Abstract
Background: In binocular stereopsis, the differential-perspective transformation between the two eyes (vertical disparities) provides an estimate of the included (vergence) angle that can be used to scale horizontal disparities (Rogers and Bradshaw, 1993, Nature, 361). In observer-produced motion parallax, the vertical-perspective transformation over time (Braunstein, 1977, Perception & Psychophysics, 21) provides an estimate of the observer's angle of rotation with respect to the scene that could also be used to scale perceived depth. Both transformations represent ‘whole field’ properties that could be integrated over the entire visual field. Purpose: The present experiment was designed to measure and compare the spatial integrative properties in the two domains. Methods: Large field displays (55° × 55°) were used to present horizontally-oriented sinusoidal corrugations in depth either binocularly (disparity information) or monocularly, (observer-produced parallax information), within a textured surround that could be masked down to 35° × 35°; 20° × 20°; or 10° × 10°. Force-choice thresholds for discriminating a difference in perceived depth were obtained firstly (in the conventional way) by varying the amount of disparity/parallax motion and secondly by varying the amount of differential perspective information (mimicking a different included angle) or vertical perspective information (mimicking a different angle of rotation). Results: Conventional Weber fractions for discriminating disparity- and parallax-defined surfaces were between 7–9% and 8–11%, respectively for different observers. The slopes of the psychometric functions obtained when the amount of differential- or vertical-perspective was manipulated decreased with a decrease in the overall size of the display, in both domains. Differential-perspective manipulations did not affect perceived depth when the display size was smaller than 10° × 10° while vertical-perspective manipulations were effective in parallax displays down to 5° × 5°. Conclusions: There is a similar pattern of spatial integration for differential- and vertical-perspective information that extends over a large portion of the visual field but the minimum critical area is significantly smaller in the parallax case.