Perceptual-Motor Recalibration on a Virtual Reality Treadmill

Abstract

 

We have demonstrated that changes in the perceptual-motor calibration of human locomotion in a real environment can be caused by manipulations of a treadmill-based virtual environment. The experiment was patterned after the real-world study by Rieser, Pick, Ashmead and Garing (1995, JEP:HPP), using a computer controlled treadmill with a 6′ by 8′ walking surface, surrounded by three 8′ by 8′ projection screens oriented to provide an approximately 180 degree horizontal field of view. In a pre-test, subjects walked without vision to previously seen targets on the floor at distances of 6m, 8m, and 10m. They then walked on the treadmill at the rate of 1.0m/s in one of three conditions: visually faster (2.0×) than the walking rate, visually slower (0.5×) and matched visual and biomechanical speeds. In a post-test, subjects performed the same visually directed walking task as in the pre-test. When exposed to the visually faster condition, subjects undershot the distances in the post-test trials by an average of 5% relative to the pre-test. Given the visually slower condition, they overshot the distances in the post-trials by an average of 9%. In the visually same condition, subjects overshot by an average of 3%. These results show that a mismatch between visual flow of a virtual environment and biomechanical walking on a treadmill will recalibrate visually directed locomotion in the real world. The effects are similar to those found using real-world visual flow and treadmill walking. As a result, we can now probe aspects of perceptual-motor calibration that have proven to be difficult to investigate outside of a virtual environment. Perceptual-motor recalibration may also provide a methodology for investigating perception of both speed and distance in virtual environments, since it remains an open question whether or not speed perception when viewing computer graphics is subject to the same compression that affects distance judgments.

 

Footnotes

 This work was supported by NSF grants IIS-00-80999 and IIS-01-21084.