Rather than a low frequency first-order stereo mechanism, the edge disparities could be mediated by responses in the second-order stereo mechanism that has received so much attention in the past decade (Hess & Wilcox,
l994; Wilcox & Hess,
1995,
l996,
1997; Schor, Edwards, & Pope,
l998; Edwards, Pope, & Schor,
l999; Langley, Fleet, & Hibbard,
1998,
l999). It is assumed that this putative mechanism responds to the disparity of the stimulus envelope, rather than to the disparity of features within the envelope. The evidence for second-order mechanisms comes from studies showing that observers can combine half-images having opposite polarity (Pope, Edwards, & Schor,
l999), different spatial frequencies (Langley, Fleet, & Hibbard;
l998), orthogonal orientations (Schor, Edwards, & Sato,
2001), uncorrelated random carriers (Elder & Wilcox,
2000) and envelopes of different sizes (Schor et al.,
2001). Like the second-order mechanisms for texture and motion, the nonlinearity that produces the envelope signal is thought to be a type of neural rectification (Elder & Wilcox,
2000; Schor et al.,
2001). Langley et al. (
l999) have argued persuasively that this nonlinearity is cortical, and Wilcox and Hess (
1996) have provided evidence that it occurs before the monocular signals are combined binocularly.