PCT not useful?

CSGnet Archive CSG1998
1

[Martin Taylor 980123 12:45]

Rick Marken (980122.0815)

The fact is that
we can't use PCT to say anything sensible about "well- established,
replicable phenomena" if we don't know whether these phenomena
involve control. And the only way to determine whether a phenomenon
involves control is to test for controlled variables.

I'm afraid I don't understand this. I thought that everything people do
(and everything other living things do) is control. If the fact of control
does _not_ suffice to describe things that always happen--or even
that usually or occasionally happen--under certain circumstances, then
that violates my belief that PCT is the correct foundational theory
for psychology. If a replicable phenomenon that describes something
that people do, or that is the result of something people do, cannot
be accounted for by PCT, then there must be something either wrong
or insufficient about PCT.

Are you saying now that PCT is _not_ correct? Or that it is only a part
of a theory, and that it does not apply to the things that are of most
interest to the man-on-the-street? It certainly sounds like one or the
other.

Or, since this was in a message aimed (for the N-ty-first time) at
discrediting Bruce Abbott, are you just pulling words out of the air
to make sure that anything he says is to be disregarded by the general
CSGnet readership?

Martin

2

[From Bill Powers (980124.1236 MST)]

Martin Taylor 980123 12:45--

..writing to Rick Marken (980122.0815)

I'm afraid I don't understand this. I thought that everything people do
(and everything other living things do) is control. If the fact of control
does _not_ suffice to describe things that always happen--or even
that usually or occasionally happen--under certain circumstances, then
that violates my belief that PCT is the correct foundational theory
for psychology.

Just a side remark here:

If we arbitrarily intepret all behavior as control behavior, without proof,
then PCT becomes an unfalsifiable theory. You can interpret anything as
control behavior if you want to. The reference condition of a raindrop is
to be on the ground. The distance above the ground is the error; the action
is to drop, and the outcome is that the error is corrected.

By always leaving the door open to a different possibility, we can make
every experiment with PCT into a test of the theory in which the result
could disprove the theory. There are experimental criteria that we say any
control system must must meet. It is always possible that a behavior we
observe is being produce by some other kind of system -- an S-R system, for
example. In any case of doubt, we can apply the Test and rule out other
explanations -- or the control-theory explanation, should that happen to be
the result.

We don't have to take the position that all behavior is control behavior,
and then defend it. PCT requires no defense. But it does require a test
that will show when it applies.

Best,

Bill P.

3

[From Tim Carey (980125.7450)]

[From Bill Powers (980124.1236 MST)]

Martin Taylor 980123 12:45--

..writing to Rick Marken (980122.0815)
>I'm afraid I don't understand this. I thought that everything people do
>(and everything other living things do) is control.

Just a side remark here:

control system must must meet. It is always possible that a behavior we
observe is being produce by some other kind of system -- an S-R system,

for

example. In any case of doubt, we can apply the Test and rule out other
explanations -- or the control-theory explanation, should that happen to

be

the result.

And could another possibility be that the particular thing we observe the
person doing is an unintended consequence of the process of control? So
even if we are observing a control system, not everything we see that
system _do_ will be an act of control .... hence the Test.

Cheers,

Tim

4

[Martin Taylor 980125 16:55]

Bill Powers (980124.1236 MST)]

Martin Taylor 980123 12:45--

..writing to Rick Marken (980122.0815)

I'm afraid I don't understand this. I thought that everything people do
(and everything other living things do) is control. If the fact of control
does _not_ suffice to describe things that always happen--or even
that usually or occasionally happen--under certain circumstances, then
that violates my belief that PCT is the correct foundational theory
for psychology.

Just a side remark here:

If we arbitrarily intepret all behavior as control behavior, without proof,
then PCT becomes an unfalsifiable theory.

Who is suggesting that? Not I!

You can interpret anything as
control behavior if you want to.

The issue is exactly the reverse. Rick seems to be asserting that there
is much that happens typically and ordinarily in the world of human
behaviour that _cannot_ be interpreted as control. In other words, he
seems already to have adopted the position that PCT _has been_ falsified.
He uses this position, without explanation, to discredit Bruce Abbott
(and now me), since neither of us believe as he claims to.

