Abstract
The visual system integrates the intrinsic bias, an internal model of the ground surface, with external depth information to form a ground surface representation. When walking a short distance from one’s home-base in the dark, the intrinsic bias remained anchored to the home-base through path-integration, indicating utilization of an allocentric reference frame (Zhou et al 2023, eLife). Here, we investigated whether the ground reference frame is also allocentric when one walks toward a goal-target in a depth cue-impoverished environment. During the experiment, the observer saw a goal-target (1.5 cd/m2) at 3.5 m amongst an array of 2x8 parallel texture elements (2.25-9.25 m) on the floor. The goal-target was extinguished 5 sec later, and the observer walked to the remembered goal-target location. But before the observer reached the goal-target location, after having walked for either 1.25, 1.75 or 2.75 m, the texture elements were extinguished. This signaled the observer to stop walking. Ten seconds later, a test-target (0.16 cd/m2) was presented (2 sec) for the observer to judge its location using the blind walking-gesturing task. Four test-target locations were measured: three distances (4.50, 5.75, 7.00 m) with 0.14 m elevation and one distance (5.75 m) with 0.5 m elevation. The average results (n=6) show judged test-target distances were longest when the observers’ initial walk toward the remembered goal-target location was stopped earlier (1.25 > 1.75 > 2.75 m); that is, when they stopped further from the goal-target (p<0.001). This indicates that upon seeing the goal-target, the intrinsic bias shifted toward the goal-target. Thus, as the spatial interval from where the observers stopped walking (to the goal-target) to the shifted intrinsic bias gradually shortened (at 1.75 and 2.75 m), the perceived distance of the test target decreased. Altogether, our data reveal the visual system utilizes an allocentric ground surface representation during navigation.