[Avery Andrews 930318.1121]
(Rick Marken (930317.1400))
I don't think I'm bluffing about anything (tho have done a bit of
guessing, some of which was a bit wide of the mark, as Martin recently
indicated). I'm with you in wanting to know how knowing p(t) lets you
describe d(t) with fewer bits. I think my little story about noise
generators is a sufficient basis for saying that information about
d(t) is somehow present in p(t), but if you don't, fine. After all,
I do say that it is present in `some mysterious sense' which means I
don't claim to know what's going on, but only to suspect that something
interesting is.
On a completely different note, I've just written a little
implementation of a simple `coordinative structure' via perceptual
control, vaguely inspired by the Abbs & Winstein work on lip
movements. There are two little points, driven by thrusters through
a medium. And there is a reference level for separation of the
points, so that when they are too far apart the thrusters drive them
together, & vice versa. The reference level is produced by a square
wave generator. Finally either of the two points can be `frozen',
whereupon the other goes further to compensate.
In the grab-bag of possible paper topics, perceptual control as an
implementation theory of coordinate structures strikes me as having
some promise.
Avery.Andrews@anu.edu.au