I actually don't think we completely disagree here. First, I definitely didn't intend the parallels with math to extend any further than the problems of percep

I actually don't think we completely disagree here. First, I definitely didn't intend the parallels with math to extend any further than the problems of perception with regard to how they are taught versus how they are practiced by scholars. Music theory is an art-related discipline and very much culturally inflected. And this makes sense. Most music theory (in the West at least?) does focus on Western classical music, in large part because Western music has a literate tradition (see Taruskin) that most other musical traditions lack, which allows us to treat the compositions themselves as texts/objects more easily than other musics. Even within Western music, we tend to use style- (read culture-) specific analytical tools. It doesn't make sense to use harmonic analysis for Stravinsky, and it doesn't make sense to look for twelve-tone rows in Mozart. In general we use a combination of language that composers used and thought with (so for Mozart, think keys and counterpoint) and techniques that arose from observing the style after the fact (in this case, for instance, sonata form, or Schenkerian analysis). All of these techniques are culturally biased because, as you've pointed out several times, music is psychological, and therefore rooted in culture. Meanwhile, it makes sense to analyze, for instance, Indian classical music using rhythmic modes. So, coming around to your analogy with elementary school grammar education, I think you're right that there is a prescriptivist parallel here -- learning that parallel fifths are not allowed, or that the second theme in a sonata always modulates to the dominant, or even that there always is a second theme in sonata, all of these do not accurately describe all music, or even all Western music of the common practice period. They are artifacts of the particular techniques being used. The original description of the sonata came into being around the same time as prescriptivist grammar (and has had a great deal of improvement since then -- read Rosen if you haven't already). Species counterpoint and Bach-style chorale harmonizing came about to describe specific styles, and to be used as exercises for composition students in the 18th-19th Centuries. The fact that those two things in particular (which I think are the most grievous examples of what you're talking about) are still included in music theory education is possibly counterproductive, especially since they are treated as sets of rules, rather than as historical examples of the composition pedagogy that the Western classical composers themselves were subjected to. Teaching Fux as anything other than "this is how Mozart and Beethoven learned to write counterpoint" is silly. So that's the prescriptivist parts. However, where we disagree is that I think there's still a lot here that's useful for descriptive purposes. It's difficult to talk about classical music within its cultural-intellectual context, which I think is necessary, without these tools. You can argue that any interval can be dissonant with the right timbre, but if those timbres are rarely used in the repertoire you're describing, it's not much use is it? The 18th and 19th Century composers thought about dissonance a certain way, and if you're analyzing their work as art objects (cultural objects) rather than physical objects, you have to at least take that into account.