Abstract
We previously defined a new cone difference space by choosing cone fundamentals such that the measured detection ellipses map into a circle (Teufel & Wehrhahn 2000). On a plane of equal luminance we chose 16 equally detectable colors situated in a circle around white. These colors can be described by the azimuth angle phi on the isoluminant plane. When the points representing the colors are equidistant in phi, they are equally discriminable. Projecting these points onto the cone contrast axes c(L) and c(M) and c(S) yields sinusoidal functions of the azimuth angle phi with characteristic phase angles. When these colors surround a small white test-field, human observers perceive shifts in color whose locus in cone space is measured. Plotting these shifts on the three cone contrast axes yields 2pi-periodic functions of phi. Different phase shifts are predicted depending on cone specific adaptation or opponent chromatic mechanisms underlying the perceived shifts. We showed in a previous paper that the shifts observed under binocular conditions can be explained solely by opponent chromatic mechanisms (Teufel & Wehrhahn, submitted). This indicates that under binocular viewing conditions the gains of all photoreceptors change simultaneously in the two eyes. Using dichoptic stimulation the effects of adaptation are quantified when a colored surround is presented to one eye only. This shows that the effects of cone specific adaptation as defined in our experiments are not perceived under normal (isoluminant) viewing conditions. This makes them a powerful tool for color constancy. A model that predicts excitatory interactions between color opponent units accounts for red-green induction. On the basis of our data a new inducing mechanism is postulated which consists of opposed M- and L-cone inputs with the latter having more than twice the magnitude of the former.
TeufelHJWehrhahnC(2000) J Opt Soc Am A, 17(6): 994–1006