[From Bruce Abbott (941222.1750 EST)]
I'm told by one source that when our VAX computer is unplugged tomorrow, we
will lose internet services and will not get them back for two weeks. Another
source says that internet service will continue. So this may (or may not) be
my last post to csg-l for a while. Whichever is the case, I wish everyone a
happy holiday and may the new year be your best yet!
Some last-minute comments:
Tom Bourbon [941222.0919]
Bruce Abbott (941221.1030 EST)
In reorganization, do not the
consequences of a set of system parameters (control or failure to control)
determine whether those parameters will be preserved in the system or
abandoned in favor of yet other variations?
Rick...has already replied to this, Bruce. I
agree with his reminder to you that the *consequences* do not "select" or
determine anything.
Note that in the above I stated what those "consequences" are: control or
failure to control. Control means that the relevant perceptions are brought
near their reference levels and held there against non-overwhelming
disturbances. Continued failure to control intrinsic variables is, as I have
been given to understand, what keeps the reorganization process going.
Establishing successful control is what brings it to a halt, and thus keeps
the current (effective) configuration of control systems present. It is in
fact the "blind variation and selective retention" of which you speak. Thus
when you write
Other than the fact
that continued survival depends on the organism "selecting" a satisfactory
configuration for the system parameter(s), there is no "selection" going on
here
you are saying essentially that the only selecting going on here is the very
selection to which I refer. What an odd way to say that you agree with me!
Martin Taylor 941222 11:10
I have the impression that Bruce understands very well indeed how control
system operate, and that he has wanted all along to deal with learning. I
have not understood why Rick insists that Bruce isn't talking about what
Bruce tries to talk about. Now Rick is saying in addition that Bruce
shouldn't talk about learning at all (and therefore he isn't?). It's
been a weird conversation to observe, and I'm sure that it must have been
as frustrating to Bruce as some of my conversations with Rick have been
to me, to be told that he isn't saying what he says.
Thank you, Martin. I was beginning to wonder if anyone reading these posts
had understood my point of view. After a while, you get to feel a bit like
Alice in Wonderland during one of her conversations with the Cheshire Cat or
the Red Queen.
There's really very little point in a discussion that consists largely
of one participant repeating "I want to talk about A" and the other
repeating "No. I'm going to pretend you are talking about B, and I'll
prove you wrong." It wastes net bandwidth and elucidates no issues.
I agree. There are plenty of interesting and important issues to discuss,
among which are the means by which organisms learn to control their
perceptions. Dennis Delprato's query about whether a separate reorganizing
system is really necessary, and Hans Blom's observations about adaptive
control systems both suggest interesting avenues worth exploring. Although an
important function of CSG-L is and will remain educational (helping newcomers
to understand the basic principles of PCT), I believe that it should also
serve as a forum in which to offer and debate alternative conceptions, develop
new models, and provide sharp comparisons between PCT and alternative views.
This is hardly possible if some of the participants are hell-bent on trying to
interpret one's every word as evidence of a lack of understanding of PCT
principles. C'mon guys, closed loop ain't easy, but it's not rocket science.
Perhaps our disagreements arise from another source.
By the way, I originally wrote ECOLI7A merely to demo the use of Turbo Pascal
objects, and to provide what I thought would be an entertaining example of
variation and natural selection at work. (The model does not include another
aspect of evolution, "like begets like," but could easily be adapted to do so
by having e. coli that live long enough reproduce themselves, usually with
only minor variations. They might even "mate" before doing so.) The model is
fun to watch and does show how the environment constrains the control
parameters of successful bugs. For example, none of the successful ones have
a negative gain parameter (they are repelled by the nutrient source when
hungry instead of being attracted to it), and few have a speed over 2.0. (Bugs
that move too fast use fuel faster than they can absorb it, even when staying
close to the nutrient source.) But when Rick went into orbit over my
suggestion that the model's ability to "find" bugs with good control parameter
represents an analog to PCT's reorganization process, I couldn't resist going
on the counter-offensive. I believe I have raised an important point about
the reorganization system in the process, which got overlooked in the rush to
find "mistakes" in my understanding of PCT. Anyone remember what that point
was?
So long for now,
Bruce