However, this has to bottom out somewhere; and we call the places where it bottoms out - goals that're valued in and of themselves, not just because they help with some other goal - terminal values.

This seems to be a near-consensus here at LessWrong. But I'm not convinced that "it bottoms out in goals that're valued in and of themselves" follows from "this has to bottom out somewhere". I grant the premise but doubt the conclusion. I doubt that where-it-bottoms-out needs to be, specifically, goals -- it could be some combination of beliefs, habits, experiences, and/or emotions, instead.

But you say, we call the places where it bottoms out goals ... (emphasis added). Of course, you can do that, and it's even true that people will pretty well understand what you mean. You can call these things goals, and do so without doing terrible violence to the language, but I'm not convinced that this is the most felicitous way of speaking about motivation and ethical learning. Whether these bottom-level items are best described as goals, or habits, or beliefs, or something quite different, depends on psychological facts which may not yet be in (sufficient) evidence.