Horizontal Disparity Pooling

Abstract

Vertical magnification in one eye causes a frontoparallel surface to appear slanted about a vertical axis. Stenton, Frisby, and Mayhew [1984 Nature (London)309 622 – 623] showed using a dot display that, for mixtures of dots drawn from two populations of vertical magnification, the amount of horizontal magnification needed to null the perceive slant varies monotonically and nearly linearly with the mixture. We find a similar, but weaker, relationship for mixtures of horizontal magnification. In one experiment, a horizontal row of dots (containing no vertical disparity) was used to measure the local horizontal magnification needed to null the slant-biasing effect of a background which consisted of a pair of transparent planes made up of dots from two populations of horizontal magnification. The nulling magnification for the dot row was generally a monotonic function of the ratio of the number of dots from the two populations. In a second experiment, observers added horizontal disparity to the biplanar display itself until one plane appeared frontoparallel. The nulling horizontal magnification was again a monotonic function of the ratio of the number of dots from the two populations. We conclude that stereoscopic slant perception is influenced by pooling both of vertical and of horizontal magnifications within a stereoscopic image.