Protanopia, Deuteranopia, Tritanopia: Designing for Color Blindness

Color blindness comes in three main types, one per cone class in the eye: protanopia (long-wavelength “red” cones absent), deuteranopia (medium “green” cones absent) and tritanopia (short “blue” cones absent). Protanopia and deuteranopia both collapse the red–green axis and together make up the common red–green color blindness; tritanopia collapses blue–yellow and is rare. Designing for all three rests on one move: separate meaning by lightness, because lightness is the one channel that survives every type.

Some form of color vision deficiency (CVD) affects roughly 8% of men and about 0.5% of women — so any product with a real audience has affected users right now. This article is about the understanding: what each type is and what it actually experiences. The step-by-step testing routine — simulate, find the collapsing pairs, confirm the fix — lives in our contrast guide and its color-scales companion; here we work from the physiology down to the design rules it implies.

What are the three types, by cone class?

Human color vision uses three cone types, each most sensitive to a band of wavelengths: long (L), medium (M) and short (S). Color is the brain’s read of how strongly the three respond. Remove or weaken one, and a whole axis of that comparison goes quiet.

• Protanopia — L-cones absent. The red end of the spectrum dims and reds slide toward green. A distinguishing trait: reds also look darker than they do to typical vision, because the missing cones were the ones catching much of red’s brightness.
• Deuteranopia — M-cones absent. The most common dichromacy. Red and green collapse toward a shared yellowish-brown, with less of the darkening protanopes report.
• Tritanopia — S-cones absent. The blue–yellow axis collapses instead; blues drift toward green and yellows toward pink or gray. It is rare — well under 1% — and, unlike the red–green forms, affects men and women at similar rates.

Most CVD is not the full “-opia,” though. The anomalous trichromacies — protanomaly, deuteranomaly, tritanomaly — are cases where a cone is shifted rather than absent, weakening an axis instead of deleting it. Taken together these milder forms are actually more common than the full dichromacies; the dichromacies are the strong end of the same spectrum, and the standard design target because a palette that holds up for them holds up for the milder forms too.

What do color-blind viewers actually see?

Not gray. This is the misconception worth correcting directly: the most common forms of color blindness do not turn the world monochrome. Red–green CVD preserves lightness perfectly and leaves the entire blue–yellow axis intact — a red–green viewer sees a vivid blue sky, yellow sun and the full range of light-to-dark exactly as anyone does. What collapses is one axis of hue distinction, not color itself.

That is the whole design insight in one sentence: certain hue pairs merge into one hue, while lightness and the surviving axis stay reliable. The world does not go gray; a specific set of confusions appears.

Which palette pairs collapse under each type?

Under protanopia and deuteranopia, anything separated only along red–green:

• success-green against error-red status colors — the classic failure;
• red-to-green diverging charts (the collapse runs right through the neutral midpoint);
• blue against purple (purple is blue plus red, and the red component fades);
• olive against orange, brown against green.

Under tritanopia, anything separated only along blue–yellow:

• blue against green (both drift toward teal);
• teal against gray;
• yellow against pink.

How do you rank the design strategies?

Three strategies, in order of how much they buy:

1. Separate by lightness first. A light color and a dark color stay light and dark under every type of CVD — and under full grayscale, the harsher version of the same stress. If two meanings differ in lightness, the difference survives no matter which cone is missing. This is the same grayscale-survival test the color-scales side of the hub leans on.
2. Add a redundant channel second. An icon, a text label, a shape, a position. Once meaning rides on two channels, losing the color channel no longer loses the meaning — the principle WCAG phrases as never relying on color alone.
3. Choose robust hues third. A blue-versus-orange axis is commonly recommended as the most CVD-durable hue pair, since it survives red–green collapse well. But hue selection scales the worst of the three: it can rescue two colors, not a six-series chart across all three types.

What does a status trio look like under simulation?

Take a conventional success/warning/error set and read it under each simulation. (Simulations are computed with the Machado matrices — a standard research model — and one caveat travels with them: matrices approximate typical dichromacy of each type, and individuals vary, so a simulation flags a risk rather than certifying how any one person sees it.)

Read the lightness column. Warning survives — it stays clearly the lightest of the three under deuteranopia, so it is still distinguishable. But success and error, which looked boldly different (green versus red), converge to a near-identical olive: their hues merge and their lightness lands within ≈0.03 of each other. For a color-blind user, a status dot in this palette carries almost no information.

The fix follows the ranking. Separate success and error by lightness — lift the green a step, sink the red a step — and even though the hues still converge, one dot is now clearly light and the other clearly dark. Add an icon (check versus cross) and the meaning survives even a black-and-white printout.

Open this trio under all three simulations in Scale Composer — the Vision selector re-renders every role through protanopia, deuteranopia and tritanopia, and you can watch success and error march toward the same olive while warning holds apart.

Where does the physiology come from?

If you want the full account — cone biology, the anomalous forms, the prevalence figures broken down by ancestry and sex — the color blindness overview on Wikipedia is a well-sourced starting point. For design work, the three cone classes and the one rule they imply — lightness survives, hue may not — carry almost all the practical weight.

How do you fix a palette that fails?

Locate the collapsing pair, then separate it. Load the status trio and pull success and error apart by lightness — set them a clear lightness step apart, switch Vision to deuteranopia, and watch the pair that merged a moment ago stay legible as one-light-one-dark. Do it once on your own palette and the rule stops being a checklist item and becomes a reflex: meaning goes on the channel that survives.