Abstract
Purpose. When attempting to reproduce exocentric spatial extents by nonvisual walking, observer accuracy is biased by the direction walked relative to the initial viewing direction (Philbeck, VSS '02). Here, we set out to confirm this and test whether body rotations imposed before walking are responsible.
Method. Observers saw 2 short vertical rods resting on the ground and separated by .5 to 3.5 m; they then indicated the rod separation by nonvisual walking. Exp. 1: The rods were aligned in either a frontoparallel or a sagittal plane. When responding, participants walked directly forward, or after turning 45 or 90 deg to the right. Exp. 2: The rods were aligned in a frontoparallel plane. Walking was executed along an axis PARALLEL to this plane, or along an orthogonal axis pointing to the nearest rod. Some trials required a 90 deg body rotation prior to walking while others did not. Sometimes this entailed viewing the rods over the left shoulder while straddling the PARALLEL axis; this could be followed by a leftward body turn to walk along the ORTHOGONAL axis or by no turn (walking straight along the PARALLEL axis). Observers walked without vision while facing the direction of walking.
Results: Exp. 1: Response errors varied systematically with walking direction; walking forward tended to produce larger responses than walking after a 90 deg right turn (mean errors: +30 vs. +16 cm); errors after a 45 deg turn were intermediate (+24 cm). Exp. 2: Regardless of whether or not a body rotation was required prior to responding, participants tended to walk farther when they walked toward the stimuli versus parallel to them (mean error: +41 vs. +17 cm).
Conclusions: Walking error varies systematically with walking direction when observers attempt to reproduce spatial extents. This robust effect cannot be explained by body rotations imposed before walking but appears to be related to head orientation during the preview or to locomotion toward the stimuli.