Swing those arms: automatic movement controlled by the cerebral cortex

Charles Darwin (1871) wrote: ‘Man alone has become a biped: and we can, I think partly see how he has come to assume his erect attitude, which forms one of his most conspicuous characters. Man could not have attained his present dominant position in the world without his use of his hands.’ Freeing the hands from locomotor duty to allow carrying, feeding and tool use has remained prominent among the evolutionary theories of how homo sapiens became bipedal. Although not as precise as when we are stationary, our extraordinary dexterity and volitional control of the forelimbs during locomotion indicates that the motor cortex can dominate drive to these muscles. However, often we don't want to do anything with our arms. Then what do we do with them?

One possibility is to do nothing – just let them hang. The rhythmic arm movements of human walking could simply be the consequence of passive locomotor dynamics, and it has been said that they act as passive inertial dampers that reduce rotation of the trunk and head (Pontzer et al. 2009) or assist in reducing metabolic energy expenditure (Collins et al. 2009). Another is that they are the result of active neural drive by a subcortical, probably spinal, locomotor pattern generator (Zehr et al. 2009). This could be a system evolved to assist our unique locomotion or a relic of our individual and collective days as quadrupeds. The constrained swing of dystonic conditions suggests an involvement of supraspinal neural drive onto the modulating activity. However, the sense of effortlessness associated with the swing of the arms as we walk certainly argues for little or no involvement from the motor cortex. In a recent issue of The Journal of Physiology, Barthelemy & Nielsen (2010) present an elegant and difficult series of experiments showing that the motor cortex does indeed contribute to muscle activity that controls the swing of the arms.

Barthelemy & Nielsen used transcranial magnetic stimulation (TMS) to seek evidence for cortical involvement in the activation of arm muscles across the gait cycle. This technique has many potential pitfalls (Burke & Pierrot-Deseilligny, 2010), which the authors have been extremely careful to avoid by validating their observations with a variety of stimulation protocols and interpreting their results with extreme caution. They first demonstrate phase modulation of TMS-evoked motor potentials (MEPs) in the deltoid muscle. The changes associated with the modulated muscle activity of walking were not different from those of an equivalent voluntary push. This MEP modulation of existing activity confirms corticospinal access to the limb muscles during walking but it reflects the excitability of spinal motoneurons, which could receive their phasic drive from non-cortical centres. It provides little information about the contribution of corticospinal projections to the on-going motoneuron drive (Nielsen, 2002).

With paired-pulse TMS, a weak conditioning pulse can activate intracortical inhibitory inputs onto corticospinal neurons 2 ms before the test excitatory pulse that generates the MEP. Measuring this short-latency intracortical inhibition across the locomotor cycle revealed phase-dependent modulation of intracortical inhibition of corticospinal drive to the upper-limb motoneuron pool. Still, this doesn't say whether or not corticospinal drive contributes to the arm muscle activity of walking because the MEP is not the natural cortical drive. Rather, it says that intracortical inhibitory neurons influence the excitability of descending corticospinal projections only if these are contributing to the arm muscle activity to begin with.

The final test was to deliver subthreshold TMS. This low-intensity stimulus does not elicit an excitatory MEP and is thought to activate only inhibitory cortical interneurons, some of which project to the primary motor cortex. Barthelemy & Nielsen (2010) show clear depression in ongoing arm muscle activity – with no sign of prior facilitation – when subthreshold TMS is delivered over the contralateral upper-limb motor cortex. These results strongly suggest that the motor cortex is part of a neural network contributing to the modulation of arm muscle activity during human walking. Similarly, potential contributions of the motor cortex to lower-limb motoneuronal drive during walking were demonstrated by the same laboratory using the subthreshold TMS technique (Petersen et al. 2001).

The subthreshold TMS technique is technically difficult and prone to inter-subject differences in activation. In this study, depression without prior facilitation of on-going muscle activity could only be obtained in 11 of 20 subjects. At this stage, however, it is probably the best tool available to investigate the contribution of the motor cortex in a motor task (Nielsen, 2002). Other stimulation techniques can provide information on task-dependent excitability of sensorimotor pathways but most fail to show with certainty the contribution of the pathway to the modulation of muscle activity during on-going, undisturbed motor behaviour.

The discovery of Barthelemy & Nielsen is an exciting advance to our understanding of human locomotion and the motor activity that generates rhythmical movement. They provide evidence that apparently involuntary rhythmical arm movements rely on a neural network that involves the motor cortex. Although these results could appear to be at odds with the spinal origin of basic locomotor activity in other vertebrates, the extent and nature of the motor cortex contribution to the entire neural network that generates human upper-limb motoneuronal activity is difficult to determine. First, motor cortical drive to upper-limb motoneurons could originate from subcortical or cortical motor centres and involve a variety of sensorimotor loops. Motor cortical drive to the arm motoneurons could also be indirect. Indeed, it is conceivable that the descending cortical output modulates activity of interneurons projecting to the arm motoneurons – for example, interneurons that are part of a subcortical or spinal central pattern generator. More research will be needed to identify and understand the neural networks driving motoneurons required for human walking. Such research will require the realisation of difficult, careful and comprehensive experiments such as that presented by Barthelemy & Nielsen (2010), along with innovative new stimulation and analysis techniques.

While evolution has freed our arms from locomotor duty, it appears that the motor cortex is still on the job. The change in excitability and cortical drive across the locomotor cycle suggests that the processes of intracortical inhibition could shape corticospinal output (Reis et al. 2008) and the movement pattern seen in the swing of the arms. Our idiosyncratic locomotor movements are signatures that allow instant recognition, often before we can identify a person by form. Who can honestly say they haven't worked at creating or copying walks of attitude, affectation or outright silliness –‘I know I have’ (Python, 1970). No doubt this is achieved by modulating output from the motor cortex. The findings of Barthelemy & Nielsen suggest that even when these patterns are or have become involuntary, the motor cortex maintains a close interest. Certainly we want to know what is happening to our limbs as we walk so that we can detect when things are not moving according to plan. Cortical efferent contribution to the locomotor network driving limb movements along with patterns of sensory reafference could be critical in providing this information.