reorganization rate

[Marc_L_Kurtzer1]

i.kurtzer (951110.1430)

To further accomodate readers adversive to previous posts this one
includes capitalization!
This post is not so much in disagreement on the discussions of
reorganization but a diversion (brain-fart, maybe) on some minor
points on reorganization.

···

-----------------------------------------------------------------
Does Reorganization preclude differential rates of reorganizing at
different levels in a posited hierarchy? It could be similar to
the non-symptotic hierarchy (NSH just for short)in having more rapid
adjustments at the lower levels either by its own merits or via the NSH.
The Reorganization system might act at different rhythms for different
levels by spurting adjustments at different rates (faster for the lower
than higher so the difference is intrinsic to the Reoganization system)
or equal spurts across all levels that by the nature of the speed of the
NSH result in more rapid transitions down the hierarchy. If it is the
later it adds to the complexity of the posited entity, but decreased
"solutions" as when the cat more quickly escapes (generally) on each
trial seems to reasonably follow. As for the thought-experiment, the
creature creature stumbles upon one higher level refernce that is
necessary for any of the lower levels to effectively get it out of the box.
So the creature is "next to the stick"--maybe step one in a program of
references--and if it not then all its wagging and scrathcing won't do a
durn thing. Now next to the stick it does other things like jump at the
shadow the stick is casting, or bite it, or..whatever. Maybe next time
its bites here and not there, but "be next to the stick and bite it"
are continuing (i.e. that the random changes are acting at a level of
greater and greater specificity).
The basic point is that some reference is varying less rapidly than
others in the process of learning and as the lower levels are being
"fine-tuned" faster the decreased reaction times are a simple
consequence of this. This could be either a property of the
Reorganization system or a side-effect of the NSH or maybe both.
I hope this was clear enough, my own thoughts are still cloudy.

Any comments?

i.

[_Martin_Taylor]

[Martin Taylor 951115 14:55]

i.kurtzer (951110.1430)

Does Reorganization preclude differential rates of reorganizing at
different levels in a posited hierarchy?

No. It is ordinarily assumed that reorganization occurs at different rates
in different levels. There are two (at least sources) of difference. One
is the assumption that control systems that usually control well will not
reorganize much. This means that the bottom levels of the hierarchy will
usually stay pretty well fixed, provided that the person doesn't suffer
some trauma to those parts of the hierarchy (such as amputation or stroke).

The second source of difference is in the actual control bandwidths of the
different levels of the hierarchy. High-level systems necessarily control
more slowly than the low-level systems on which they depend. This means
that they respond more slowly to disturbances (most of the effects of
the buffeting delivered by the outer world are countered by lower-level
control systems) and to the effects of their own ouputs on their own
perceptions. Reorganization can't usefully happen faster than the effects
of control, because there is no way for the reorganizing system to determine
whether a change has been beneficial, and if good changes are removed as
easily as are bad changes, the net result is not a change for the better.

The net result of this is that higher-level systems should be reorganized
more slowly, but with higher probability than high-level systems. That
sounds like a contradiction, and it is, if "higher probability" means
higher probability per unit time. The way around that is to propose that
higher probability relates to the time-scale of the control system itself,
rather than to clock time. There's a benefit in this, because it removes
any need for external clocking in the hierarchy. All the effects are derived
from local conditions.

But the whole reorganization system is speculation, anyway. It just seems
to account for quite a few observations that have been made about learning,
and to be plausible in that it relies on nothing that the system can't
observe. It's worth keeping until it's either disproved by experiment, or
someone thinks of a more plausible and simpler approach to learning.

I hope this was clear enough, my own thoughts are still cloudy.

