Still fighting the hydra

CSGnet Archive CSG1997
1

[From Bruce Abbott (971201.1550 EST)]

Rick Marken (971201.1000) --

Bruce Abbott (971201.1215 EST)

Sorry to disappoint you, Hydra, but I'm not so easily vanquished.

Boy, do I know that! I didn't really expect that you would be
vanquished. Don't think there is any way the Hydra can win this
fight.

Oh sure there is. You can just keep raising new spurious arguments until I
just give up trying to get you to seriously consider my position.

PCT is like heaven. The people who want to get here are dying to
get in;-) Their old misconceptions about human nature have to die
before they can be reborn as control theorists. You seem to be
comfortable with your misconceptions. So enjoy!

I agree with Mark Twain: if heaven is anything like the way it's described,
I don't want any part of it. I'd rather be alive. Anyway, I doubt if I'm
as comfortable with my "misconceptions" as you appear to be with yours.

I was talking about what is observed in the steady state. There is
a nice, systematic, IV-DV relationship between disturbance and
output

Yes. There is. But you said:

But when we hold, say, reference signal constant, then variation
in disturbance causes variation in output.

This statement is false because disturbance variations do not
_cause_ output variations in a closed loop system. Your statement
would have been true if you had said "disturbance variations are
systematically related to output variations" (as you did the
second time around) or "disturbance variations _appear to_ cause
output variations, but don't".

Now this is precisely what I have been talking about. Under my definition
of "cause" the statement is true; under yours it isn't. I have pointed out
the difference in meaning and noted that it only adds confusion to apply
one's own definitions to another's words after the other has made clear what
_he_ means by them.

I've heard a lot recently about how awful it is that psychologists sometimes
adopt words in common use and then apply special meanings to them within a
particular research context. It seems to me that this engineering
definition of cause represents a case in point -- a term borrowed from the
common vernacular and given a more specific meaning. I think the average
joe would be surprised to hear that turning the light switch "on" does not
cause the light to come on (if everything is working properly). The airline
pilot would be equally mystified (and alarmed!) to learn that pulling back
on an aircraft steering column does not cause the elevator to rise (even
though that relation is mediated by a servomechanism). The folks at the
Centers for Disease Contol would be equally surprised to learn that smoking
does not cause cancer. When palentologists suggest that the extinction of
the dinosaurs may have been caused by an asteroid collision, they do not
mean by this that every dinosaur was hit by the asteroid, but that the
asteroid set off a chain of events which eventually lead to the mass
extinction. The limited engineering definition simply does not cover the
most common uses of the term "cause," in or out of science.

I am not arguing that engineers should change their more limited definition
of cause, as I appreciate the need for it within the context in which the
definition is applied. I am arguing that when psychologists do the same
(for similar reasons), it is no sin.

As for substituting the phrase "systematically related to," I find it an
inadequate substitute, because it fails to convey the essential feature that
"cause" does: a change in Variable A (the cause) _leads to_ a change in
Variable B (the effect). The term "cause" asserts a directional property in
the systematic relationship that "systematically related to" does not.

The assertion that A causes B may be true whether or not it is also true
that B causes A, so by saying that A causes B, one has not ruled out
feedback or the possibility of circular causality.

I am not suggesting that the introduction of the term "cause" (however
defined) is necessary or fruitful when discussing how closed-loop systems
(including control systems) work. I can certainly get by without it. My
argument is directed against the notion that the use of this causal language
by someone necessarily implies that they hold to a particular model of
causality, i.e., stimulus-response. It seems to me that this is exactly the
inference often drawn here. It is the only explanation I can come up with
for silly ideas like the notion that psychologists believe incentives "make"
people do things.

I know how simple proportional control systems work.

The only thing you don't seem to know is how living control
systems work;-)

One of the things _you_ don't seem to know is that I do. You have a strong
motivation for believing that I don't. After all, if I did, then our
disagreements might hinge on something more substantive, and that is a
possibility you do not wish to consider.