I dispute this assertion, and would replace it with a claim that _if_
PCT is an accurate description of human behaviour, _then_ it should be
able to account for phenomena that are always observed, usually observed,
or even sometimes observed when humans behave singly or in groups.
If there is such a phenomenon, as Rick claims, that cannot be accounted
for within PCT, then he is right: PCT has been demonstrated to be either
inadequate or false. I asked him which was the case, to which his helpful
response was that I would be incapable of understanding. Nice.

Mind you, I don't really believe that Rick thinks PCT has been falsified,
which is why I left him an out that he disdained to take.

+Tim Carey (980125.7450)

+And could another possibility be that the particular thing we observe the
+person doing is an unintended consequence of the process of control? So
+even if we are observing a control system, not everything we see that
+system _do_ will be an act of control .... hence the Test.

I assume that this is actually the way that a lot of commonly observed
phenomena will be accounted for within PCT. In particular, those things
that are consequences of the side-effect coupling among control systems,
such a tribalism and group conflict.

I don't a phenomenon falsifies PCT by being an unintended consequence
of control actions, especially when the consequence is quite predictable
if you consider the effects of one control system disturbing another.

Martin

5

[Martin Taylor 980125 16:40

Rick Marken (980124.1300)]

Me:

The fact is that we can't use PCT to say anything sensible about
"well- established, replicable phenomena" if we don't know whether
these phenomena involve control. And the only way to determine
whether a phenomenon involves control is to test for controlled
variables.

Martin Taylor (980123 12:45) -

I'm afraid I don't understand this.

And I'm sure you never will.

That's really not a very helpful reply, is it? Or in your mind, perhaps
it is intended to be helpful, and it's just another of your pronouncements
that I don't understand because of my impoverished mind. Which?

Martin

6

[From Bruce Gregory (980125.1744 EST)]

Martin Taylor 980125 16:40

Rick Marken (980124.1300)]

Me:

The fact is that we can't use PCT to say anything sensible about
"well- established, replicable phenomena" if we don't know whether
these phenomena involve control. And the only way to determine
whether a phenomenon involves control is to test for controlled
variables.

Martin Taylor (980123 12:45) -

I'm afraid I don't understand this.

And I'm sure you never will.

That's really not a very helpful reply, is it? Or in your mind, perhaps
it is intended to be helpful, and it's just another of your pronouncements
that I don't understand because of my impoverished mind. Which?

Gotta go with Martin on this one. The actions of organisms are all part of
their efforts to control. Rick knows this perfectly well. We need the test,
not to tell _whether_ organisms are controlling, but _what_ organisms are
controlling.

Bruce

7

[From Bill Powers (980125.14534 MST)]

Martin Taylor 980125 16:55 --

Rick seems to be asserting that there
is much that happens typically and ordinarily in the world of human
behaviour that _cannot_ be interpreted as control. In other words, he
seems already to have adopted the position that PCT _has been_ falsified.
He uses this position, without explanation, to discredit Bruce Abbott
(and now me), since neither of us believe as he claims to.

I read Rick differently. He is saying that he is interested only in control
phenomena, because only they are relevant to PCT. In this way we avoid
getting embroiled in arguments about whether some phenomenon does or
doesn't involve control. It isn't necessary to say that everything everyone
studies is really control behavior -- even if that is possibly true. All we
need to say is that we are interested only in studies that reveal
controlled variables. After all, why should we have to untangle anyone
else's procedures, logic, and analysis, or do their experiments over again
for them?

The other thing he is saying is that to investigate control phenomena, it's
necesary to do the test -- to find out what is being controlled. No study
that omits looking for controlled variables is of any interest in PCT,
because the Test is the only way to be sure you know what is being
perceived and controlled, if anything (as sure as possible).

I dispute this assertion, and would replace it with a claim that _if_
PCT is an accurate description of human behaviour, _then_ it should be
able to account for phenomena that are always observed, usually observed,
or even sometimes observed when humans behave singly or in groups.

Yes, if the theory applied to all behavior it would apply to all behavior.
But the catch is proving it. You can think of a way to interpret behavior
as control in a given circumstance, but if challenged you have to be able
to prove it, according to acceptable criteria. All Rick is saying is that
if you want to make statements about all kinds of behavior, you have to be
ready to put your experimental results where your mouth is.

