Abstract
Purpose. When observers indicate the magnitude of previously-viewed exocentric spatial intervals by walking without vision, their responses often do not show evidence of the depth foreshortening typical of other behavioral indications (e.g., visual matching: Loomis et al., 1992). Is the visuomotor system accurately registering spatial layout, or is walking influenced by biases that do not depend on stimulus depth? This study tests the second hypothesis by presenting all stimuli in a frontoparallel plane, thereby removing the depth component. Method. Eleven observers binocularly viewed a single white line (23 – 183 cm long) on each trial. The stimuli were projected onto a wall in a well-lit room and were oriented either horizontally or vertically. The task was to close the eyes and reproduce the line length, by walking either straight ahead or after first turning 90 deg to the right. In a separate block, observers gave verbal estimates of line length. Results. (1) Both verbal and walking responses showed a horizontal — vertical illusion of about 13%. (2) Regardless of the line's orientation, observers walked farther when walking straight ahead than if they turned before walking. (3) Walking responses were subject to large overshooting errors (37% of the physical line length). Conclusions. Despite its reputation as an accurate indicator of spatial layout, blindfolded walking is subject to a variety of errors when used to indicate the magnitude of exocentric intervals, some of which are clearly perceptual in origin. At least some of walking's apparent insensitivity to perceptual depth foreshortening can be explained by response biases that do not depend on depth cues.