Monocular location provides a signed 'depth signal' for discriminating the sign of large disparities

Purpose. Current models of disparity processing predict that large disparities are detected by coarse-scale mechanisms. However, observers can correctly identify the sign of large diplopic disparities, even if the targets are composed of high frequency information at low cortrast (Wilcox & Hess, Vis. Res., 1995). This suggests the existence of an alternative mechanism that processes large disparities. Here we tested whether the location of a single monocular half-image (nasal or temporal) provides a signed depth signal that could contribute to the alternative mechanism. Methods. Flanked by a pair of black reference lines, the stimulus was a bandpass target (center 8cpd) consisting of one high contrast half-image paired at random with either a low-contrast, or a blank, half-image. The target had a large disparity (up to 90 min) which was randomly either crossed or uncrossed. The bright half-image was presented to either eye at random so that disparity sign did not correlate with its visual direction (left or right of centre). Observers were asked to discriminate the disparity sign of the target. Results. When one eye contained a bright half-image and the other a low-contrast half-image, performance was well above chance for a range of large disparities (6-60 min). Observers were also able to detect the depth sign of large 'disparities' when one eye contained no target at all, but performance was poorer than when a low-contrast half-image was present. Conclusion. Large disparities can be processed by a mechanism that utilizes information about monocular target location. However, monocular location cannot completely account for the disparity sign discrimination performance, because a near-threshold half-image allows better discrimination than a zero-contrast half image.