Near-optimal spatial integration of optic flow information for direction of heading judgments

Abstract

Purpose: We tested whether human observers maintain ideal patterns of information use across the visual field (VF) in determining direction of heading (DOH) from optic flow. Methods: Six participants aged 18–22 fixated centrally while monocularly viewing flow patterns simulating straight-line movement in depth through a 3D cloud of dots. DOH on base trials (20% of all trials) was set at 3°, 6°, or 9° eccentricity along the 4 diagonal axes of the VF. On perturbed trials (half of all trials), three VF quadrants indicated 6° DOH along one axis and one quadrant indicated 3° or 9° DOH along the same axis (a 3° perturbation). On 30% of trials, the DOH was chosen randomly from a circle with 10° radius. Stimulus information within 10° of fixation and outside 20° eccentricity was occluded. Subjects indicated their perceived DOH with an optical pointer. Multiple linear regressions were performed on DOH responses to determine relative weights given to each quadrant. Quadrants were labeled spatially, e.g. upper-right and lower-left, and functionally, e.g. DOH-containing quadrant, adjacent quadrants and opposite quadrant. Results: Subjects gave greatest and lowest weight to the quadrants containing the most and least useful information respectively, as predicted by the Crowell & Banks (1996) ideal observer model. Quadrants adjacent to that containing the DOH also differed in horizontal and vertical influence on perceived DOH. However, while an ideal observer would weight each spatial quadrant equally across the information-balanced conditions, human observers deviated from this pattern, giving more weight to quadrants above the horizontal midline. Conclusions: While the ideal observer model explains some of the complex patterns human observers exhibit in weighing information across the visual field, it did not predict their bias to rely on the upper half of the visual field in determining direction of heading from optic flow.

NIH Training Grant 5T32-EY007125 and NIH Core Grant P30EY0131 to the Center for Visual Science, R01-EY017939 to DK, and an unrestricted grant to the University of Rochester's Department of Ophthalmology from the Research to Prevent Blindness Foundation. KRH is supported by a Lew R. Wasserman Merit Award from the Research to Prevent Blindness Foundation.