I am confident that you will prevail against the hydra, preserving
your ingnorance of PCT. But the fight is fun even if the end result
is known.

Speaking of ignorance, I suggested in another post that you, a trained
"cognitive" psychologist, might know something about utility theory.
Perhaps it is just too painful to recall, given your present perspective?

Regards,

Bruce

2

[From bruce gregory (971201.1700 EST)]

Bruce Abbott (971201.1550 EST)

I am not arguing that engineers should change their more limited definition
of cause, as I appreciate the need for it within the context in which the
definition is applied. I am arguing that when psychologists do the same
(for similar reasons), it is no sin.

It seems to me that the problem is not one of terminology, but
one of the underlying model. You understand control models but
you treat them as simply one model among many. This allows you
to use language in apparently inconsistent ways. (This is
more than an appearance, for the models differ so greatly
that the terminology that they appear to share is in fact
inconsistent.) For reasons that I share, Rick is not fond of
many of the models you invoke. They are, as far as we are
concerned more like smoke and mirrors than they are like
Newton's laws. As long as you are committed to using these when
Rick is around he will point out their inadequacies.

As for substituting the phrase "systematically related to," I find it an
inadequate substitute, because it fails to convey the essential feature that
"cause" does: a change in Variable A (the cause) _leads to_ a change in
Variable B (the effect). The term "cause" asserts a directional property in
the systematic relationship that "systematically related to" does not.

The assertion that A causes B may be true whether or not it is also true
that B causes A, so by saying that A causes B, one has not ruled out
feedback or the possibility of circular causality.

That is a perfect example of the problem. In a world ruled by
circular causality attempts to talk in terms of linear causality
invariably look like arrant nonsense.

I am not suggesting that the introduction of the term "cause" (however
defined) is necessary or fruitful when discussing how closed-loop systems
(including control systems) work. I can certainly get by without it. My
argument is directed against the notion that the use of this causal language
by someone necessarily implies that they hold to a particular model of
causality, i.e., stimulus-response. It seems to me that this is exactly the
inference often drawn here. It is the only explanation I can come up with
for silly ideas like the notion that psychologists believe incentives "make"
people do things.

I have no idea what you are talking about. As nearly as I can
tell the word "cause" functions for you much as the word God
does for some theologians. It works in mysterious ways that
cannot fully be described or explained.

One of the things _you_ don't seem to know is that I do. You have a strong
motivation for believing that I don't. After all, if I did, then our
disagreements might hinge on something more substantive, and that is a
possibility you do not wish to consider.

I thought you have been arguing that there are no substantive
differences between you and Rick--that he was not understanding
your use of language. You seem to want it at least both ways.

Speaking of ignorance, I suggested in another post that you, a trained
"cognitive" psychologist, might know something about utility theory.
Perhaps it is just too painful to recall, given your present perspective?

I don't think it is fair to remind Rick of his misspent youth.
We all have things we would rather forget. Present company
excluded, of course...

Bruce

3

[From Rick Marken (971201.1530)]

Bruce Abbott (971201.1550 EST) --

I agree with Mark Twain: if heaven is anything like the way it's
described, I don't want any part of it.

Since I was talking about "PCT heaven", I think it's nice that
you have finally admitted that you want no part of PCT. I would
find you a lot more tolerable if you would just admit that you
are arguing against an idea (PCT) that you find intolerable
instead of trying to make us think that you are arguing for an
idea (PCT) that you love.

Under my definition of "cause" the statement is true; under yours
it isn't.

This is more than a word problem. It's an understanding problem.
I think that it is very possible that you do understand what's
going on here -- that you can't study a closed loop system in
the same way that you study an open loop system -- and that you
simply reject the implications (that psychology has to start the
study of behavior all over again, using a new methodology). You
may just be trying to use words to defend the psychological
status quo that you so admire. Perhaps this is why you say things
like what you say to Bill in Bruce Abbott (971201.1700 EST):

What other method is there, than to manipulate some variable under
controlled contitions and observe what happens to other variables
in the system?

