMAs similar to those iden-tifiedin monkeys (Fig. 2.2(b)). Cortical surface stimulation via implanted subdural electrode grids up to 30 mm anterior to the Rolandic fissure (i.e., 20 mm anterior to the precentral gyrus) produces motor effects in humans (Uematsu et al., 1992). Motor related cortical potentials derived from electroen-cephalography suggest that both the SMA and M1 are normally involved in self-paced movements (Tarkka and Hallett, 1991; Ikeda et al., 1992). Studies of regional cerebral blood flow showing activation of NPMAs - including SMA, PMv, PMd, CMAd, CMAr, CMAc - during various motor tasks suggests that the human brain contains NPMAs that are homologous to the NPMAs identified in non-human primates (see Picard and Strick, 2001 for review). And finally, resections of NPMAs such as the SMA, where the surgical resection leaves M1 intact, produces an initial hemiparesis that almost completely resolves (Laplane et al., 1977; Krainik et al., 2001). Though the correspondence between particular activated regions in humans and those more precisely defined in monkeys has yet to be fully understood, all the above evidence indicates that human voluntary movement control normally requires NPMAs along with M1 and that voluntary movement is controlled via a parallel, distributed process.
Implications of NPMA organization for recovery after brain injury
Compared with a hierarchical control process, a parallel, distributed control process would allow for better recovery after damage to one part of the system. The NPMAs are therefore well-suited to provide compensatory control of voluntary movement after damage to M1. With their individual loosely somatotopic representations and corticospinal projections, neurons in NPMAs could provide compensatory control of spinal motoneurons after damage to the M1. Although each NMPA has its own unique inputs and neural activation patterns in relation to various aspect of movement control, subsets of neurons in each area have activation patterns that are similar to Ml neurons (Tanji and Kurata, 1982; Shima et al., 1991; Cadoret and Smith, 1997; Boudreau et al., 2001). Furthermore, neurons in NPMAs are more frequently related to bilateral movements than neurons in M1 (Tanji et al., 1988), and thus may be able to exert control over spinal motor neurons via both corticospinal pathways. Electrical stimulation in patients with structural cortical lesions, and magnetic stimulation in patients with traumatic quadriplegia, both have been found to evoke muscle contractions from a much wider area of cortex than just M1, suggesting that NPMAs have developed more output to motoneurons (Levy et al., 1990; Uematsu et al., 1992). Imaging studies indicate that PM and SMA are more active during finger movements in hemiparetic subjects compared to control subjects (Weiller et al., 1992, 1993; Seitz et al., 1998), suggesting that these NPMAs are providing compensatory control of voluntary movements after damage to M1. Interestingly, many imaging studies after stroke show increased activity not only in the NPMAs described above, but also in parietal regions and in ipsilateral M1. Thus NPMAs, which play a role in generating normal voluntary movements, take on an increasingly important role after brain injury.
We speculate that as the particular situations that maximally engage each NPMA are better understood, rehabilitation professionals will be better able to promote functional recovery. For example, if a particular patient has a lesion that affects M1 and the human equivalent of PMv but spares the SMA, movements made in response to visual cues might be most severely impaired, whereas internally generated movements that are mediated in part by the SMA might be relatively spared. The most effective rehabilitative approach might then primarily employ internally generated movements, or teach the patient compensatory strategies to substitute internal cues in situations where visual cues normally suffice. In this way, each patient's neurorehabilitation might be better tailored to their particular brain injury.
Was this article helpful?