If there is such a phenomenon, as Rick claims, that cannot be accounted
for within PCT, then he is right: PCT has been demonstrated to be either
inadequate or false. I asked him which was the case, to which his helpful
response was that I would be incapable of understanding. Nice.

Well, you're right to ignore that kind of belligerent insult.

Best,

Bill P.

8

[From Rick Marken (980125.2120)]

Martin Taylor (980125 16:55) --

If there is such a phenomenon, as Rick claims, that cannot be
accounted for within PCT, then he is right: PCT has been
demonstrated to be either inadequate or false. I asked him which
was the case, to which his helpful response was that I would
be incapable of understanding. Nice.

Bill Powers (980125.14534 MST) --

Well, you're right to ignore that kind of belligerent insult.

Assuming that this is a refernce to my [Rick Marken (980124.1300)]
brief reply to Martin, let me make a few brief points:

First, Martin didn't ignore the "insult".

Second, it was not belligerent; it was tres belle:-)

Third, it was not an insult; it was a prediction (see below).

Fourth, it was not even what I said. I didn't say (as Martin says
I said) that Martin ".. would be _incapable_ of understanding" what
I said about conventional behavioral research. I said that I was
"...sure Martin _never would_ understand it. That is a _prediction_
based not on Martin's mental capabilities (which are considerable)
but on my estimation of some of the perceptions which he is
oontrolling for -- essentially the same perceptions as those Bruce
Abbott is controlling for.

I answered Martin's (980123 12:45) post where he said:

I'm afraid I don't understand this.

With my confident prediction

And I'm sure you never will.

instead of a detailed reply because we have been over this same
territory with Martin many times before. My prediction was not
a belligerent insult but a white flag of surrender; it was my
admission that there is nothing I can say or do that will get
Martin to understand the PCT point of view on conventional
psychological research (the point of view you described _again_
so well in your reply [Bill Powers (980125.14534 MST)] to Martin's
[Martin Taylor 980125 16:55] misinterpretations of my descriptoin
of that same point of view) because the PCT point of view is a
big disturbance to other perceptions that Martin is controlling;
perceptions like "new theories must account for existing data",
"conventional behavioral data is good data and any theory of
behavior must account for it" and "PCT is just another theory of
behavior (as conventionally understood)".

Best

Rick

···

--

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

9

[From Bruce Gregory (980126.10000 EST)]

Bill Powers (980124.1236 MST)

Martin Taylor 980123 12:45--

..writing to Rick Marken (980122.0815)
>I'm afraid I don't understand this. I thought that everything people do
>(and everything other living things do) is control. If the fact of control
>does _not_ suffice to describe things that always happen--or even
>that usually or occasionally happen--under certain circumstances, then
>that violates my belief that PCT is the correct foundational theory
>for psychology.

Just a side remark here:

If we arbitrarily intepret all behavior as control behavior, without proof,
then PCT becomes an unfalsifiable theory. You can interpret anything as
control behavior if you want to. The reference condition of a raindrop is
to be on the ground. The distance above the ground is the error; the action
is to drop, and the outcome is that the error is corrected.

By always leaving the door open to a different possibility, we can make
every experiment with PCT into a test of the theory in which the result
could disprove the theory.

Not likely. In order to make any experiment a test of PCT, we
would have to convince ourselves that we had "done everything
right" and to rule out the possibility that we had not been
clever enough to guess which variable was being controlled. It
appears to be easier to demonstrate successful applications of
PCT than to subject it to a critical test of the sort Popper
loved so much. This is true of Newtonian physics, as well, so I
doesn't count as a black Mark against PCT.

There are experimental criteria that we say any
control system must must meet. It is always possible that a behavior we
observe is being produce by some other kind of system -- an S-R system, for
example. In any case of doubt, we can apply the Test and rule out other
explanations -- or the control-theory explanation, should that happen to be
the result.

Again all we can rule out is that the environmental variable
which we are testing is controlled, not that the behavior is the
result of control of another variable.

We don't have to take the position that all behavior is control behavior,
and then defend it. PCT requires no defense. But it does require a test
that will show when it applies.

Since PCT is the only theory we have of purposeful action, it
behooves us to assume that all behavior is linked to the
exercise of control. At least until a better theory comes along.
(Note that the anomalous advance of the perihelion of Mercury
was not seen as a counter instance to Newtonian theory until
General Relativity provided an explanation.)

