Abstract
If stereo channels of different scales were truly independent of each other, the components of a surface with a large disparity would appear smeared out across space. Instead, we see a single solid surface. To identify factors behind this perceptual coherence, we investigated interactions among the components of a squarewave grating. The target stimulus consisted of squarewave components with independently manipulated disparities. The fundamental (F) was placed in one depth plane, while the higher frequency components (mF—missing fundamental) were placed together in another depth plane. The task was to compare the perceived depth of the whole target stimulus to that of a conventional squarewave with the same spatial frequency (0.4 cpd) and contrast (20%) but varying disparity. For all four observers, F and mF cohered in depth: The appearance of the target was indistinguishable from a conventional squarewave when F and mF were in cyclopean phase. The perceived depth of the target was consistent with the disparity of mF. Randomizing the relative phase between F and mF changed the appearance of the target, but not its perceived depth. Removing the 3F component from mF changed the appearance of the target but did not alter the coherence. For one of the two observers, this switched the perceived depth to be consistent with the disparity of F. But increasing the contrast of the 5F component re-established the dominance of the higher harmonics. Perceived depth is determined by interactions among stereo spatial frequency channels. These interactions can be either coarse-to-fine or fine-to-coarse. The direction of the interaction depends on the relative salience of all the competitive scales. Therefore, Marr and Poggio's coarse-to-fine direction of interaction is not unique.