Metaphorical theory, perceptual control

CSGnet Archive CSG1996
1

[From Rick Marken (961016.0900)]

Bill Powers (961016.0500 MDT) to Martin Taylor (961015 11:30) --

Thanks Bill. That was a beautiful post.

After reading Martin's (961015 11:30) reply to me, I realized that Martin's
approach to developing a theory of "social control" is very simliar to
Kepler's early approach to developing a theory of planetary orbits. Kepler
was a mystic who was convinced that the planetary orbits had to be
constrained in some way by the mathematics of the "perfect solids". I was
reminded of the similarity of Martin's and Kepler's approaches to theory when
Martin (961015 11:30) said:

Attractor theory, if you want to call it that (I prefer "dynamics") argues
that these three are the only possibilities for interacting systems.

Suddenly I had this image of "Martin" Kepler saying "Perfect solid theory (I
prefer "gnosis") argues that these twelve orbital ratios are the only
possibilities for planets moving on perfect spheres".

Bill Benzon (961016) --

Frankly Rick, there are times when H/PCT seems like a series of verbal
formulae to translate things into a certain verbal style.

I find that when PCT starts to seem like that, it's time to download the PCT
programs and demos and see how the model actually works. Did you take a look
at my spreadsheet hierarchy model? I know that it's a little abstract but I
think that, when you know what to look at, it can be a very helpful way to
visualize the behavior of a hierarchy of control systems: from inside and out.

So tell me about the level or levels where references are set for search,
pursuit, and consume and the relationship between this level or levels and
the intrinsic variables.

The references are for perceptions. "Search", "pursue" and "consume" are
names for what we see when organsms control many different perceptions.
Determining what perceptions are actually controlled, and in what
hierarchical manner, is a job for research. But we can make some reasonable
guesses based on observation of the perceptual variables we control when we
are "searching", "pursuing" and "consuming". That's what I did when I
described the variables controlled by the hunting lion; I tried to look at
the world from the lion's perspective.

Is there some kind of weighting function to determine which of some 10s or
100s of intrinsic variables will be serviced now?

We don't know how control of perceptions relates to control of intrinsic
variables. We have a theory, but very little data. And there are very few
people collecting the necessary data. So I can only speak theoretically. In
theory, no, there is no weighting function; all intrinsic variables are
controlled at the same time -- all the time.

And what controls the relationship between search, pursuit, and consume?

I think you should take a look at my spreadsheet model. Or Bill's Little Man.
What you will see is that you can categorize the observable behavior of a
hierarchy of control system in various ways at various times; you can see
behavior as "searching", then "pursuing" then "consuming". But this is just
the observer's way of categorizing the results of the actions of a hierarchy
of control systems that are varying references (and, hence, perceptions) at
all levels in order to maintain control of all variables in the context of a
world of continuously (and unpredictably) varying disturbances.

Reminds me of that old joke about the new guy in prison arriving and hearing
the inmates calling out numbers, with laughter following each number. You
could save yourself some key strokes if you assigned numbers to the various
phrases, with "X clearly didn't know what behavior is . . . .) being a prime
target.

I think this is a great idea. We do repeat the same things over and over
again on CSGNet, usually saying them in somewhat different ways, in an effort
to get people to understand some of the basic facts about how control systems
work. Of course, we usually get nowhere because most of us (because we ARE
control systems) are far more interested defending our existing ideas then
learning new ones. I think it would be great to be able to just type numbers
as my reply to the many misconceptions about control and control systems that
are repeatedly posted here on CSGNet. I agree that "X clearly didn't know
what behavior is" would probably be number one on my list, followed by:

2. Behavior is control.

3. Control systems control perceptual input.

4. Disturbance resistance gives the illusion that behavior is caused by
external events.

5. Changing references give the illusion of spontaneous change in behavior.

6. (corollary to 4 and 5) The results of nearly all conventional behavioral
science research is misleading and of no use to PCT.

