How Do We Tell Colors Apart? — Cone Cells and Opponent Processing
Follow light as it stimulates the retina's three types of cone cells and the brain re-sorts those signals into red–green, blue–yellow and brightness channels, to understand how color discrimination works.
Color is not a property attached to objects; it is a perception created jointly by light, the eye and the brain. How well you tell colors apart is shaped by the characteristics of each of these three stages.
Stage 1: Three types of cones
The retina has three types of cone cells that handle color in bright light. They are usually named after the wavelengths they respond to best: L (long-wavelength, peaking around 560nm), M (medium-wavelength, around 530nm) and S (short-wavelength, around 420nm) cones.
A single cone cannot report wavelength. Whatever wavelength it absorbs, it just fires the same kind of signal. Color is determined by the ratio of responses across the three cones. That is why two lights with completely different spectra look like the same color if they stimulate the three cones in the same ratio (metamerism). It is also why a screen can mimic countless colors using only red, green and blue light.
The sensitivity curves of the L and M cones overlap heavily. Red–green discrimination comes from the small difference between these two curves, so if one cone type is genetically missing or its curve is shifted, red–green discrimination weakens. Studies report red–green color vision deficiency in about 8% of men and about 0.5% of women of Northern European descent (Birch, 2012).
Stage 2: Opponent channels
Cone signals do not travel to the brain as they are. On the way through the retina and the thalamus they are recomputed as differences.
- Brightness channel: the sum of L + M signals. It handles light and dark and the shapes of things.
- Red–green channel: the difference L − M.
- Blue–yellow channel: the difference S − (L + M).
This arrangement is called opponent processing. Hering proposed it in the 19th century based on the observation that nobody can imagine a “reddish green” or a “yellowish blue”, and Hurvich and Jameson (1957) backed it up quantitatively. The a* (green–red), b* (blue–yellow) and L* (lightness) axes of CIELAB are modeled on this structure.
Where are we most sensitive?
If you gradually shift a single wavelength of the spectrum and measure the smallest wavelength change people can notice, sensitivity is sharpest around cyan (about 490nm) and yellow–orange (about 580–590nm) and dullest at both ends of the spectrum (Wright & Pitt, 1934). These are the regions where the signals of the two opponent channels change most steeply with wavelength.
The special case of S cones
S cones make up only about 5–10% of all cones and are almost absent from the very center of vision (the innermost part of the fovea). As a result, differences in the blue–yellow direction are especially hard to see over small areas. This is one reason blue and yellow rounds feel harder once the grid grows to 8×8 and the squares get small.
Stage 3: The brain's interpretation
The brain reinterprets color using the surrounding context. The same gray looks lighter on a dark background and darker on a light one (simultaneous contrast). Color constancy, which lets you see white paper as white even when the lighting changes, is another correction made by the brain. This test fixes the area around the grid to neutral gray so these context effects do not vary from round to round.
Age and color discrimination
With age, the lens gradually yellows and lets in less short-wavelength light. As a result, blue–yellow discrimination tends to decline first, and the effect is more pronounced in dim lighting (Knoblauch et al., 1987). Even for the same person, results can fluctuate with fatigue, lighting and time of day.
What this test measures
HueSense looks at the end result of all three stages, namely your ability to notice the difference between two colors on a screen. It is not a medical test of individual cone function, and it does not determine whether you have a color vision deficiency.
References
- Hurvich, L. M., Jameson, D. (1957). An opponent-process theory of color vision. Psychological Review 64(6), 384–404
- Wright, W. D., Pitt, F. H. G. (1934). Hue-discrimination in normal colour-vision. Proceedings of the Physical Society 46(3), 459–473
- Birch, J. (2012). Worldwide prevalence of red-green color deficiency. JOSA A 29(3), 313–320
- Knoblauch, K. et al. (1987). Age and illuminance effects in the Farnsworth-Munsell 100-hue test. Applied Optics 26(8), 1441–1448