Bruce

10

[From Bruce Nevin (980126.1155 EST)]

(Martin Taylor 980125 16:40)--

Rick Marken (980124.1300)]

The fact is that we can't use PCT to say anything sensible about
"well- established, replicable phenomena" if we don't know whether
these phenomena involve control. And the only way to determine
whether a phenomenon involves control is to test for controlled
variables.

Martin Taylor (980123 12:45) -

I'm afraid I don't understand this.

And I'm sure you never will.

That's really not a very helpful reply

In another context, Rick (980121.1445) said

the perceptions I'm controlling by doing this include
1) seeing people get PCT right 2) testing my own mastery of PCT
3) providing entertainment for others on the net who understand
PCT. There are probably many more.

Maybe Rick could now add to this list of perceptions whose control is more
important to him than tact or even than (IMO) effectiveness in controlling
1 and 3, and probably all three.

Help me to understand, Martin. What can we appropriately use PCT to talk
about other than control?

  Bruce

11

[From Bruce Nevin (980126.1210)]
(Martin Taylor 980125 16:40)--
(Martin Taylor 980125 16:55)--

OK, I got Martin's intention backward. (Reading sporadically and out of
sequence.)

I do not think it is credible that all "observed behavior" of an organism
is a reflection of control.

o There are disturbances that overwhelm control.

o There are effects of the environment on the physical disposition of the
organism without affecting perceptions that the organism is at the moment
controlling.

In other words, without the Test, you cannot distinguish behavior from
successful interference with control or manipulation without disturbance to
controlled perceptions, both of which can have the appearance of "observed
behavior."

The shift from the observer's point of view to that of the observed
organism is fundamental. Is there any way to make this shift other than the
Test? Any such alternative would be a valuable addition to PCT methodology.

  Bruce Nevin

12

[From Rick Marken (980126.1020)]

Bruce Gregory (980126.10000 EST)

In order to make any experiment a test of PCT, we would have to
convince ourselves that we had "done everything right"

We don't have to do _everything_ right; all we have to do right
are the steps that constitute The Test for controlled variables.

and to rule out the possibility that we had not been clever
enough to guess which variable was being controlled.

The Test allows us to rule out this possibility.

Bill Powers:

we can apply the Test and rule out other [non-PCT] explanations --
or the control-theory explanation, should that happen to be
the result.

Bruce G.

Again all we can rule out is that the environmental variable
which we are testing is controlled, not that the behavior is the
result of control of another variable.

But the behavior we are interested in _is_ that of the environmental
variable that we suspected of being under control. Once we find
that that variable is _not_ controlled, we have ruled out a PCT
explanation of its behavior.

You seem to be saying that any particular behavior is almost certainly
the result or a side effect of controlling. This may be true (at least
for the behavior of living systems). But it is irrelevant to how we
go about studying and modeling control. Just because a behavior, like
bar pressing, is likely to have something to do with the controlling
done by a control system doesn't mean that we have to try to find a
PCT explanation for it; if we don't know how the behavior relates to
controlled inputs, any explanation of that behavior is going to
be either a fantasy (at best) or a delusion (more likely).

Perhaps a concrete illustration will help. One of the basic phenomena
of conventional psychology is the effect of schedule of reinforcement
on response rate. For ratio schedules (that are not too demanding)
the finding is that response rate (presses/sec) is directly related
to schedule ratio (presses/reinforcement); the higher the ratio
(the more presses required for a reinforcement) the higher the
rate of response (more presses/sec). Martin (and you, apparently)
say that PCTer should explain this phenomenon using PCT or admit
that PCT is useless. OK. It turns out that Bill Powers did build a
PCT model of this phenomenon (I'm honored to say that he started it
at my house while on a visit to LA long ago); so PCT can explain
this observation. But what is the model explaining?

In order to build the model, Bill had to assume that the organism
was controlling some variable; I believe he guessed that it was
rate of food input. Given this assumption and some parameter
twiddling Bill was able to build a model that fit the data with
impressive accuracy. The success of the model might convince you
that response rate is, indeed, an aspect of controlling; that it
is the means by which the organism controls its rate of food input.
If this were your conclusion, however, you would be wrong -- big
time; the success of the PCT model was completely misleading.

