We usually think that we – meaning our nervous systems – are in control of what our bodies do. Our intuitions lead us astray: in many cases, it seems that body movements only result from interactions among the musculoskeletal system, the external world and the nervous system. In rapid behaviors like running, physical interactions can be dominant, and the nervous system – particularly higher centers in the brain – may only have a small supporting role.

In a recent paper in Physical Review E, Shinya Aoi, Tsuyoshi Yamashita and Kazua Tsuchiya provide a nice example of this complex coupling between nervous system, body and environment. Using a computational model, they examined how gait changes come about – why a dog might shift from a walk to a trot or vice versa – and found that coupling between the body and the environment could cause the gait to change, with no explicit shift in the neural control pattern.

To examine this effect, the researchers built a computational four-legged animal. It has a ‘nervous system’ – a simulation of a neural circuit, present in the spinal cord of all vertebrates, called a central pattern generator (CPG), that produces the pattern of muscle activity for locomotion. The CPG controls a ‘body’ – a mechanical model of the four legs, each with a hip and a knee, but no ankle. Crucially, the forelimbs and hindlimbs are linked by a rotational spring joint at the waist. Each leg can contact the ‘environment’, which has a slightly springy surface. And finally, the CPG receives sensory input by phase resetting: each time a leg reaches a particular angle forward, the CPG is pushed towards a specific phase in the step cycle.

The researchers set up the model so that each pair of left and right legs alternated with one another, but they didn't specify how the forelimbs and hindlimbs should be coordinated. They found two stable coordination patterns: a pace, in which both legs on the left side alternate with both legs on the right; and a walk–trot pattern, in which the model shifted from walking at low speeds to trotting at high speeds.

There was not a single speed at which the model shifted from a walk to a trot; instead, the shift speed depended on whether the model's speed was increasing or decreasing. If the model speeded up from a slow walk, it only shifted to a trot at a relatively high speed. If it slowed down from a high speed trot, it carried on trotting for a long time, even at speeds much slower than the first transition speed (when speed was increasing and the model shifted up to a trot), until the quadruped finally shifted to a walk at a relatively low speed. This effect, called hysteresis, has been observed in cats, horses and humans, among others.

Interestingly, the hysteresis also arose when the researchers changed the stiffness of the waist joint. At low stiffness, the model would trot; at high stiffness, the model would walk; but in between, it depended on whether the stiffness was increasing or decreasing.

In Aoi's model, the hysteresis in the gait transition was purely an effect from the body's interaction with the ground. For animals, the nervous system probably contributes to some degree, but these results suggest that mechanics may play a crucial role in determining how and when a gait changes.

Hysteresis in the gait transition of a quadruped investigated using simple body mechanical and oscillator network models
Phys. Rev. E