Sounds like it's to the point, but it's a misdirection. PCT doesn't
say that we should study living systems using some method other
than "manipulating a variable under controlled conditions and
observing what happens to other variables in the system". PCT explains
how to do this manipulation when some of the system variables _may
be_ in a closed loop relationship. You are trying to use words
to make it seem that conventional psychologists have been studying
behavior the right way all along -- because it's the only way to study
behavior. But this is self-deception. There is a difference -- a BIG
difference -- between what conventional psychologists do when they
"manipulate some variable under controlled contitions and observe what
happens to other variables" and what a control theoriest does when
he does this. The difference (which you steadfastly refuse to
recognize as important) is described in my "Dancer..." paper.

When I said I would hold other conditions constant while varying
the disturbance, I did not include the output, which is after all
the dependent variable whose changes are to be accounted for.

But the output is not _just_ a dependent variable. It is also a
_confounding_ variable because it varies along with the IV and
contributes to the variance in the DV (the output itself). Calling
the output a DV is another linguistic trick; an attempt to make
it seem like there is no closed loop (effect of output on itself)
when there is. It's just another example of the kind of dopey things
you have to say to defend an incorrect point of view (the causal
model of conventional psychology).

But I don't know any other way to discover a system's organization
than to study it systematically, manipulating variables and
observing changes in others, and taking it apart if possible to
examine the functional properties of the components. Of course,
if you already have a model to test (such as a control model), you
can measure the relevant variables in the real system (if you can
get at them) and observe whether their changes correspond to those
of the model under the same conditions.

See. There you go with "the only way to study behavior is by
manipulating variables under controlled conditions and observing
changes in other variables". Yes. That's the only way to do it. But
you have to do it knowing that the behaving systems might be closed
loop. So the IV and DV may not be independent or dependent in the
way you think they are. There is now a section of my "Dancer.."
paper that addresses this issue. It's called "When to Use the TCV"
(they made me abbreviate Test for the Controlled Variable). The
answer is "whenever there is any possibility that you are dealing
with a closed loop system". So you always use the TCV when you are
studying a living system. If the system is actually open loop, the
TCV will tell you what you would have learned from a conventional
IV-DV approach -- so there is no price to pay for inappropriate use
of the TCV. If, however, you use the conventional IV-DV approach
to study what turns out to be a closed loop system there is a BIG
price to pay -- the behavioral illusion.

Both conventional IV-DV methods and the TCV can be described as
"manipulating a variable under controlled conditions and observing
changes in other variables". But the two methods are different in
some very important ways -- ways that you could learn about by
reading my "Dancer.." paper. You _could_ learn about it, that is,
if you weren't busy defending the demonstrably indefensible
status quo to yourself.

I think PCT would get a lot more milage out of your interest in
it if you would devote your energies to writing a paper about
"Why PCT Sucks". So far, your efforts to write the paper on
"PCT: The great new model of operant behavior" have been manifestly
unimpressive. Why not get in touch with yourself Bruce. Do what you
clearly love to do. Tell conventional psychologists what a dangerous,
wrongheaded piece of crap PCT is. Then maybe they'll pay attention
to it;-)

Best

Rick

ยทยทยท

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

4

i.kurtzer (971201)

[From Bruce Abbott (971201.1550 EST)]

to Rick Marken

Speaking of ignorance, I suggested in another post that you, a trained
"cognitive" psychologist, might know something about utility theory.
Perhaps it is just too painful to recall, given your present perspective?

Did this theory make predictions of the caliber of r=.98 ?
Do we know of any theory that does?
If a theory did make predictions of that accuracy while another didn't, which
theory would one choose?
Which theory would be more true?

i.