If we never walked through the fog, we'd never get anywhere, because the
fog isn't going to clear away by itself. Good on yer, mite:-)

Martin

[Peter_J_Burke]

I am constantly amazed about the real lack of attention to the question of
(re)organization in the PCT model. There seems to be more arm-waving and
hocus-pocus than real investigation of, or real proposals of, or real models
of, exactly how it works. Without this PCT cannot really claim to be viable.
I put this problem on the shelf next to the arm-waving and hocus-pocus that
reinforcement theorists use to talk about a "history of reinforcement" in
order to account for things that they cannot account for.
Peter

···

At 03:31 PM 11/15/95 EST, you wrote:

[Martin Taylor 951115 14:55]

But the whole reorganization system is speculation, anyway. It just seems
to account for quite a few observations that have been made about learning,
and to be plausible in that it relies on nothing that the system can't
observe. It's worth keeping until it's either disproved by experiment, or
someone thinks of a more plausible and simpler approach to learning.

------------------------------------------------------------------------
Peter J. Burke Phone: 509/335-3249
Sociology Fax: 509/335-6419
Washington State University
Pullman, WA 99164-4020 E-mail: burkep@unicorn.it.wsu.edu
------------------------------------------------------------------------

[Bill_Leach1]

<[Bill Leach 951117.21:02 U.S. Eastern Time Zone]

Message: 48249 on Fri, 17 Nov 1995 14:43:02 -0800
Author : "Peter J. Burke" <burkep@UNICORN.IT.WSU.EDU>

This is "gonna" be one of those "I should probably not send this for 24
hrs following my writing of it" type messages!

I am constantly amazed about the real lack of attention to the question
of (re)organization in the PCT model. There seems to be more arm-waving
and hocus-pocus than real investigation of, or real proposals of, or
real models of, exactly how it works. Without this PCT cannot really
claim to be viable. I put this problem on the shelf next to the
arm-waving and hocus-pocus that reinforcement theorists use to talk
about a "history of reinforcement" in order to account for things that
they cannot account for. Peter
------------------------------------------------------------------------

You are entitled to your opinion of course but examining my perception of
your opinion results in my NOT stating what I would like to say only
because this is a public forum and in general I don't believe such
language results in any communications taking place.

Should PCT have the opportunity to deal with one tiny fraction (say one
part in 1E6) of the research effort that goes into other so called
"behavioural science" then possibly PCT will have a foundation upon which
rational scientific research can be conducted in fields related to
learning and reorganization.

Since Frans has managed some how (probably at his own expense) to join us
here on the net maybe we can here first hand some genuine reorganization
type research results from two people that are interested in serious
research (Frans Plooij & Hedwig Van De Rijt-Plooij).

"Without this PCT ..."

At some point this is true but not in comparison with other "theories".
Where tested PCT WORKS... not some of the time but ALL of the time. This
is an important distinction between PCT and "others".

PCT is NOT answering the "big" questions that are being asked and where
relevent often can not do anything more than suggest an approach.
However again!, the major difference is that PCT does tell you how to
tell if you are "on track" or only "blowing smoke" as opposed to ...

The theory itself REQUIRES a random reorganization method of some sort.
Some of the requirements are based upon this requirement. Others are
only suggested by PCT based observation as appearent specifications.

I think it is a little much to "demand" that PCT answer questions about
an area of behavioural research where there are no two researcher that
will even agree as to the nature of the questions much less a possible
model (and that is even when their models are NOT required to actually
work)... especially when there are no more than about a dozen people that
have ever conducted serious PCT research efforts.

<flame off>

-bill

[Richard_Marken]

[From Rick Marken (951117.2130 PST)]

Peter J. Burke (951117) --

I am constantly amazed about the real lack of attention to the question of
(re)organization in the PCT model.

I think there are at least two reasons for this lack of attention. I don't
know if you will think they are _good_ reasons, but they are reasons
nevertheless: 1) it's necessary to understand the organization of behavior
before trying to understand its reorganization. You have to know, for
example, what variables people control when they walk before you can
try to understand how they learn to walk: 2) an awful lot of behavior is
very stable -- at least for adult humans; there doesn't seem to be a lot of
reorganization going on. It's easy to find examples of variables people are
controlling -- and they keep controlling those variables over long periods
of time. It's difficult to teach an old dog (like me) new tricks (new
variables to control) because we like the tricks we do; they work for us.

