Abstract
The steering dynamics model of human locomotion (Fajen & Warren, 2003) demonstrates that steering could be controlled on the basis of current visual information, without relying on an internal path plan or model of the environment. To investigate whether locomotion is under on-line or off-line control, participants walked a slalom course of five goals while viewing conditions were manipulated. Virtual displays of goal poles (1.44 m high,.12 m diameter) standing vertically on a ground plane were presented stereoscopically in a head-mounted display (63°H × 53°V) in the VENLab (12 m × 12 m), while head position was recorded. The task was performed under five viewing conditions: (1) FullView: all goals in the slalom course were visible. (2) Preview1: only one goal immediately ahead of the participant was visible. (3) Preview2: the two goals immediately ahead were visible. (4) Block1: all goals were visible except the one immediately ahead. (5) Block2: all goals were visible except the two immediately ahead. If the path of locomotion is planned in advance or based on an internal world model, one might expect steering to improve with a longer preview; more critically, the effect of occluding the next couple of goals should be minimal. Analysis of passing errors indicated that steering was no more accurate or precise in the Full View than the Preview 1 condition (ns), indicating no advantage of a longer preview. In contrast, performance significantly deteriorated as visual information about the next one or two goals was removed (p < .001). This pattern of results is consistent with the view that human locomotion is controlled on-line based on current visual information about the next one, or possibly two, goals.
Supported by NIH R01 EY010923.