Abstract
The ground surface can be used as a reference frame for coding object location (Gibson, 1950; Sedgwick, 1986). But at times in the real world scene, we do not see the full expanse of the ground surface from the feet to the target object. The direct path could be obstructed by another object (obstacle) such as a sofa or an opaque fence. While the obstacle does not in itself occlude the view of the target, we wonder if the partial occlusion of the ground surface leads to errors in judging the target location. To test this, observers performed four different distance judgment tasks on the ground from a viewing distance of 5, 6 and 7 m. In the occlusion condition, a box (51 × 165 × 54 cm) was placed on the ground in between the target and observer. The distance between the box and the observer was fixed at 2 m. Compared to baseline (no occluding box), observers significantly underestimated target locations in tasks that required them to judge the target location along a direct path through the occluding box. These include the perceptual matching, throwing, and blindfolded walking tasks, which were performed after removing the box. Performance, however, was accurate in a fourth task — a “triangulation” blindfolded walking task (Philbeck et al, 1997) — that required observers to perceptually bypass the occluding box by looking to the side of the box and then to the unobstructed view of the target. Altogether, our results show that accurate distance computation is achieved when the entire ground surface is sampled, even if having to use a roundabout strategy. Further, this raises the possibility that a single target may not always be judged at a unitary location, but depends perhaps, on how the visual system picks the information.
Supported by a research grant from A&S (UofL); SCO Research Funds