It turns out that the relationship between schedule ratio and
measured response rate has nothing to do with controlling. Bruce
Abbott (of all people) discovered that it is an artifact of the way
response rate is measured (by dividing the number of presses by
the length of a session). When the schedule ratio is low,
organisms spend more of their time eating than pressing; when
the schedule ratio is high, organisms spend more time pressing
(at the _same_ rate they did in the low schedule ratio condition)
than eating. The apparently higher rate of pressing in the ratio
schedule is simply a result of the fact that more presses (and
less eating) occurs in the high ratio than in the low ratio
schedule sessions.

So Bill ended up writing a lovely and convincing PCT model of a
phenomenon that doesn't involve control; indeed, of a phenomenon
that (being an artifact) doesn't even exist. If I understand you
and Martin correctly, this modeling effort should be counted as
an important success because it showed how PCT could explain one
of the basic observations of conventional psychology. I guess I
don't see it that way; I see it as something PCTers should avoid
like the plague; and the way to avoid it is by not taking the data
of conventional psychology at face value. The way to avoid it is
by redoing conventinoal experiments and including a Test to see
if any variables in these experiments are under control.

I'm quite sure that PCT models can be built that mimic any of
the basic findings of psychological reseach; heck, I'm sure that
PCT models can be built to mimic the behavior of the planets
(which I'm pretty sure are not controlled by anyone). PCT is
just a set of differential equations that can probably be made to
behave in nearly any way you want; so you can probably model
_any_ behavior with PCT. But this doesn't strike me as a useful
(or interesting) way to approach PCT. PCT is useful for undersanding
purposeful behavior; so the first thing to do before trying to
use (apply) PCT is to make sure that what you are trying to
understand is, indeed, an example or aspect of purposeful behavior.
The first thing is always _phenomena_ - and the main phenomena of
interest to PCT are controlled variables. Conventional behavioral
research tells us nothing about controlled variables -- that is
why PCT has nothing useful to say about the phenomena "discovered"
be conventional research -- and that is why I no longer feel like
it's necessary to _prove_ PCT by fitting it to the results of
conventional psychological research

Best

Rick

···

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

13

[From Bruce Gregory (980226.1438)]

Rick Marken (980126.1020)

Bruce Gregory (980126.10000 EST)

> In order to make any experiment a test of PCT, we would have to
> convince ourselves that we had "done everything right"

We don't have to do _everything_ right; all we have to do right
are the steps that constitute The Test for controlled variables.

Assuming we have tested all possible controlled variables.

> and to rule out the possibility that we had not been clever
> enough to guess which variable was being controlled.

The Test allows us to rule out this possibility.

No, the Test allows us to rule that the variable investigated is
not a controlled variable. The Test tells us nothing about other
variables. In order to test PCT (and show it is not adequate) we
would have to conduct the Test on every possible variable.

Bill Powers:

> we can apply the Test and rule out other [non-PCT] explanations --
> or the control-theory explanation, should that happen to be
> the result.

Bruce G.

> Again all we can rule out is that the environmental variable
> which we are testing is controlled, not that the behavior is the
> result of control of another variable.

But the behavior we are interested in _is_ that of the environmental
variable that we suspected of being under control. Once we find
that that variable is _not_ controlled, we have ruled out a PCT
explanation of its behavior.

All we have ruled out is that the behavior is produced by
efforts to control the variable we have identified.

You seem to be saying that any particular behavior is almost certainly
the result or a side effect of controlling. This may be true (at least
for the behavior of living systems).

Thanks for acknowledging this. It was the only point I was
trying to make.

But it is irrelevant to how we
go about studying and modeling control. Just because a behavior, like
bar pressing, is likely to have something to do with the controlling
done by a control system doesn't mean that we have to try to find a
PCT explanation for it.

Presumably we do not _have_ to try to find a PCT explanation of
_anything_.

if we don't know how the behavior relates to
controlled inputs, any explanation of that behavior is going to
be either a fantasy (at best) or a delusion (more likely).

This is always true.

