Help me understand color theory from a mechanistic perspective
July 19, 2013 9:28 AM Subscribe
How can we (as precisely as possible) characterize what colors go well together? What underlying physical/biological or mathematical explanations underpin the recommendations of artistic color theory? I am particularly interested in the culturally invariant component of these models, if any such components can be extracted.
posted by Jpfed to Science & Nature (6 answers total) 19 users marked this as a favorite
We have a pretty good idea of why certain sounds are harmonious/dissonant. We may have started out with "notes from this scale sound good together", but we eventually came to understand "notes are characterized by their fundamental frequencies; harmony between notes is greater for rational intervals q/k where q and k are small integers" AND THEN we came to understand that this is partially explained by the perceptual interference created between harmonics that lie within a critical band's width of one another. I could even write a computer program that could rank different sets of frequencies by how harmonious/dissonant it estimates those sets to be, and I would expect it to perform pretty well.
The physiological component of audio judgments is something that can be pretty mechanistically characterized (which you'd expect- much of the process is done by machines that happen to be made with carbon). The cultural component is not so easily characterized; I'm comfortable deciding that it is irrelevant for my current purposes.
I am looking for an analogous understanding of color judgments.
In particular, if I have a set of colors (e.g. as defined numerically in a color space of your choice) what sorts of numeric relationships might help me predict how harmonious/dissonant that set of colors might be? Why do sites like this
make the recommendations they do? What sorts of mechanisms (e.g. "your thalamus has to do this thing, and here's how dissonant colors don't work as well with that") might help explain the particular form of the numerical models?