Ask the internet how many colors a person can see and you will get one number back almost every time: ten million. It is a tidy, confident figure, and it is repeated so often that it has the ring of settled fact. It is not. Nobody has ever sat a human down and counted ten million distinguishable colors, and the people who study this for a living give answers that range from about one million to figures in the hundreds of millions depending entirely on what you mean by the question. The honest short answer is that the number is unknowable to a single digit, and the reasons it is unknowable are more interesting than any one figure.
We think about this a lot, because the whole premise of the Color Memory Game sits on the gap between the colors you can tell apart and the colors you can actually pin down from memory. Those are not the same number, and they are not even close.
Where the famous ten million comes from
The ten million figure traces back to mid-twentieth-century color science, where it was offered as a rough order-of-magnitude guess rather than a measurement. The logic was simple: estimate how many steps of hue, how many steps of saturation, and how many steps of brightness a typical observer can distinguish, then multiply the three together. Pick reasonable-sounding numbers for each axis and you land somewhere around ten million. Pick slightly different ones and you land at five million or twenty. The figure was never wrong so much as it was never precise. It was a back-of-the-envelope product of three educated guesses, and the envelope stuck.
The trouble with the multiply-three-axes method is that it assumes the steps are independent and evenly spaced, and they are neither. The number of distinguishable saturations depends on the hue. The number of distinguishable brightnesses depends on both. The color solid that holds all visible colors is a lumpy, irregular shape, not a tidy box, so multiplying its rough dimensions overcounts badly in some regions and undercounts in others.
The more careful number is lower than you would guess
When researchers stopped multiplying axes and instead counted the actual volume of distinguishable colors inside a proper perceptual space, the number came down. The most cited careful estimate comes from Pointer and Attridge, who in 1998 worked out how many colors fall inside the solid of real surface colors and arrived at roughly 2.28 million (Pointer & Attridge, 1998). That figure counts the colors of physical objects, the things light can bounce off and reach your eye as, which is a narrower set than the colors you can produce with pure colored light.
So the careful answer for real-world surfaces is closer to two million than ten. The gap matters because it shows how much the answer depends on the question. Are you asking about the colors of objects, the colors a screen can emit, or the colors a laser could in principle stimulate? Each of those is a different volume, and each gives a different count.
Why no single number can ever be right
There are three reasons the count refuses to settle, and they are worth naming because they are the same reasons two colors can look identical to one person and different to another.
- It depends on the threshold. Counting colors means counting how many you can tell apart, which requires a definition of just barely different. Set the threshold tight and the count balloons; set it loose and it shrinks. There is no neutral place to set it.
- It depends on the gamut. The colors a phone screen can show, the colors a printer can lay down, and the colors that exist in nature are three different sets. A count is only meaningful once you say which set you are counting.
- It depends on the observer. Color vision varies from person to person more than most people realize. Two people with entirely normal vision will draw the boundary between blue and green in slightly different places, so even the threshold itself moves depending on whose eyes you ask.
The cone math, and the people who break it
The reason the number lives in the millions at all comes down to three cones. Most people have three types of cone cell in the retina, each most sensitive to roughly short, medium, or long wavelengths of light. Every color you experience is your brain reading the relative firing of those three channels. If each channel could resolve about a hundred distinct levels, three channels together would in principle encode on the order of a hundred cubed, a million, distinguishable combinations. That is the rough order of magnitude the careful estimates confirm. The machinery has three dials, and a million-ish is what three dials of that resolution can produce.
Which raises an obvious question: what if you had four dials? A small number of people, almost all women, carry a fourth type of cone thanks to the genetics of how color vision is inherited. Having the extra cone is not enough on its own; the brain has to actually use the fourth channel as independent information. For years researchers struggled to find anyone who genuinely did, until Jordan and colleagues identified a woman whose color discrimination could only be explained by a working fourth channel (Jordan, Deeb, Bosten & Mollon, 2010). A true tetrachromat of this kind would, by the same multiply-the-dials logic, distinguish on the order of a hundred times more colors than the rest of us, which is where the eye-catching figure of around a hundred million comes from. Treat that number the way you should treat ten million: a theoretical ceiling from multiplying dials, not a headcount anyone has run.
The number that actually matters is much smaller
Here is the part we find genuinely useful, and it is where the whole question turns inside out. The colors you can see number in the millions. The colors you can name number in the dozens at most, and the basic color words almost every language settles on number around eleven. And the colors you can hold accurately in memory for even a few seconds number, realistically, around one. The funnel from seeing to naming to remembering is brutal, and most of the loss happens in the last step.
That is the whole reason the game is hard, and it is worth sitting with. You almost never fail a round because you could not see the target color. You could see it perfectly. You fail because the instant it disappears, your visual memory compresses those millions of distinguishable shades down to something far coarser, a rough gist of hue with the saturation and brightness already drifting. We wrote about exactly how that compression works in the science of color memory, but the headline is this: discrimination is cheap and memory is expensive. Your eyes are extravagant and your memory is a miser.
This is also why we do not score guesses by counting how many of the millions of possible colors sit between your answer and the target. Raw distance in a color space does not match how wrong a guess looks to a human, because the millions are not evenly spaced in perception. Two colors a small numeric step apart can look identical or clearly different depending on where in the space they sit. So every guess is judged with a perceptual difference formula built to match human ranking, which we explain in what is CIEDE2000. The count of colors you can see is a fun fact; the count of colors you can tell apart in the region that matters is what a fair score has to respect.
So, how many?
If you need one number to repeat at a dinner table, say a few million for the colors of real objects, with ten million as the loose popular figure and a couple of million as the careful one. But the more honest answer is that the question has no single number, because it has no single meaning. It depends on how different two colors have to be before they count as two, on which set of colors you are counting, and on whose eyes are doing the counting.
And the number that should actually impress you is not how many colors you can see. It is how few you can keep. Millions stream in every time you open your eyes, and almost none of them survive the walk from perception to memory. If you want to feel that gap directly rather than read about it, that is what every round of the game is for. Try a single color in the daily challenge, or string a few together in solo, and watch how fast something you saw with perfect clarity turns into a guess. The seeing was never the hard part. Training the part that remembers is the whole point, and a few of the drills that help are collected in train your eye for color.