Measuring the depth induced by an opposite-luminance (but not anticorrelated) stereogram

The same-sign hypothesis suggests that only those edges in the two retinal images whose luminance gradients have the same sign, known as same-sign edges, can be stereoscopically fused to generate a perception of depth. If true, one would expect that the magnitude of the depth induced by an opposite-luminance stereogram (eg one where the figure in one stereo half-image is black and the figure in the other is white) should be determined by the disparity of the same-sign edges. Despite the considerable work on the same-sign hypothesis this prediction has yet to be verified. Here we confirm this prediction for a particular opposite-luminance stereogram and discuss possible reasons why it is not true for opposite-luminance stereograms that are presented briefly or where each stereo half-image contains many elements.