Splitting the difference: New insights into distinguishing features of human versus feline models of adaptive locomotor control

If you open most any neuroscience textbook and begin reading about the neural control of locomotion, you are almost guaranteed to find information gleaned from experimental studies of cats with spinal transections. Since the turn of the 20th century, these studies have demonstrated the remarkable capacity of the spinal cord to generate hindlimb locomotion and have contributed to the prevailing idea that pattern generating networks within the spinal cord are primarily responsible for coordinating mammalian locomotion. In recent years, studies of non‐invasive spinal stimulation have supported the hypothesis that spinal pattern generators may exist in humans (Gerasimenko et al. 2015). The potential existence of human spinal pattern generators has sparked a range of therapeutic interventions for patient populations as varied as individuals with spinal cord injury, stroke survivors and people with Parkinson's disease. Despite the fact that studies of the spinal control of locomotion have dominated our collective understanding of locomotor control, few investigations have examined the degree to which spinalized animals can replicate key features of human locomotion.

One of the central features of human locomotion is that we typically move our legs in an alternating, antiphase manner and take steps of similar length. This pattern is also accompanied by equal double support times, which correspond to the time within a gait cycle when both feet are on the ground. Each of these variables, relative phasing, step length and double support time, reflects features of interlimb coordination, and right–left symmetry of these features is a hallmark of healthy gait. In fact, even when people are exposed to the unnatural task of walking on a split‐belt treadmill, where one belt moves faster than the other (Dietz et al. 1994; Reisman et al. 2005; Finley et al. 2015), they eventually adopt a walking pattern characterized by taking steps of equal length. This behaviour is quite robust as split‐belt adaptation can be observed across a wide range of ages and in people with neurological impairments that spare the cerebellum. Given the robustness of changes in interlimb coordination during split‐belt adaptation in humans, one might assume that this behaviour would be observed in animal models of locomotion that have classically been used to understand the neural control of mammalian locomotion. One way to test this hypothesis, and perhaps gain insight into the role of spinal circuits in the control of interlimb coordination, would be to evaluate the capacity of animals with spinal transections to adapt to walking on a split‐belt treadmill.

A paper by Kuczynski et al. (2017) in this issue of The Journal of Physiology does precisely this by examining whether cats demonstrate similar changes in interlimb coordination to humans when adapting to walking on a split‐belt treadmill. Similar to results from humans, when cats with complete spinal transections initially walked on a split‐belt treadmill, they exhibited an abrupt increase in interlimb asymmetry characterized by right–left differences in step length and double support times. However, unlike the behaviour routinely observed in human studies, spinalized cats failed to modify interlimb coordination and reduce asymmetries in step length or double support times after extended exposure to split‐belt walking. Moreover, experiments in neurologically intact cats, who had both spinal and supraspinal networks available to facilitate adaptation, failed to adapt interlimb coordination. In other words, the absence of interlimb adaptation in spinalized cats is not merely a phenomenon that results from a lack of supraspinal input.

While the differences between the results presented by Kuczynski et al. and those from previous human studies are readily apparent, the reason for these differences remains to be determined. The authors offer two potential explanations for this discrepancy. First, they suggest the possibility that balance requirements may partially drive adaptation in humans. Indeed, bipedalism presents unique challenges to balance control, and it is likely that the locomotor control centres within the human nervous system have adapted novel strategies to maintain balance and coordinate the legs during walking. However, it has yet to be established if maintaining interlimb symmetry is the best strategy to achieve dynamic balance, particularly in the context of walking on a split‐belt treadmill. A second intriguing explanation that the authors propose is that these results may reflect intrinsic differences in the flexibility of interlimb control between humans and quadrupeds. For example, intact and spinalized cats routinely switch between antiphase gaits, such as walking, and in‐phase, non‐alternating gaits such as galloping. In contrast, humans prefer antiphase, alternating locomotor patterns and rarely hop (in phase locomotion) as a chosen form of locomotion. Therefore, the way in which we adapt to walking on a split‐belt treadmill may reflect an experience‐dependent bias toward antiphase alternation.

In summary, the results from Kuczynski et al. are significant in part because they identify a fundamental difference in the control of interlimb coordination during walking in humans and cats. Since the maintenance of interlimb symmetry is an essential feature of normal human walking, these results highlight limitations in the degree to which our understanding of locomotor control in the feline model provides meaningful insight into the neural control of human locomotion. Indeed, this study, and others before it, are important reminders of the need to identify inter‐species differences, and not just similarities, in the neural circuits subserving the control of motor behaviour. Acknowledging these differences can be a powerful tool for driving further inquiry into the phylogenetic development of neural control strategies unique to human motor control.

Additional information

Competing interests

None declared.