Abstract
Conventional trichromatic color vision models1 postulate that long- (L), middle- (M), and short-wavelength (S) cone signals reorganize into two color-opponent channels where L+S (red) and M (green) cone signals cancel each other, and L+M (yellow) and S (blue) cone signals cancel each other.
This scheme is inconsistent with the hues that remain when S cone or L+M cone contributions are removed.2 When S-cone contributions are removed (e.g., by foveal or small field tritanopia, S-cone saturation, genetics, disease, toxicity, or light damage), vision becomes dichromatic, with short wavelengths appearing greenish blue, long wavelengths appearing reddish orange, and wavelengths near 570 nm appearing achromatic. Conversely, when L+M cones are suppressed by white light adaptation, the S-cone hue signal appears more red than blue.
These findings imply that L cones signal reddish orange, M cones signal greenish blue, and S cones signal bluish red. Also, S and M cones modulate color appearance in a complex manner, with S cone red cancelling M cone green to reveal blue near 470 nm and L+M cone yellow cancelling S+M cone blue to reveal green near 500 nm. Also, since removing S cones removes yellow, yellow must be a contrast color that requires an S-cone off signal. Similarly, selectively suppressing L+M yellow signals may weaken S cone blue signals by reducing chromatic contrast.
Some recent single-cone color naming results3,4 seem inconsistent both with classical color opponent models and the more complex model above. Ultimately, however, models of opponent-color organization must explain both single-cone and conventional color appearance.
The research upon which this presentation is based was supported in part by National Eye Institute grant EY07327. The author has no commercial relationships related to this work.