Stereoscopic perception of relative depth with reversed-contrast half
images differs in several important respects from stereopsis with matc
hed-contrast half images. Thus, reversed-contrast images show no corre
lated shift in visual direction, indicating that the sensory-fusion me
chanism ignores opposite-sign edges; one experiment addressed this asp
ect of the problem. Mainly, this was a quantitative study of opposite-
contrast stereopsis, in which stereoacuity was measured as a function
of bar width by means of narrow-band stimuli. Acuity was about an orde
r of magnitude worse for reversed-contrast than for matched stimuli, b
ut the ability to see valid (disparity-dependent) depth was not altoge
ther last even with wide (1 cycle deg(-1)) reversed-contrast bars. It
is generally believed that depth with opposite-contrast stimuli is med
iated by interaction between binocular stimuli components that have th
e same sign of contrast. Perceived depth was measured as a function of
disparity and thus one of the predictions of that 'same-sign hypothes
is' was tested experimentally; then, the magnitude of same-sign compon
ents was manipulated within the reversed-contrast stimuli, and thus th
e general prediction of the same-sign hypothesis was tested. The resul
ts show conclusively that the same-sign hypothesis cannot account for
opposite-contrast stereopsis; its mechanism remains unknown.