Color Game - Test Your Color Memory Online

Each round of the color game deals three colors, one at a time. A color fills the card for a few seconds with a countdown, then it is gone, and you rebuild it with three sliders: hue, saturation and brightness. Lock it in and the original appears beside your color, scored from 0 to 10.

There is nothing to name and no color theory to know. The game measures one thing: how much of a color survives the seconds between seeing it and setting the sliders. That is less than it feels like while the color is on screen, and the reasons are some of the best studied in the psychology of memory.

To test colors you have carried for years instead of seconds, the Logo Game recolors one part of a real company logo and asks you to put the brand's color back.

Color reached English in the Middle Ages from the Anglo-French culur and the Old French color, from the Latin color, which Old Latin wrote colos and which first meant a covering. It shares a root with celare, to hide: a color was the coat on the outside of a thing, the part the eye meets, long before anyone tied it to light.

Colour was the usual English spelling from the 1300s, and scholars tidying words back toward their Latin made color an alternative a century later. Noah Webster did not invent the American form, he settled it. His Compendious Dictionary of 1806 dropped the u from honor and its kind, a choice he called common practice in this country, and his American Dictionary of 1828 fixed color for the United States while British usage kept colour. The game is the same under either spelling, and the International Commission on Illumination, whose formula scores it, writes colour.

In 2004 Wilken and Ma flashed a few colored squares for a tenth of a second, blanked the screen, and asked people to click the color of one of them on a wheel of 252 hues. Answering on a continuous scale instead of saying same or different turned a memory into a number: how many degrees round the wheel the answer landed from the truth. The method is now called delayed estimation, and it is one of the standard tools for measuring visual memory.

Weiwei Zhang and Steven Luck used it in 2008, with 180 colors laid in a circle in CIELAB, the color space this game scores in, all equally light. Their answer, in Nature, was that people hold a high-resolution copy of two or three colors and keep nothing at all of the rest: a color is not remembered blurrily, it is remembered or it is gone.

A year later they stretched the wait. With three colors to hold, people lost 26 percent of them after one second and the same after four, but 39 percent after ten. The colors they did keep were no less precise for the wait. Zhang and Luck called it sudden death: a remembered color holds, then drops out whole.

This game asks for one color at a time, well inside that limit, so a low score here is rarely a color forgotten outright. It is a color that drifted, and the drift has a direction.

On 7 February 2015 Cecilia Bleasdale photographed a dress in a shop in England, one she was thinking of wearing to her daughter's wedding on the Scottish island of Colonsay. The family could not agree whether it was blue and black or white and gold. Caitlin McNeill, a guest at the wedding, posted the photo on Tumblr on 26 February, BuzzFeed picked it up that evening, and its post drew 28 million views in a day. The dress, from the British retailer Roman Originals, was royal blue with black lace.

Rosa Lafer-Sousa, Katherine Hermann and Bevil Conway asked 1,401 people that year: 57 percent saw blue and black, 30 percent white and gold, 11 percent blue and brown. The photo is washed out and gives no clear clue to the light it was taken in, so each viewer's brain supplies one. Assume cool daylight or shadow and the brain subtracts blue, leaving white and gold; assume warm indoor light and the blue stays. Pascal Wallisch surveyed more than 13,000 people online and found four in five of those who thought the dress was in shadow saw white and gold, against about half of the rest. In his sample the majority flipped, to white and gold.

Nobody in the argument was misremembering anything: the photo was in front of them. The game draws each color on a plain card with no scene around it to guess the light from, so what stands between you and the original is memory alone.

Michael Pointer and Grenville Attridge counted in 1998 by dividing CIELAB color space into cubes one just-visible step across and keeping the cubes that real surfaces can reach: 2,279,381 of them. The rounder figure you often see, ten million, comes from a 1975 reference book, and as Pointer and Attridge pointed out, it came with no working at all.

Words cover far less. Brent Berlin and Paul Kay compared color words in 98 languages in 1969, and their book is the source of the idea of basic color terms. English has eleven: black, white, red, green, yellow, blue, brown, orange, pink, purple and gray. A color you can name keeps a label in memory, and the label is a tiny part of what your eyes could separate. Hard mode leans on that gap: it deals the muted, dark and in-between colors that have no word of their own.

Nor does everyone see the same millions. About 1 in 12 men and 1 in 200 women of European descent have some form of color blindness, nearly all of it red-green. At the other end, Gabriele Jordan and colleagues tested 24 women who carry a gene for a shifted green cone in 2010. One of them, known in the paper as cDa29, told apart mixtures of light that look identical to everyone else, on every test they gave her: the clearest evidence yet of a person with four channels of color vision instead of three.

Splitting a color into what it is, how strong it is and how light it is goes back to the painter Albert Munsell, whose A Color Notation of 1905 called the three hue, chroma and value: what tells a red from a yellow, a strong color from a weak one, a light color from a dark one. The sliders are its screen-age cousin, HSB, also called HSV. Alvy Ray Smith wrote the conversion in 1974 for SuperPaint, an early paint program at Xerox PARC, so artists could pick colors by the qualities they see instead of by amounts of red, green and blue light, and published it in 1978.

The three interact. At zero saturation every hue is the same gray, and at zero brightness every color is black, so a hue error costs little on a pale or dark color and a lot on a vivid one.

The sliders start at half saturation and half brightness, on a hue at least 60 degrees from the color's, so the starting point gives nothing away. The three colors in a round are at least 50 degrees apart in hue, so no round repeats a color.

Slider numbers would make an unfair score. Hue 359 and hue 1 are the same red, and at zero brightness every hue is black. So the game scores what you would see. It converts both colors to CIELAB, the CIE's color space of 1976, built so that equal distances stand for similar visible differences, and measures the gap with CIEDE2000, the CIE's color-difference formula of 2001, written ΔE00.

A ΔE00 of 0 is an exact match. Dentists, who match tooth shades for a living, have measured where visible begins. Rade Paravina and colleagues showed pairs of ceramic samples to 175 observers at seven sites in 2015: half could see a difference of 0.8, and half would accept a difference of 1.8 as a match. A difference of 10 is plain side by side; 30 is another color.

Points follow a smooth curve, 10 × exp(−(ΔE00 ÷ 30)^1.4). With the power above 1, the curve is gentle on a near miss and steep further out, so the right color a shade off still scores well and a different color scores little.

The scoring is the same in every mode. The CIEDE2000 formula was built for surface colors under set lighting, so on a screen it is a well-tested ruler rather than a laboratory measurement. The game's version is checked against the 34 test pairs Gaurav Sharma, Wencheng Wu and Edul Dalal published in 2005. Your best round and average stay on your device for each mode, and a shared round lists the scores without the colors, since every round deals new ones.