The visual system relies on its ability to represent the ground surface for use as a reference frame to localize object. To learn the temporal requirement for forming the ground surface representation, we varied stimulus duration (0.15, 0.5, 5s) to uncover its impact on judged target accuracy and precision under various conditions. A 0.27deg light target was placed either on the floor or 0.5m above, at a distance of 3.75, 5 or 6.25m. Judged distance and height were measured with the blind walking and gesturing tasks. Exp 1-Monocular view in the dark: target duration did not affect the accuracy of position (distance between the judged and physical target positions) nor angular declination. This finding is not surprising since judged location in the dark is determined by the internal bias of the visual system and angular declination (Ooi et al, Nature ‘01). However, the precision for angular declination (within-subject standard deviation of repeated measurements) improved with target duration. Exp 2-Monocular view with parallel vs convergent LED (2x3 array, distance range: 1.5-4.5m) background: predictably the convergent array creates a false perspective that shifts the perceived eye level downward (Wu et al, VSS03) leading to reduced judged angular declination. Supporting this, we found that the false perspective reduced judged angular declination, and similarly so for all three durations tested. Together, Exps 1 and 2 suggest that the process of representing the ground surface reaches an asymptotic performance for accuracy by 0.15s, consistent with Philbeck’s (Perception ‘00) finding in the light environment. Exp 3-Monocular vs binocular view with parallel LED-array background: for targets 0.5m above the ground, position accuracy was better with binocular than monocular viewing with test duration. This means that relative binocular disparity with respect to the ground is used for egocentric localization of a suspended target (Meng & Sedgwick, P&P ‘01).

Support: NIH grant EY014821