Perhaps a concrete illustration will help. One of the basic phenomena
of conventional psychology is the effect of schedule of reinforcement
on response rate. For ratio schedules (that are not too demanding)
the finding is that response rate (presses/sec) is directly related
to schedule ratio (presses/reinforcement); the higher the ratio
(the more presses required for a reinforcement) the higher the
rate of response (more presses/sec).

Martin (and you, apparently)
say that PCTer should explain this phenomenon using PCT or admit
that PCT is useless.

I certainly don't say this. Nor do I have any evidence that
Martin would say this.

OK. It turns out that Bill Powers did build a
PCT model of this phenomenon (I'm honored to say that he started it
at my house while on a visit to LA long ago); so PCT can explain
this observation. But what is the model explaining?

In order to build the model, Bill had to assume that the organism
was controlling some variable; I believe he guessed that it was
rate of food input. Given this assumption and some parameter
twiddling Bill was able to build a model that fit the data with
impressive accuracy. The success of the model might convince you
that response rate is, indeed, an aspect of controlling; that it
is the means by which the organism controls its rate of food input.
If this were your conclusion, however, you would be wrong -- big
time; the success of the PCT model was completely misleading.

It turns out that the relationship between schedule ratio and
measured response rate has nothing to do with controlling. Bruce
Abbott (of all people) discovered that it is an artifact of the way
response rate is measured (by dividing the number of presses by
the length of a session). When the schedule ratio is low,
organisms spend more of their time eating than pressing; when
the schedule ratio is high, organisms spend more time pressing
(at the _same_ rate they did in the low schedule ratio condition)
than eating. The apparently higher rate of pressing in the ratio
schedule is simply a result of the fact that more presses (and
less eating) occurs in the high ratio than in the low ratio
schedule sessions.

So Bill ended up writing a lovely and convincing PCT model of a
phenomenon that doesn't involve control; indeed, of a phenomenon
that (being an artifact) doesn't even exist. If I understand you
and Martin correctly, this modeling effort should be counted as
an important success because it showed how PCT could explain one
of the basic observations of conventional psychology.

You do not understand me correctly. Martin can speak for
himself. Notice that you are not talking about a behavior, but
about a putative relationship between measures of behavior.

I guess I
don't see it that way; I see it as something PCTers should avoid
like the plague; and the way to avoid it is by not taking the data
of conventional psychology at face value. The way to avoid it is
by redoing conventinoal experiments and including a Test to see
if any variables in these experiments are under control.

A perfectly reasonable suggestion. Perhaps Martin agrees.

I'm quite sure that PCT models can be built that mimic any of
the basic findings of psychological reseach; heck, I'm sure that
PCT models can be built to mimic the behavior of the planets
(which I'm pretty sure are not controlled by anyone). PCT is
just a set of differential equations that can probably be made to
behave in nearly any way you want; so you can probably model
_any_ behavior with PCT.

When one is learning PCT even hypothetical explanations can be
illuminating. At least I find many of Bill's illuminating,
even though I know he has not conducted the Test in most of
these cases.

But this doesn't strike me as a useful
(or interesting) way to approach PCT. PCT is useful for undersanding
purposeful behavior; so the first thing to do before trying to
use (apply) PCT is to make sure that what you are trying to
understand is, indeed, an example or aspect of purposeful behavior.

Excellent advice.

The first thing is always _phenomena_ - and the main phenomena of
interest to PCT are controlled variables. Conventional behavioral
research tells us nothing about controlled variables -- that is
why PCT has nothing useful to say about the phenomena "discovered"
be conventional research -- and that is why I no longer feel like
it's necessary to _prove_ PCT by fitting it to the results of
conventional psychological research.

Perfectly reasonable. Just notice that _I_ am talking about
behavior and _you_ are talking about "phenomena 'discovered' by
contentional research.

Bruce

14

[From Bruce Nevin (980226.1547 EST)]

Bruce Gregory (980226.1438)--

Just notice that _I_ am talking about
behavior and _you_ are talking about "phenomena 'discovered' by
contentional research.

Help! I've lost track completely of what behavior you are talking about.
Perhaps a reference to a CSG datestamp will do?

  Bruce Nevin

15

[From Bruce Gregory (980126.1620 EST)]

Bruce Nevin (980226.1547 EST)]

Bruce Gregory (980226.1438)--

>Just notice that _I_ am talking about
>behavior and _you_ are talking about "phenomena 'discovered' by
>contentional research.

