Color and Reality
gmilburn.ca
gmilburn.ca
Thank you for confirming the ultimate weirdness of the universe for me.
Interesting idea but in the long run the truth is that it doesn't really matter.
The theory of physicalism states that an object's color is an intrinsic property of matter, that its color is determined by its atomic properties. This theory is largely discredited in modern philosophy.
The theory of relationalism states that an object's color is determined by the atomic properties of the object in combination with physical properties of the environment: ambient light, reflection/refraction, diffusion, etc.
The theory of projectivism states that an object's color is a combination of its physical properties and the perception in the viewer; namely, that color is an experience tied to a dual system of subject and perceiver. This is the generally accepted theory today.
There's a large body of work on the philosophy of color (most of it sitting in my girlfriend's bookshelf). See http://plato.stanford.edu/entries/color/ for an example of the depth of the topic.
You might think this isn't a color demonstration, but it is actually a compelling example that beyond just the spectrum of the photons hitting your eye, the brain is doing a lot of additional work involving what we call "colors". Further evidence: The broad similarities between what attributes cultures ascribe to colors. Yes, there are differences too, but there's a lot of correlations there, added by your brain, which comes preprogrammed with some meaning-ish things. (It's not that direct, but since I'm working in English here that sentence will have to do.) We know we share similar "red" and "green" experiences because we share their clashing effect. I'm not sure if there's anybody who has ever moved from not-red-green-colorblind to red-green-colorblind, but in a real sense, a red-green colorblind person is seeing a different color, with different properties and different aesthetics. Their aesthetics aren't wrong, just different.
(Related: It is frequently complained that aliens on TV are just humans with a bit of latex stuck to them, but there are numerous other ways in which aliens are just "humans with bits stuck on", and one of the most notable is how many alien civilizations basically look good on TV. I'm willing to buy that an alien race building a real engineering artifact will build something with a form that appeals to us, just as many aliens-on-earth do (spiders, for instance), but their colors ought to be alternatively garish beyond belief, subdued beyond our ability to recognize (10 million shades of blue and combinations of blue), or just plain random (a gamut twice ours, perhaps, or a race with no equivalent to vision at all, though I find that one hard to swallow), as they work with colors that may happen to share the same wavelengths of light (or not) but otherwise have nothing in common with human colors.)
In other interesting thoughts, when we can directly stimulate the human optic nerve, I look forward to the seeing what happens when we move out of the CIE color space. IIRC, the boundaries of the CIE color space represent the point where the three color receptors in the eye would need a negative photon to generate that result, which can't exist in real life. This implies to me that such values could be generated by direct stimulation, though. Are there colors that no human has ever seen, but nearly every human is equipped to experience in much the same way as any other color, in some sense as much a color as any of the ones we know? Perhaps some of us will even find out, some day.
Also, there is a difference between how one sees color and how well one is able to distinguish one color from another.
The CIE color model was based upon, if memory serves, testing done on 25 white french males, about the worst possible test group so far as color acuity is concerned. They should have chosen asian women, who have the highest color acuity as a population, that is, the ability to distinguish shades from one another.
What does this mean? Color is a perceptual quality, not a physical one, so the ability to distinguish colors is intimately tied to their definition.
Since most people have fairly similar cone spectral responses, basing the model on an Asian woman instead of a French man wouldn’t really change it much. If using CIE colorimetry didn’t work pretty well, we wouldn’t still be using it all these decades later.
And CIE colorimetry works, but I can tell you from experience, it does not work "very well". Rather it is adequate given the DeltaE tolerance of any given project. That is why the world is moving to spectrographic models, and there are already alternatives in the works which stand to replace CIE/LAB/XYZ, for precisely the reason that they have proved inadequate to the task of modern color management.