The Bible is Close Too

[From Rick Marken (920825.2100)]

I feel a loose canon coming on.

Forgive me, Bill, for I care not what I do.

penni sibun (quoting Agre I think) says:

faced w/ an empirical phenomenon to explain, our
first explanatory recourse should be to dynamics,
not to machinery.

and Powers replies:

I don't call that an explanatory recourse; it's just
describing the dynamics of the phenomenon
(usually from the standpoint of an unspoken
theory).

and Martin Taylor says

But it is precisely the dynamics of the control
hierarchy that describe the modelled behavior.

and I say:

Talk about generous interpretations. The Agre
quote was ruling out mechanisms (which I read as
"models") to explain observations. In that context,
dynamics means, to me, the time course of
behavioral phenomena. Bill read it exactly that
way too, apparently -- dynamics refers to a
DESCRIPTION, not an explanation of a phenomenon.
If mechanisms are ruled out then its bye-bye
control hierarchy (an invented explanatory
mechanism designed to explain the dynamics of
the phenomenon of control).

it seems to me that you and she both have the
same intention of simplifying mechanism to a
point at which it can be directly tested through
the dynamics.

My intention is to explain the phenomenon of
control. The first order of business is to develop
experimental situations where we can demonstrate
the phenomenon reliably and in detail (maybe it is
these details that you would call "dynamics"?).
Then we see if we can build a control model to
produce the results. Dynamics, per se, are not
fundemental to PCT; although the world is dynamic
(and, hence, so are the actions that are part of the
process of controlling perceptual consequences of
that world).

Martin says:

Penni wrote back to me asking why she kept
getting "jumped on" for making these kinds of
comment. I don't know why. I would have
expected compliments.

I think Bill Powers has been MORE than diplomatic
in his response to the quote by Agre. But the fact
of the matter is (and since nobody likes me
anyway I'll say it) that the Agre quote was a pretty
concentrated heap of prattle. No one means to
"jump on" anyone -- as I said, there seems to be a
good deal of tolerance on the net. But if, when
turning the words that are posted on the net into
our own imaginings, we come up with nonsense
(and that's the best I could do with the Agre quote)
then we (I anyway) say it. This doesn't mean that we
think Agre (or penni or anyone) is dumb. He is just
WRONG (at least from my perspective). But it may
be that I'm turning his words into all the wrong
imaginings -- so show me what you mean with a
model (mathematical or computer or mechanical);
then there might be less misunderstanding.

I'm willing to believe that the Agre quote is the
embodiment of the greatest wisdom of the ages.
But it sounded like complete drivel to me (and Bill
tried to politely explain why it did so to him as
well). It would really help if penni (or anyone else
who has an alternative to PCT) could describe ways
we could demonstrate the meanings of the words
to ourselves. If you want to know how to
demonstrate the phenomenon (and model) of
control to yourself, I suggest ordering my Mind
Readings book for suggestions.

Martin says:

There is some kind of a fog here. Could it be the
smoke from our respective burnouts?

The fog seemed to be entirely in the Agre quote.

Avery Andrews (920825) says:

This discussion is gotten completely out of hand,
and, like Martin, I'm rather baffled by it.

I respectfully submit that there is no way that
prose of the sort attributed to Agre could lead
anywhere but to an "out of hand" discussion.

I suspect that people haven't done enough
homework on the other guys' stuff to justify the
things they're saying about it.

And I suspect that this is a load of crap! I get this
ALL THE TIME when I try to publish experimental
tests of other people's theories. I have read more
articles on other peoples theories and models than
I even want to admit. I've desperately tried to find
testable working versions of those models so that
people could not say what you are saying. But no
matter what I do, if I create a situation that
exposes the failures of the model, I am quickly told
that I don't understand the model or that the
model doesn't really work that way. Either I'm
really stupid (a serious possibility -- just ask my
kids) or there is a very cute shell game going on
out there in ScienceLand that I hadn't intended to
play.

I rather suspect that `interactionists' have an
inadequate appreciation of how feedback control
systems actually work.

Actually, what they have is no idea what feedback
control system DO. They know zip about control so
they obviously have no interest in figuring out how
control systems work. Many of these people (ai types mainly)
have no apparent interest in science, for that matter.
They just like to play with models (or philosophical
abstractions) and have no idea that there is a REAL
phenomenon out there (called control) BEGGING for
an explanation.

but I also think that PCT is pretty shaky in the
area of perception, where Chapman in
particular has done useful work.

What in the world is "shaky" about the PCT view of
perception??? What perceptual processing has
Chapman put in his models that is so "useful". The
Agre and Chapman article that I read reflected one
of the most naive and unsophisticated views of
perception I've ever seen coming out of grown-ups.
Have you read Behavior: The control of
perception?? Have you ever seen a behavioral
model with a more sophisticated, deeper,
functional understanding of perception (maybe you
mean the brilliant perceptual model of stimulus
sampling theory? dynamical systems theory?
what???).

But they really are very similar.

And the biblical description of genesis is "very
similar" to the cosmological/evolutionary
description of genesis.

Doesn't anyone read my rants? Remember. All
these people with theories that are "very similar"
to PCT forget that little part about how organisms
CONTROL and that what they control is
PERCEPTION. Did anyone read my "Blind men"
paper. Did anyone read Bill's Psych Review article.
Should I write it in RED. How about repeating it:

Behavior is controlled perception;
Behavior is controlled perception;
Behavior is controlled perception;
Behavior is controlled perception

That doesn't make it true, but I hope it does make
it NOTICEABLE. Now ask all these people with
similar theories if the above collection of phrases
strikes them as being a good description of their
model of behavior.

