P1-22: Upper-Lower Asymmetry in Slant Perception and Natural Scene Statistics

This study investigates the upper-lower asymmetry of slant perception by measuring our daily environment and conducting a psychophysical experiment. We hypothesize that the performance of slant perception varies according to the viewing direction because the visual system adapts to the probability distribution of environmental slants. In the environmental measurement, we sampled 17 locations (2 indoor and 15 outdoor ones) from our daily environment. At each location, a 3D laser scanner was used to measure the spatial layout by collecting radial distances to the surrounding surfaces. On the basis of the measurements, we calculated the probability distribution of the magnitude of surface slant, which was analyzed for each upper and lower area with respect to the eye level. Consequently, the distribution for the lower area, compared to that for the upper area, showed a remarkably high peak at the horizontal (i.e., ground-like) slant. In the psychophysical experiment, participants observed a surface with a random-dot texture, which was presented in the upper and lower viewing directions. For each direction, the participants performed two tasks: adjusted the surface's slant to be vertical (i.e., wall-like) and horizontal (i.e., ceiling- or ground-like). The results show that the response is most accurate for the horizontal task in the lower direction. This finding, coupled with the distribution of environmental slants, suggests that the visual system is sensitive to the ground-like slant that often exists in our daily environment. This asymmetrical property of slant perception is consistent with our hypothesis