7. The Test for Controlled Variables must be the first step in the study
of the behavior of living control systems.

There are many more PCT "standards". I suppose they all sound like slogans;
but I hear them as concise descriptions of what I know (from formal and
informal research) about human controlling and what I know (from computer
modeling) about the operation of the hierarchical PCT model of human
controlling. Even if I were able to reply to posts "by the numbers" I would
feel compelled to try to explain how my answer relates to my experience of
what seems to me the "reality" of perceptual control.

Best

Rick

2

Rick Marken (961016.0900) sez:

I find that when PCT starts to seem like that, it's time to download the PCT
programs and demos and see how the model actually works. Did you take a look
at my spreadsheet hierarchy model? I know that it's a little abstract but I

Well, I don't have Excel. But to tell the truth, having Excel wouldn't
make much difference to me. I adopted PCT around 20 years ago for certain
limited purposes (and published my first "intro to PCT" stuff in 1976 in
the opening section of an article on "Cognitive Networks and Literary
Semantics"). I don't need to be convinced that the hierarchy is a real good
thing and so is reorganization; I think it is a permanent contribution to
the study of human behavior in the way that "F=ma" is a permanent
contribution to mechanics. Once it's recognized, and it will be, it's not
going to disappear. But..... for one thing, human behavior is a lot more
complex than mechanics.

Let me tell you a story so you have a better sense of where I'm coming
from. In 1974 I hooked up with David Hays, who was teaching in the
linguistics department at SUNY Buffalo. He was doing research on computer
models of language, with a particular interest in semantics. The approach
I learned from him was what is called a cognitive or semantic network (and
is a variety of what came to be called connectionism). So, we have people
doing things with language via computers. One of Dave's students did some
work on driving a car; another did text analysis for stories exhibiting a
pattern of tragedy.

One of the epistemological problems you have doing this work is that one
can dream up any number of computer models that will perform more or less
to the current standard (which has never been very high). For some the
mere fact that you've got a computer program doing something constitutes
what they call an "existence proof." At least the model is explicit enough
to get a computer to do something. Hays and I were never much impressed
with that: existence of what? But, in general, how do you choose among
models? Of course, we know the best answer to that question: get empirical
evidence. But that is easier said than done.

And still, you'd like to put some constraints on your theoretical
inventiveness. Dave read BP's book and liked it a lot. For one thing, the
hierarchy provided some justification for what otherwise seemed like one of
many arbitrary facts about language, that verbs govern nouns. If you think
of verbs as being the linguistic coorelates of sequences and/or programs,
and nouns as corresponding to configurations, then you have a good reason
(or at any rate, as good a reason as PCT theory is a theory) for beginning
to see why verbs govern nouns; it has to do with the temporal
characteristics of the stack (from which they are derived). So, the stack
became a way of justifying certain semantic and syntactic relationhips.
Beyond that, reorganization sure seemed elegant.

Dave had everyone in his research group read B:TCP and various of us spent
many hours over 7 or so years working on getting HPCT & cognitive networks
into the same framework. Hays wrote a book, I wrote a dissertation, and
together we wrote the paper I've mentioned several times. That paper was
more or less the end of that phase of our work. When then went on to
cultural evolution.

So, Dave and I (& others) have logged many hours dealing w/ HPCT. If Dave
& I have ended up thinking that HPCT can't do all that one would want of a
model of human behavior, it is not a casual conclusion. Did we explore
every possiblity that HPCT presented before we added other stuff? Of
course not. We began with other stuff and we worked to keep it. We did
not come to HPCT w/ the proverbial tabula rasa. We were mostly interested
in what's going on in the upper levels of the stack. Our particular kind
of intellectual entertainments have more detail about that stuff than any
version of HPCT I've seen. That doesn't make our entertainments more
scientific, but it is a reason for me to prefer them and to urge you to
work harder on your own models.

Verbal reformulations and 3-level demos aren't going to convince me that
Hays and I have gathered a lot of unnecessary intellectual machinery. We
didn't gather that machinery together because we want to make our
intellectual lives as complex as possible. We did it because we thought it
was necessary. In the end, it was a mass of neural evidence of all kinds
that forced a lot of that stuff on us. And the most sophisticated work we
did is but a promissory note on invesitgations as yet not even begun.

