Lizhu Yan¹ (), Lingling Bai¹, Teng Leng Ooi², Zijiang He¹; ¹University of Louisville, ²The Ohio State University

Reliable ground surface representation is vital for accurate absolute distance judgment. We proposed the ground surface is represented by a Sequential Surface Integration Process (SSIP) (Wu et al., 2004, Nature), wherein the ground around the observer’s feet is represented before the farther regions. This way, the visual system can use the near depth cues to form a reliable ground representation, which is then used to sequentially integrate with texture gradient cues in the more distant ground regions to form a global ground representation. We further tested the SSIP hypothesis by varying the spatiotemporal availability of external depth cues. Four conditions were tested with binocular and monocular viewing. (i) Dark: no texture. (ii) Full-texture: a 2x5 parallel texture array (4s) spanning 2-8 m in the dark. (iii) Near-to-far sequence: the texture pair nearest to the observer at 2 m was first presented (4 s) and followed sequentially by the 3.5m, 5m, 6.5m, and 8m pairs (0.75 s each). The presentation sequence was run twice. (iv) Far-to-near sequence: the texture presentation sequence was the reverse of the Near-to-far condition. The test target (0.20°, 2s) was presented in the dark 1s after the texture removal. The observers’ (n=8) task was to judge the target location (4.5, 5.75, 7.0m @ 0.14m height and 5.75m @ 0.5m height) using the blind walking-gesturing paradigm. Judged distances in the Near-to-far and Full-texture conditions were similar and most accurate. Confirming the SSIP hypothesis, judged distances were significantly more accurate in the Near-to-far than the Far-to-near conditions (p<0.001). The data trend was similar with binocular and monocular viewing (p>0.05). No significant difference was observed between the Far-to-near and Dark conditions (p>0.05). Judged angular declination was similar among the four conditions (p>0.05), suggesting observers maintained reliable angular declination even as distances could not be accurately judged.

Acknowledgements: NIH R01EY033190