Help! I've lost track completely of what behavior you are talking about.
Perhaps a reference to a CSG datestamp will do?

[Bruce Gregory (980125.1744 EST)]

Gotta go with Martin on this one. The actions of organisms are
all part of their efforts to control. Rick knows this perfectly
well. We need the test, not to tell _whether_ organisms are
controlling, but _what_ organisms are controlling.

Bruce

16

[From Bruce Gregory (980126.1732 EST)]

Rick Marken (980126.1400)

Bruce Gregory (980226.1438)--

> the Test allows us to rule that the variable investigated is
> not a controlled variable. The Test tells us nothing about other
> variables.

I see the Test as aimed at finding variables that _are_ controlled,
not at eliminating variables that are not. The Test is iterative;
as hypotheses about the variable under control are eliminated we
come up with new hypotheses until we either 1) find a variable
that is under control or 2) we give up and go play raquetball.

I was referring to Bill's statement about testing PCT. Successes
increase our confidence in PCT, but failures at this point only
suggest we haven't been smart enough to guess the right CV.

> In order to test PCT (and show it is not adequate) we would have
> to conduct the Test on every possible variable.

Not at all. We test PCT by seeing how well it explains control --
actual examples of control that have been discovered by Test. That's
why our tracking experiments are so important; they are tests of PCT
that show (every time) that PCT is an adequate model of this
particular proven example of purposeful behavior. We continue to
test PCT by seeing whether or not it is adequate to account for
other demonstrated examples of control. Controlling the optical
path of a baseball is another demonstrated example of control
(demonstrated by the Tests conducted by McBeath, et al) and, lo
and behold, PCT seems to handle this kind of control rather
nicely (it even suggested that the controlled variable may be
different than the one -- Linear Optincal Trajectory -- suggested
by McBeath et all; further Tests are needed to find out).

Yes. This is exactly what I said above. I was simply pointing
out that there are no crucial experiments that allow us to
_reject_ PCT as Sir Karl would have us do.

Me:

> But the behavior we are interested in _is_ that of the environmental
> variable that we suspected of being under control. Once we find
> that that variable is _not_ controlled, we have ruled out a PCT
> explanation of its behavior.

Bruce G.

> All we have ruled out is that the behavior is produced by
> efforts to control the variable we have identified.

I guess what I'm saying is that PCT suggests that it is
ridiculous to pick some arbitrary behavior (like bar press
rate) and then go off and try to find the variable it might
be involved in the control of. The behavior might _not_ be
involved in the control of any variable (as in the operant
scheduling example I gave). Just because some behavior (like
bar press rate) is interesting to me doesn't mean that it is
involved in the control of some variable. It might just be
a bunch of stuff that's happening.

I don't think I follow you. I _assume_ (perhaps wrongly) that
whatever the rat is doing is purposeful and therefore involves
the control of one or more EVs. Nothing a LCS does is "just a
bunch of stuff that's happening" unless the LCS is
"unconscious".

> Presumably we do not _have_ to try to find a PCT explanation of
> _anything_.

That's sort of my point, mainly because it's going about things
ass backwards. I don't think it's reasonable (or even possible)
to try to find PCT explanations of all the randomly noticed
behaviors we see (in terms of the variables that these behaviors
are involved in the control of); what we should be trying to
do is finding the variables that organisms control , trying to
develop PCT explanations of this control and _then_ pointing
out what kinds of behaviors one _might_ see when an organism
is busy controlling these variables.

> Perfectly reasonable. Just notice that _I_ am talking about
> behavior and _you_ are talking about "phenomena 'discovered' by
> conventional research.

Do you think, then, that PCT should be used as an explanation
of "behavior" but not of "phenomena 'discovered' by conventional
research"? How do I distinuish these two? Is "bar press rate"
a behavior to be explained by PCT?

No. It's a "phenomenon". As you say, the rate may be an
artifact, but pushing the bar is not -- it is a purposeful
action.

Is the "relationship between
bar press rate and reinforcement schedule" a phenomenon that does
not need to be explained by PCT?

Right.

