Rob Harris <firstname.lastname@example.org> responded with:
> Why would the system need to "perceive" the color to stop at the
> light? A light sensor WOULD do fine.
I think this is precicely right! Color detecting machines are far more aware, intelligent and functional than we are about color. Not only can such machines resolve much better but obviosly they can be set up to produce complex behaviors based on colors we can't even see (let alone remember) the difference between.
The kicker difference is: when you ask a human what "red" is like, he honestly tries to express the very real and phenomenal sensation. Any abstract machine that represented color with mere abstract representations rather than a real phenomenal qualia, though it could be programmed such that it seemed to respond much more "humanly" than a human could, its descriptions of what red was like would be a blatent lie since abstract representations of knowledge and awareness information aren't phenomenal or conscious at all.
If a machine really had been given the proper machinery we have to produce, say, a salty qualia, it could then honestly respond with something like: "Oh THAT'S what salt tastes like!". But if it was representing the sodium chloride with abstruct numbers or something, though it could easily act even more convincing, it would still be blatently lieing about that fact that it could feel walt a salt quale was like. And if we really knew what "salty" was and precicely what machinery really produced it, and could examine the machine's representations objectively looking for such. Only then could we know if it was lying or not.