Abstract
Perceiving opponent hues (e.g., red and green) as components of a uniformly colored region of space cannot be explained within Hering's opponent theory. Here we demonstrate that the classical formulation of this theory cannot account for perception of colors resulting from certain chromatic contrasts. In a series of stimuli generating a continuum of colors produced by complementary chromatic induction, our subjects were asked to set bounds along the continuum on when they first (or no longer) saw “redness” or “greenness.” The results demonstrate an overlap between the set of colors in which a red component is perceived and the set of colors in which a green component is perceived. This overlap constitutes perception of opponent mixtures explicitly forbidden by the opponent theory. In a control stimulus sequence, in the absence of complementary chromatic induction, the two sets of colors do not overlap, which is consistent with the classical prediction of red and green being mutually exclusive. Our finding and previous results (Crane & Piantanida, 1983, Science, 221:1078–1080; Billock et al. 2001, JOSA A, 18, 2398–2403) support a revision of Hering's theory. We conclude that the opponent structure of classical perceptual color space results not from opponent hues occupying polar positions of a single perceptual dimension, but rather from a projection of the four-dimensional unipolar chromatic hyperspace onto a subspace whose dimensions emerge in response to the visual environment. Such a space allows any unique hue combinations but typically gets reduced to unipolar two-dimensional chromatic percepts. Showing that “forbidden” hue combinations, previously reported only under artificial image stabilization and in our own lab using neon color spreading, can be present in normal viewing conditions opens a new paradigm in the experimental study of the dimensionality and structure of perceptual color space.
Supported in part by NSF SBE-0354378 at Boston University.