Bruce

17

[From Bill Powers (980126.1637 MST)]

Bruce Gregory (980126.1732 EST)--

I was referring to Bill's statement about testing PCT. Successes
increase our confidence in PCT, but failures at this point only
suggest we haven't been smart enough to guess the right CV.

Failure of a proposed CV to pass the test is absolute: there's no way to
interpret it as a success. But success is equivocal, because it tells us
only we have a variable that is related to the true controlled variable.
The attempt to refine the definition of the controlled variable by using
the test is endless. You quit when you'e sure enough for practical
purposes. Consider Rick's "area versus perimeter" test for controlled
sizes. Are those the only two choices? Of course not. But we least we can
tell which is the better choice.

Best,

Bill P.

18

[From Bruce Gregory (980130.1335 EST)]

It occurs to me that in the same way that the place to look
for your model of the world is "out there" rather than inside
your head, "out there" is also the place to look for your
intentions. Forget what you want. Look to your actions. They
will tell you what your intentions are.

Bruce

19

i.kurtzer (970130.1000)
<< [From Bruce Gregory (980130.1335 EST)]

It occurs to me that in the same way that the place to look
for your model of the world is "out there" rather than inside
your head, "out there" is also the place to look for your
intentions. Forget what you want. Look to your actions. They
will tell you what your intentions are.

  >>

I am near sure that what you mean is " the controlled perceptions subsequent
of output" with action as a understood shorthand. Cool?
However, from a certain abstraction this "search for structure of our
intentions" has seemed extremely problematic and there has been a _long_ and
heated--of course some by insincere or downright silly flames--debate
centering on this---on where does the meaning of our percepts point to, hinge
on, etc. So would it be to the "external" world or other percepts---roughly,
for the former we have the material monists and dualists; one the other we
have the idealists and pragmatists. I am of the opinion that this problem can
resolved given a reconsideration of my favorite Mr. W. James with a
hierarchical twist.
But just for laughs here's one of the principle scenarios that philosophers
have racked their brains over, I suppose since the Scholastics.
Who did Oedipus care to sleep with, his mother or his wife.

the question seem to resolve to "where is the structure", mi noodle or that
other real world; that is, as the question has been framed so far.
on the side, one of the practical difficulties of this possibly questionable
question is in the field of learning:when did the rat get to know about that
funny bar, or click, or what not. The way its been adressed so far is
basically justified by saying that the rat _already_ knew it, but differently.
As this practically troubles me please feel free to comment.

i.

20

[From Bruce Gregory (980201.1612 EST)]

i.kurtzer (970130.1000)

<< [From Bruce Gregory (980130.1335 EST)]

It occurs to me that in the same way that the place to look
for your model of the world is "out there" rather than inside
your head, "out there" is also the place to look for your
intentions. Forget what you want. Look to your actions. They
will tell you what your intentions are.

>>

I am near sure that what you mean is " the controlled perceptions subsequent
of output" with action as a understood shorthand. Cool?

My intention was to say that by becoming aware of our actions we become aware
of our intentions (or the reference levels associated with our exercise of
control). These reference levels are not perceptions, so they don't "show up"
for us directly.

However, from a certain abstraction this "search for structure of our
intentions" has seemed extremely problematic and there has been a _long_ and
heated--of course some by insincere or downright silly flames--debate
centering on this---on where does the meaning of our percepts point to, hinge
on, etc. So would it be to the "external" world or other

percepts---roughly,

for the former we have the material monists and dualists; one the other we
have the idealists and pragmatists. I am of the opinion that this problem

can

resolved given a reconsideration of my favorite Mr. W. James with a
hierarchical twist.

Yes, I too am a fan of WJ. But here I am simply talking about the scientific
"received view" which, as Bill has pointed out, tells us that all we have
direct access to is our perceptions. Therefore whatever we perceive as out
there must be a construct from these perceptions. (Which is not to deny that
there is something "out there".)

But just for laughs here's one of the principle scenarios that philosophers
have racked their brains over, I suppose since the Scholastics.
Who did Oedipus care to sleep with, his mother or his wife.

on the side, one of the practical difficulties of this possibly questionable
question is in the field of learning:when did the rat get to know about that
funny bar, or click, or what not. The way its been adressed so far is
basically justified by saying that the rat _already_ knew it, but

differently.

Could you say a little more about this? I'm not certain that I follow you.

Bruce