There seems to be more arm-waving and hocus-pocus than real
investigation of, or real proposals of, or real models of, exactly how it
works.

I don't know where you got that impression. Maybe you don't know that
both Bill Powers and Tom Bourbon have built working models of a
reorganizing system. The reorganizing system in these models is a meta-
control system that changes the functions and parameters in control
loops with the goal of achieving better overall control of variables in the
environment (actually, I've done a version of this myself with the
spreadsheet hierarchy). These models demonstrate the feasibility of the
"E. coli" approach to learning how to control.

There have also been a number of investigations of reorgnization in real
organisms. I have mentioned (and am happy to mention again) Dick
Robertson's demonstration of clear "bouts" of reorganization as people hit
on higher order solutions to a control game (Robertson and Glines, The Phantom
Plateau Returns, Perceptual and Motor Skills, 61, 55-64, 1985). I really
think it's worth it to read this paper; Robertson blazed a trail that few have
followed up; if you are really interested the study of reorganization, Dick's
article is the trail head.

Other investigations of real life reorganization can be found the work of
the new guy on the net (and his wife) -- Frans and Hedwig Plooij. The
Plooij's have studied developmental reorganizations that occur in the
early life of chimps and humans. They have made some fascinating
naturalistic observations of changes in the way variables (like "breast milk
consumption") are controlled (the variables under control are inferred
from _lack of_ effect of natural disturbances -- an observational version of
The Test). I highly recommend their paper in the "Control Theory" issue of
the American Behavioral Scientist (Developmental transitions as successive
reorganizations of a control hierarchy, v. 34, Number 1, Sept/Oct, 1990).
Reorganization is the name of the game when primates (chimps and humans)
are growing up.

There has not been lot of work done on reorganization -- yet; but what has
been done is a lot more than just "arm-waving and hocus-pocus"

Without this PCT cannot really claim to be viable.

I don't think PCT can claim to be viable (able to survive) while it is
just being done by a few old men (like yours truly). If people would
spend more time teaching their students how to model and study purposive
systems and less time arguing that "PCT couldn't possibly mean THAT
[insert rejection of your favorite assumption about human behavior
here]" then maybe PCT would become viable. Let's start teaching kids PCT
and leave the arguing to the old people who prefer those other perspectives
on behavior anyway.

Best

Rick

[Marc_L_Kurtzer1]

i.kurtzer (951118.1535)

>[Martin Taylor 951115 14:55]
>
>But the whole reorganization system is speculation, anyway. It just seems
>to account for quite a few observations that have been made about learning,
>and to be plausible in that it relies on nothing that the system can't
>observe. It's worth keeping until it's either disproved by experiment, or
>someone thinks of a more plausible and simpler approach to learning.
>
I am constantly amazed about the real lack of attention to the question of
(re)organization in the PCT model. There seems to be more arm-waving and
hocus-pocus than real investigation of, or real proposals of, or real models
of, exactly how it works. Without this PCT cannot really claim to be viable.
I put this problem on the shelf next to the arm-waving and hocus-pocus that
reinforcement theorists use to talk about a "history of reinforcement" in
order to account for things that they cannot account for.
Peter

You are absolutely correct that there has been much more speculation than
empirical demonstration of control in its relation to learning and
development. There has been some--Pavlovski, Hershberger, Robertsons,
Plooijs, Boubon and his grad. students--but this is only a beginning.
There have been several real investigations, real proposals, and real
models of how some "types" of learning might occur, but again there is a
serious dearth. This in itself does not question the viability of
PCT for 1) how could it be falsified if it has not yet been tested,
2) it is/would _not_ be a test of PCT per se, but rather a test of an
one possible extension of PCT. Lastly, I honestly urge you to join the
very small group that has made some real investigations, real proposals,
and real models into this subject and maybe you just might answer your
own question and a few of ours.

i.

