I suggested a probabalistic framework in which to analize it. Take the red-orange problem. The probability that a given wave of light is classified as red is a function of its length. Thus for wavelength a, the probability that it's red may be 1. Then as you move along the spectrum, that probability dereases and eventually reaches zero. Next, you move on to orange, then yellow, and so on.
You could estimate this function by showing different wavelengths of light to a large sample of people and asking whether it is red or orange.
What do you gentlemen think?
I think the "problem" begs the question as to how we apprehend the colors "red" and "orange" to begin with. In fact, the act of posing the question/problem/solution assumes that the terms "red" and "orange" will be recognized by most (all?) people. In fact, children have no difficulty recognizing red, orange, and "reddish-orangish."
Am I addressing your question?