On color theory

This is a repost of an article I published on Tumblr in 2017.  As many things I publish there, it was written somewhat extemporaneously, but it stood out as something I wanted to eventually import here.

There are really two color theories, the scientific theory of color perception, and the aesthetic theory of choosing color palettes.

The former is quite interesting, containing some surprising facts: yellow is the brightest color, many shades of green can’t be produced by modern displays, white is defined arbitrarily.

The latter is a hodgepodge of various historical ideas and a collection of overgeneralized advice. When I’ve read about aesthetic color theory online my impression is that much of it is either already taught to children or else it is not very good. Here is my attempt to identify some non-bullshit principles of color theory.

1. Recognize prior meanings and associations. Making a color palette is an exercise in referencing prior meanings, and also generating new meanings for other people to reference (or possibly you want a low-key color scheme that is very difficult to reference). Contrary to certain advice on the internet, single colors do not have very strong meanings–they’re simply used in too many different contexts. The strongest meanings in my experience come from holidays, flags, and natural objects. All these meanings are of course culturally and contextually dependent.

2. Choose contrast level carefully. High-contrast colors emphasize boundaries, while low-contrast colors may make them difficult to see. For some reason, many people think that high-contrast high-saturation colors are ugly, although I think it depends a lot on context. I’d put that in the category of “prior associations”–people have some prejudices about what brightness levels are appropriate where.

3. Know at least a little about color perception. Most importantly, be aware that about 5% of people are red-green color-blind.

4. Some possibilities that people sometimes forget about: Black and white are colors too. Not all the colors need to be fully saturated. You can completely change the color palette simply by changing how much of each color there is.

5. For large numbers of colors (at least 3 or 4), choose color combinations with low information density. Two common techniques are analogous colors (which are nearby on the color wheel) or triads/tetrads (which are evenly spaced along the color wheel). The particular technique isn’t important, but you want the color scheme to be something people can process easily. For contrasting colors, it helps if most of the colors are “canonical” colors, the kind that can be described in one word. For instance, “lime green, dusty blue, mustard yellow, maraschino cherry red” is just too complex and weird. “Green, blue, yellow, red”, on the other hand, is the color scheme of Google. (Although to be fair, even if Google’s colors are weird we’ve had a long time to get used to them.)


  1. stroppy says

    Interesting post!

    My 2 cents: From a practical stand point, for the various purposes to which color is put, color can be very hard to get right and requires a fair amount of training to use effectively. It’s not always well served by flat out assertions.

    There is of course the physical EM band of radiation shooting into the biologic (eye, nerve, brain), and there are models of color for scientific purposes (CIE lab, Munsell…). And there are also numerous modern models which, for instance in Photoshop, a user may want to switch between for various manipulations and output (RGB, CMYK, CIE lab…).

    Where the rubber hits the road, in making thematic maps with complex keys for example, there are rules of thumb which are helpful. Depending on purpose, you might use discrete steps, or continuous ramps. Creating perceptually even steps or meaningfully blended changes requires a keen eye and understanding of subtle optical illusions caused by; size of area, juxtaposition, and the need to direct movement of the eye…not to mention how to tune variables of value, saturation and hue.

    There are cultural, emotional and psychological components to how people perceive color, but I have to say, most of what people assert about this is of little use, unless they’re working off a style sheet–which to me is just a conceptual jail.

    Giving colors names is also a slippery area. Different cultures categorize color differently. In the end naming colors tends to be part creative art form in its own right and part marketing exercise.

    Well, IMO anyway. I do tend to gas on…

  2. Jenora Feuer says

    I will admit that, when choosing clothes, I tend to go with a combo of one dark, one light; also one saturated, one… low-chroma, I guess. So black pants with bright coloured shirts, navy blue pants with white shirts, that sort of thing. Contrast along multiple axes. That’s easier to do when you’re only dealing with two things, of course.

    Even people who are colour-blind can see differences in luminance, so UI designs have to keep that in mind. And the standard red/green traffic lights actually deliberately have a green chosen with enough blue that it’s still visibly different from the red light to people who are red-green colour-blind.

    As for ‘giving colours names’, Randall Monroe actually did a survey on that a few years back:


    Some of the joke names people submitted to the survey are amusing in their own right.

Leave a Reply

Your email address will not be published. Required fields are marked *