Subject: Misguided Wow

[From Bruce Abbott (950507.1925 EST)]

Rick Marken (950507.1230)

By the way, Bill, I ran your PCT-based re-write of Hans' model-based
control program. It works like a charm -- the perceptual signal, xt, is
kept _precisely_ matching the sinusoidally varying reference signal,
xopt, despite aribitrarily varying disturbances to the perception. As
Bruce Abbott said "WOW".

I hope I am not detecting sarcasm in your quote. I think Hans' model should
be evaluated on its OWN merits, not on those YOU'VE imposed. The model was
designed to function within an environment in which the disturbance waveform
is fairly regular (i.e., a function that can be fairly well described by the
system's internal world model equation, given the right parameters), and to
continue to function for short periods (based on that model) when input to
the perceptual function is lost. I don't see your ordinary error-based
control system doing that.

Hans has provided us a teaching device--one designed to clearly illustrate
basic principles--not the ultimate adaptive controller. I applaud that
effort. So it DOESN'T resist arbitrary disturbances; I see no special
difficulty in applying these principles to ordinary perceptual control
systems that do, such as represented by Bill's modification. Such a system
might, for example, build and update its model while _operating_ on the
basis of the "real world" perceptual input, then switch to the model during
short periods when that input is lost. If those periods are not too long,
the internal model should provide reasonable estimates of the actual (but
unknown) values of the controlled variable, even if the disturbance is
fairly arbitrary (NOT random!). The internal model would allow the system
to run in imagination mode.

I agree with Bill that the strategy involving statistical testing is
unlikely to be taking place in the nervous system, but I doubt that Hans was
suggesting that it does (it's a model of a piece of engineered hardware, not
of the nervous system). Let's not let the details of implementation obscure
the principle the model was intended to illustrate. Perhaps something
valuable may be learned from examining, thinking about, and discussing Hans'
model and the demonstrations he provided.