Abstract
Chromatic information in the retina and geniculate is primarily encoded within mechanisms that compare signals in the L vs. M cones or the S vs. L and M cones. Accordingly these “cardinal directions” have played a central role in modern color science both for specifying chromatic stimuli and interpreting perceptual processes. Because individual observers vary widely in their spectral sensitivity, the stimuli that isolate the cardinal axes differ across observers. Procedures have been proposed for empirically determining the axes, yet these are cumbersome and rarely used, and most studies instead specify the directions based only on the standard observer. We examined how the cardinal directions depend on individual differences in the luminance sensitivity of observers, which are routinely measured. We modeled variations in spectral sensitivity based on estimates of the normal variability in lens and macular pigment density, cone optical density and spectral peak, and cone ratios. The tilt in the equiluminant plane along each axis was then compared to the chromatic rotations within this plane for a given observer. The degree of luminance tilt and chromatic rotation were strongly correlated along both axes (r≈0.85 for S; r≈0.80 for LM). Luminance sensitivity depends only on the L and M cones, but variability in the S cones had no (S) or little (LM) effect on the correlations. We conclude that individual differences in the cardinal directions can be approximated with good accuracy simply from the empirical luminance corrections that are already widely applied in studies of color vision.