Interaction of visual and idiothetic information in a path completion task.

To assess the contribution of visual and vestibular information on human navigation, five blindfolded subjects were passively displaced along two sides of a triangular path using a mobile robot. Subjects were required to complete the triangle by driving the robot to the starting point either blindfolded or in full vision in a 7x6-m and a 38x38-m room. Room dimensions exerted a significant effect on performances: in the smaller environment blindfolded responses were always too short whereas subjects correctly reached the starting point when visual feedback was allowed. On the contrary, in the larger room subjects correctly responded while blindfolded but drove significantly farther than requested in full vision. Our data show that vestibular navigation is highly sensitive to both stored (knowledge of environment) and current visual information.