So, for better or worse, what you're telling me just doesn't sound
convincing. Now, if you can produce a great deal of hard empirical
evidence, then you can get my attention. But pleas to give HPCT a chance
won't cut it. I've been there, done that.

We don't know how control of perceptions relates to control of intrinsic
variables. We have a theory, but very little data. And there are very few
people collecting the necessary data. So I can only speak theoretically. In
theory, no, there is no weighting function; all intrinsic variables are
controlled at the same time -- all the time.

And you really think that you don't need some explicit mechanism to direct
motoric action in service of one or three of some 10s or perhaps 100s of
these variables? You think that these variables will automatically conjure
up a reference level at, say, the program level, and the dog will happily
trot off in pursuit of a bone or a fire hydrant or a male or female dog or
a warm cozy doghouse?

···

********************************************************
William L. Benzon 518.272.4733
161 2nd Street bbenzon@global2000.net
Troy, NY 12180 http://www.newsavanna.com/wlb/
USA
********************************************************
What color would you be if you didn't know what you was?
That's what color I am.
********************************************************

3

Bill Benzon 9961016.1500)

A very clear statement of where you are coming from. Could you
remind me why you are posting to CSGnet?

Bruce

4

Bill Benzon 9961016.1500)

A very clear statement of where you are coming from. Could you
remind me why you are posting to CSGnet?

"Way back when" I started posting I had made use of the reorganizing system
in recent work cultural evolution. So, I wanted to see if anyone here had
any reaction to that work. The answer seems to be no.

However, that post did lead Kent to point out his paper on collective
control and I will probably make use of that idea. Some of Martin's talk
of attractor basins may proove useful (can't tell at this point). And I
found BP's paper on the origins of life, which has some ideas I think I can
recast for cultural evolution (perhaps using Kent's notion). One reason
I've been pursuing the emotion/reorganization stuff is that perhaps it
would come around and bump into the cultural evolution stuff. So far, no
dice.

I've got another purpose as well. I'm an associate editor of the Journal
of Social and Evolutionary Systems (info at:
http://www.cinti.com/connect-ed/jses) and I'm looking for articles. At
least one member of your group, Gary Cziko, has published there. The scope
of the journal is quite wide and least some kinds of PCT articles would be
within our purview (and the PCT perspective would not count against the
article, even if it was on a topic where I think PCT inadequate). So, you
folks have a standing invitation to submit to JSES. Check out our website
to get an idea of what we're about and, if you have any questions, don't
hesitate to query me or the editor (Paul Levinson: plevinson@cinti.com).

Bill B

···

********************************************************
William L. Benzon 518.272.4733
161 2nd Street bbenzon@global2000.net
Troy, NY 12180 http://www.newsavanna.com/wlb/
USA
********************************************************
What color would you be if you didn't know what you was?
That's what color I am.
********************************************************

5

[Martin Taylor 961017 12:15]

Rick Marken (961016.0900)

It's really interesting how you manage to turn things around. But before
I get into that, I have to agree with you (for a change):

Bill Powers (961016.0500 MDT) to Martin Taylor (961015 11:30) --

Thanks Bill. That was a beautiful post.

Yes it was. Beautiful and poetic. But surely totally irrelevant.

Now:

After reading Martin's (961015 11:30) reply to me, I realized that Martin's
approach to developing a theory of "social control"...

This is where I am amazed at your ability to turn things around. I asked
_you_ whether, as it seemed from your posting, you believed in the
existence of "social control," and now you accuse _me_ of developing
a "theory of social control"!!!! Utterly amazing.

Three years ago, you claimed to understand my position on this, as well as
the mechanisms I think operate. My position hasn't changed, though I may
have a better understanding of how the mechanisms work, and the manner
in which social and linguistic conventions drift and shift over time. But
your understanding seems to have vanished into the sunset.

Social control, indeed :frowning:

Martin