Search for a command to run...
The source of visual motion is inherently ambiguous such that movement of objects in the environment can evoke self-motion illusions and postural adjustments. Theoretically, the brain can mitigate this problem by combining visual signals with other types of information. A Bayesian model that achieves this was previously proposed and predicts a decreasing gain of postural response with increasing visual motion speed. Here we test this prediction for discrete, unidirectional, full-field visual rotations in the frontal plane of standing subjects. The speed (0.75-48 deg s(-1) ) and direction of visual rotation was pseudo-randomly varied and mediolateral responses were measured from displacements of the trunk and horizontal ground reaction forces. The behaviour evoked by this visual rotation was more complex than has hitherto been reported, consisting broadly of two consecutive components with respective latencies of ∼190 ms and >0.7 s. Both components were sensitive to visual rotation speed, but with diametrically opposite relationships. Thus, the early component decreased with faster visual rotation, while the later component increased. Furthermore, the decrease in size of the early component was not achieved by a simple attenuation of gain, but by a change in its temporal structure. We conclude that the two components represent expressions of different motor functions, both pertinent to the control of bipedal stance. We propose that the early response stems from the balance control system attempting to minimise unintended body motion, while the later response arises from the postural control system attempting to align the body with gravity.