···

Peter J. Burke (burkep@UNICORN.IT.WSU.EDU) wrote:

At 03:31 PM 11/15/95 EST, you wrote:

[Peter_J_Burke]

<[Bill Leach 951117.21:02 U.S. Eastern Time Zone]

You are entitled to your opinion of course but examining my perception of
your opinion results in my NOT stating what I would like to say only
because this is a public forum and in general I don't believe such
language results in any communications taking place.

Ooops. Sorry to have apparently hit a tender nerve. Clearly you share a
feeling that this is an important topic, and it seems that you share the
opinion that not a lot is being done right now (for what ever reason).
And, I am glad to hear about the few who are doing this research, and the
work they are doing.

Should PCT have the opportunity to deal with one tiny fraction (say one
part in 1E6) of the research effort that goes into other so called
"behavioural science" then possibly PCT will have a foundation upon which
rational scientific research can be conducted in fields related to
learning and reorganization.

I get discouraged and depressed reading all the posts arguing about how
the cat can get out of the box, etc. I see nothing in these discussions
that add to our understanding of PCT. We know how the model works once it
is set up (though what perceptions are controlled, and how is an open
question, but that is for empirical research, not for theoretical
development, and the contribution is minor from the perspective of the
latter more important (to me) question.

PCT is NOT answering the "big" questions that are being asked and where
relevent often can not do anything more than suggest an approach.
However again!, the major difference is that PCT does tell you how to
tell if you are "on track" or only "blowing smoke" as opposed to ...

There is no question (in my mind) that PCT has a much better approach
than most others to understanding behavior, but I still say work needs to
be done on (re)organization and plausalble models need to be built and
experimented with. PCT remains terrible incomplete without a strong
theory of (re)organization.

I think it is a little much to "demand" that PCT answer questions about
an area of behavioural research where there are no two researcher that
will even agree as to the nature of the questions much less a possible
model (and that is even when their models are NOT required to actually
work)... especially when there are no more than about a dozen people that
have ever conducted serious PCT research efforts.

Empirical research is good once you have some theoretical guides. I would
like to see the modelers generate some plausiable models that might serve
as theoretical guides to the process of empirical research.

Peter

···

On Fri, 17 Nov 1995, William R. Leach wrote:

[Peter_J_Burke]

[Peter Burke (951118.1443)}

I don't know where you got that impression. Maybe you don't know that
both Bill Powers and Tom Bourbon have built working models of a
reorganizing system. The reorganizing system in these models is a meta-
control system that changes the functions and parameters in control
loops with the goal of achieving better overall control of variables in the
environment (actually, I've done a version of this myself with the
spreadsheet hierarchy). These models demonstrate the feasibility of the
"E. coli" approach to learning how to control.

Thank you. No, I was not aware of these. If they are availabe on the net
let me know where. If there are not, perhaps Bill and Tom can post them
to me privately. This is exactly the stuff I would like to see and am
glad to learn about it.

There have also been a number of investigations of reorgnization in real
organisms. I have mentioned (and am happy to mention again) Dick
Robertson's demonstration of clear "bouts" of reorganization as people hit
on higher order solutions to a control game (Robertson and Glines, The Phantom
Plateau Returns, Perceptual and Motor Skills, 61, 55-64, 1985). I really
think it's worth it to read this paper; Robertson blazed a trail that few have
followed up; if you are really interested the study of reorganization, Dick's
article is the trail head.

More goodies. Thank you.

There has not been lot of work done on reorganization -- yet; but what has
been done is a lot more than just "arm-waving and hocus-pocus"

Again, I am very glad to hear this. If others out there have knowledge of
other work that exists, I would appreciate learning more, though I have
been given a good start. Thank you.

Peter

···

On Fri, 17 Nov 1995, Richard Marken wrote: