Is locomotion guided by general-purpose solutions that apply across similar tasks, or by task-specific solutions that depend on task constraints? I will provide evidence from two domains suggesting that the information and control strategies used to perform similar actions differ from task to task.

To walk to a stationary target, one could simply align the direction of locomotion with the target by turning by an amount corresponding to the angle that separates the direction of locomotion and the target. Although this simple strategy may be effective for legged locomotion, it may not apply to the case of steering a vehicle for at least two reasons. First, most vehicles are designed with controllers that produce changes to the curvature of the observerís path, rather than to the direction of locomotion. Thus, whereas control strategies for walking need to inform the observer how to change direction of locomotion, strategies for steering should inform the observer how to change path curvature. Second, because vehicles have greater mass and travel at higher speeds than walking observers, the suddenness with which changes in direction can be made is limited by momentum. Thus, whereas walking observers can make nearly instantaneous changes in direction, vehicle controllers are forced to change directions gradually by making gradual changes to path curvature. In a computer simulated driving task, drivers modulated steering on the basis of information about the time-to-passage of the target, turning more rapidly toward nearby targets. Because driving introduces constraints on the rate of turn, drivers and walkers appear to use different information and strategies to guide locomotion to a target.

Walking observers also appear to use different information and strategies to walk to stationary and moving targets. Warren, Kay, Zosh, Duchon & Sahuc (2001) found that observers use both egocentric direction and global optic flow to align their direction of locomotion with the stationary target. To walk to a moving target, at least two control strategies could are possible. (a) Centering: keep the direction of locomotion aligned with the target (similar to walking to a stationary target). (b) Interception: maintain a constant heading angle between the direction of locomotion and the target. Both strategies could rely on either the egocentric direction of the goal, the global optic flow of the stationary background, or the local optic flow of the target. In a series of experiments in which observers walked to a moving target, I found that observers adopt an interception strategy by maintaining a roughly constant non-zero angle between the direction of locomotion and the target. Behavior was also unaffected by manipulations of the global and local optic flow, suggesting that observers rely upon egocentric direction to intercept a moving target. Why does global optic flow influence walking to a stationary target, but not walking to a moving target? To walk to a stationary target, observers could simply null the motion parallax around the target. Because motion parallax is affected by object motion, however, it does not provide useful information for walking to a moving target. This provides further evidence that the control of locomotion depends on constraints that are specific to the task.

References:

Warren, W. H. J., Kay, B. A., Zosh, W. D., Duchon, A. P. and Suhac, S. (2001). Optic flow is used to control human walking. Nature Neurosci. 4(2): 213-216.