Perceiving a Surface Behind a Contour

Sang Chul Chong and Oakyoon Cha

Abstract

In a visual scene, pictorial depth cues of overlapping are prevalent. Consequently, it is important to exclude objects in front of a surface in order to properly perceive it. The current study investigated whether the visual system excludes contours in the front to judge surface properties. Participants were asked to judge whether the average orientation of 60 oriented Gabors was clockwise or counterclockwise. In the contoured condition, 9 out of 60 Gabors were aligned to be integrated to a contour. In the scattered condition, 9 Gabors were randomly intermixed among others. In the grouped condition, 9 Gabors were aligned perpendicular to their orientation, and thus were not perceived as a contour. Only in the contoured condition, participants judged the average orientation as the opposite orientation to that of the contour. We hypothesized that this trend was due to participants omitting the orientations of 9 Gabors in a contour in computing the average orientation. To test this hypothesis, we fit each participant’s responses in the scattered condition to a simple drift-diffusion model. In the drift-diffusion model, signals plus noises were accumulated to a certain threshold at which the decision is made. Average orientation of 60 Gabors plus Gaussian noise summed up from 60 noise sources were accumulated as evidence. Using the fitted noise level and threshold for each of the participants, we simulated each participant’s responses in the contoured condition by accumulating signals and noises from only 51 Gabors excluding Gabors in a contour. The simulation results well described the pattern of behavioral responses. In conclusion, behavioral responses and the modeling results suggest that visual features of a contour are ignored when one judges the overall property of a surface.