Part of the problem, it seems to me, is that people
are too eager to be believed rather than
understood.

I don't want to be believed. I want to be
understood. That's why we develop all these demos
and experiments and models. No need for belief;
just LOOK and TEST.

It should be fine if Penni or anyone
else is skeptical about the truth of PCT

Absolutely. And she is free to maintain her
skepticism. I'm not objecting to her skepticism.
I'm only saying that the stuff quoted in Agre is
drivel. But I am willing to be convinced that I am
wrong. Just show me how Agre's theory explains
the one phenomenon that I know exists (because I
can demonstrate it) and that seems to characterize
everything that we call human behavior --
CONTROL.

I wish I had time to spend the morning trying to
hose this down, but I don't, so I'll just end with a
relevant quotation:

>One can seldom be sure whether the silence of

other scholars indicates agreement or
disagreement, indifference or incomprehension.

  > Wilbur Maas, _Greek Metrics_.

I'll give you odds that its the last three.

There. Now Bill can play GOOD COP.

Regards

The loose canon

(Yes, I know how to spell cannon)

ยทยทยท

**************************************************************

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)

[from Avery Andrews 920826.1615]

  Rick Marken (920825.2100)

The Agre
quote was ruling out mechanisms (which I read as
"models") to explain observations.

No, it wasn't. But it probably has to be read in context to be understood
properly. It was advocating that mechanism by reduced to a minimum
(but a *sufficient* minimum, which puts C&A in a different basket than,
say, Bickhard, who seems to have no sense at all of what `sufficient
mechanism' might look like). He wants people to get prizes for eliminating
unnecessary mechanism.

>My intention is to explain the phenomenon of
>control.

Fine, but it is also necessary to get people to suspect that control
is central to ordinary human activity, and that they don't already
understand it properly. This, I think, requires having a large
library of examples that will `grab' people, & picking the ones that
will appeal to people from particular backgrounds (given the fragmentation
of the human `sciences' there won't be any `one size fits all' solution.
Somehow, for some reason, that fact that you need feedback to point
at things or stand upright doesn't seem to get to AI-ers, I think because
they don't seem `cognitive' enough. The right examples, I think, have
a capacity to burn holes in people's minds that mere arguments don't
(academics are trained to win arguments, and people from Harvard and
MIT base their egos on the the idea that they're (almost, if they're
humble) the smartest person they know). I'm hoping that the steering
example will prove effective, but there must be lots more.

>Actually, what they have is no idea what feedback
>control system DO. They know zip about control so
>they obviously have no interest in figuring out how
>control systems work.

I *think* that this is true. So what's needed is to get them interested.

>any of these people (ai types mainly)
>have no apparent interest in science, for that matter.

Definitely false of the Interactive AI crowd, from what I've seen.

>What in the world is "shaky" about the PCT view of
>perception??? What perceptual processing has
>Chapman put in his models that is so "useful".

Well, could you build a system that could look at a video screen &
process what it sees there well enough to play the game? Chapman can't
really do that either, but Sonja embodies a lot of work on the problem.
Interestingly, in his thesis proposal, he said that Sonja' perceptual
system would be like Pengi's, but her central processor would be much
more complicated. Interestingly, the reverse proved to be true:
`deciding what to do is easy when you know what's in front of you'.
The main reason I didn't become a PCT-er at age 20 or so was an inabilty
to get anywhere on understanding how high-level perceptual functions
work - e.g. not a clue about how language is actually processed, for example.

Now, I think it's time to look again, partly because there has been
real progress on the perceptual front, also because you & Bill have
gotten me to begin to suspect that maybe the problem of unknown mechanisms
for perception is not as serious as I thought it was (if you can
demonstrate that the critter is controlling for X, then you know it
contains an X-detector, even if you don't know how to build one. And
you know that you ought to look for one).

Avery.Andrews@anu.edu.au

[Martin Taylor 920826 11:00]
(Rick Marken 920825.2100 canon fire)

Dynamics, per se, are not
fundemental to PCT; although the world is dynamic
(and, hence, so are the actions that are part of the
process of controlling perceptual consequences of
that world).

In this quote is the heart of the resistance to using information theory
and the results of psychophysical studies. Rick is controlling for the
perception that dynamics is irrelevant to the operation of a control system,
whereas I believe that the control system is defined by its dynamic behaviour.
Not everything can be relegated to the result at infinite time, as the
stability equations often trotted out assume. It matters that the control
system respond fast enough to changing events in the world (disturbances).
It matters (sometimes) whether there are overshoots in the control function.

Why is there a conceptual problem with understanding the importance of the
differing numbers of degrees of freedom in different parts of the hierarchy?
Because they depend on the dynamics, and dynamics is predetermined to be
necessarily unimportant. A static analysis is simply not good enough.

I don't know why Rick, in particular, is controlling for this reference,
because it contradicts his own presentation (I've forgotten the title of
the paper for the moment) pointing out the congruity between the rates at
which certain actions can be performed and the rates at which the relevant
perceptions can be performed. If that isn't a consequence of dynamics, what
is? Why are the so-called "slowing factors" included as a naive method of
ensuring stability in the hierarchy simulations if not for reasons of dynamics?

Dynamics is central to the importance of PCT as a psychological theory. As
a static theory it only presents a trivial truth that has some amusing
consequences. You'll never persuade the world of its beauty by repetitive
ranting based on that trivial truth.

Forgive me, Bill, for I care not what I do.

Your resistance to disturbances leads the outside observer to doubt the
truth of this claim.

Martin