[From Rick Marken (920601 12:50)]
Oded Maler (920601) says:
Rick, you are so predictable.
Of course. Should I try to be unpredictable? This strikes me as a
compliment rather than a complaint -- though you seem to be saying
it as though it's a problem. Was there something wrong with Galileo's
predictable response to philosophers who argued that the world must
be stationary. Is there something wrong with my daughter's predictable
response to my claim that 3 times 3 is 10? Yep, I'm happily predictable.
I'm sure that at some level in your
hierarchy A&C and Beer evoke the same percepts.
Yes. I perceive thesed models as OPEN LOOP. If I'm wrong, that's fine
with me. So far (after 20 years of dealing with these kinds of models)
I have not been wrong yet -- such models are either designed to be
as open loop as possible or they deny the significance (or existence) of
those aspects of themselves that actually are closed loop. So far, I have
found only PCT to be based on the observation that organisms control and
that what they control is their own perceptual experience.
Anyway your feeling
is incorrect and their work is not using the buzzwords of dynamical
systems.
Just a guess based on minimal data. It sure sounded like they were relying
on physical constraints to solve the "planning" problem. If not, fine.
And if the model is closed-loop control of perceptual input variables, so
much the better.
Since in the sequence/program level your favorite theory still does not
have much to offer beyond qualitative hand-waiving, maybe don't be so quick in
dismissing others' attepmts to approach such problems.
I didn't mean to seem dismissive, though when it comes to being dismissive
I think PCT has more often been the dismissee rather than the dismisser.
I've tried to understand and build models based on a lot of behavior theories
besides control theory. I have yet to run across one that actually tries to
model control. Thus, I have yet to find one that is architecturally equivalent
to PCT (ie. it is build around the control of perceptual variables).
I don't know that PCT types have been hand-waving when it comes to control
of higher level variables -- we admit that we don't know how to do it
yet. (In order to do it we have to learn how to build systems that can
perceive the degree to which a sequence or program is taking place. Right now
I think the idea is to walk -- literally -- before we can do calculus problems.
I think the "little man" model -- which produces realistic behavior in a
realistic environment -- is far more impressive and contributes more to our
understanding of the organization of behavior than models which produce
complex sequences and programs that only barely resemble what people do -- and
in unrealistic environments to boot. I imagine that is a predictable
preference). I do assume that control of sequences and programs, etc is
based on the same principle as control of other variables -- it is control
of perception. If this is true, then the approaches that you say I should
not dismiss must be wrong. The only way these approaches could be right
is if the sequences and programs are not controlled variables. For the sake
of all those people working real hard on open loop models of sequential
and programmatic behavior, I hope they aren't.
Predictably best regards
Rick
···
**************************************************************
Richard S. Marken USMail: 10459 Holman Ave
The Aerospace Corporation Los Angeles, CA 90024
E-mail: marken@aero.org
(310) 336-6214 (day)
(310) 474-0313 (evening)