Controlled variables

[Bruce_Abbott]

[From Bruce Abbott (960123.1815 EST)]

Rick Marken (960123.1300) --

In PCT we use the term "result" or "consequence" rather than "act" to refer
to a perceptual consequence of action. But I can see that "act" is defined
in the dictionary as "a result of action" so your usage was quite
appropriate. I would prefer to stick with the term "result" when we are
talking about the result of actions. But I also like going with the
dictionary; so I'm sorry about the misunderstanding.

Good; I'm glad that we've gotten our "act" together!

Here's my (well, Bill's) theory: The rat learns to control the perception of
"a struck key" because ingestion and digestion of the grain that results from
controlling this variable _slows or stops_ the process of reorganization. The
"incentive" (the grain) doesn't "lead to the establishment of a reference"
for striking the key; rather, it allows the _cessation_ of reorganization;

Or in words you would rather not hear used, by reducing intrinsic error it
leads to the cessation of reorganization, which leaves the current candidate
reference in place or "established." Hmmm, so the occurrence of the
incentive _does_ lead to the establishment of a reference for striking the
key, in your view. For a moment there I thought you were disagreeing with
me. (;->

the organism stops trying to control other variables; it keeps controlling
for "a struck key" because doing so keeps the organism in control of an
intrinsic variable (like blood suger level). It is the error in the system
controlling this intrinsic variable that drives reorganization. As intrinsic
error decreases (because whatever the organism is currently controlling
produces, as a side effect, the desired intrinsic perception) the (possibly
random) changes in the variables being controlled ("a struck dot on floor",
"a struck dot on ceiling", "a movement of wings", etc) stops; the
pigeon keeps controlling for "a struck key".

This it where I have trouble with this explanation. Reorganization is said
to come into play when "intrinsic" variables are in "persistent" error, and
to stop when this error is "corrected." That's a fairly slow process. I
don't see how one brief grain-delivery (or even several) is going to have
much impact on the level of deprivation (intrinsic error), yet the pigeon
goes back and repeats what it just did (in terms of result of action).
We've been over this ground before, I know, so there probably isn't much
point in rehashing the issue until we get honest-to-goodness real data to
argue from; I just want it known that I have difficulty with the explanation
as it stands. The basic idea -- that you stop looking for a solution to a
problem the moment you find it -- seems reasonable enough, but the mechanism
involved in identifying the problem and the solution seems too sluggish and
too dependent on error at the highest level in the hierarchy.

Regards,

Bruce

[R_Marken]

[From Rick Marken (980307.1250)]

Bill Powers (980307.0708 MST) --

The problem here is that you and Jeff are using "disturbance"
differently.

That is surely an aspect of it. But it still doesn't explain
statements like:

But then what is the physical system that allows o to have an
impact on d?

Even if Jeff means DS when he says d there is still no "physical
system" that allows o to have an impact on DS, at least not in
the "coin game" example I gave (unless the "physical system" is
the neural implementation of the perceptual function). The apparent
effect of d on o (it's hard to talk about the "effect" of DS)
exists because both contribute to the value of a perceptual
variable as independent inputs to a perceptual function, not
because there is any effect of o and d on each other _or_ on
some common physical variable.

Again, I think the central problem here is the "controlled
variable". I can't see how anyone who knew what a controlled
variable (or controlled perception) was could continue to do
research using conventional methodology and think they were
learning something about behavior. I can't see how anyone who
knew what a controlled variable (or controlled perception)was
could believe that it is possible to learn about controlled
variables by simply observing IV-DV (disturbance-output)
relationships.

I think people are able to fit PCT into the cause-effect framework
by imagining that disturbance-output relationships are _equivalent_
to controlled variables. I think that's why we hear so little about
controlled variables from CPs who are would-be PCTers; they must
think they have discovered controlled variables when they see a
relationship between a distrubance and an output variable. I think
this is how it is possible for there would-be PCTers to believe
that conventional research provides (and has always provided)
data relevant to PCT.

I think the whole thing (understanding PCT) comes down to knowing
what controlled variables are, accepting the possibility that they
exist and knowing how to demonstrate their existence. That's why
I think the Robertson/Goldstein paper is very important. As I
recall, it was a great first step at showing how to Test for
fairly abstract controlled variables (like "self esteem").

Best

Rick

···

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

[Bill_Powers1]

[From Bill Powers (980308.0133 MST)]

Rick Marken (980307.1250)--

The problem here is that you and Jeff are using "disturbance"
differently.

That is surely an aspect of it. But it still doesn't explain
statements like:

But then what is the physical system that allows o to have an
impact on d?

Jeff may be thinking of d as the perturbation of qi away from its reference
level. With that idea in mind, it makes at least some sense to speak of the
impact of the output on this perturbation (working against it). It's not
quite the right way to analyze the system -- the actual perturbation
reaches its maximum, normally, at about the same time that the output
produces the maximum opposing effect, so the cancellation of disturbance
effects has already happened.

Maybe that's not what Jeff thinks. Perhaps we should ask.

Best,

Bill P.

[R_Marken]

[From Rick Marken (980308.1030)]

Bill Powers (980307.1759 MST) --

The problem that is creating all our difficulties is this variable
we're calling qi. That variable is really a fiction, one we use
because we can't observe anyone else's perceptual signals.

...So where, starting with this model, does the idea of the input
quantity come from, an input quantity that is controlled instead
of a collection of v's? It comes from the observer. The observer
receives inputs from the same or a similar collection of v's, and
the observer's perceptual input function combines them to generate
a perceptual signal. It is this perceptual signal that the observer
calls "qi" and projects, conceptually, into the environment to take
the place of the collection of v's.

This is an excellent example of the value of our net discussions.
This is exactly the point I was trying to get at in an earlier post:

Rick Marken (80306.1400) --

"Controlled variable" referes to the observer's _perception_ of
the variable that is being controlled by the controller (as a
controlled variable).

But I couldn't quite get to the best part. Bill did get to it
when he notes that the variable qi, which we call the "controlled
variable" in our PCT diagrams and represent as though it actually
_existed_ as a variable in the environment (we write in on the
"environment" side of the disgram), is really a FICTION!

There is no variable qi in the environment; it exists only as a
perception in the observer. Perhaps one way to clarify this point
is to change the notation of the basic control system diagram. Bill
suggests going back to the notation used in his Science article
(see diagram in LCS, p.66). I like this idea but it gives us no
special symbol for the observer's perception of the function of
environmental variables (v's) that is the "controlled variable" (qi).
I modestly suggest keeping the "circle of v's" notation (LCS p.66)
to represent the environmental variables that make up the controlled
variable and that we continue to refer to the function of v's that
is under control as the "controlled variable", But I also suggest
that we symbolize the controlled variable as p' rather than as qi.
The idea is to explicity note that the controlled variable is a
_perception_ (hence the p) but that this perception is the observer's
perceptual "estimate" (hence the "prime" symbol) of the perception (p)
controlled by the control system.

I think our tracking experiments have had the unfortunate side
effect of encouraging a belief in qi as a real environmental
variable. However qi is conceived of in these experiments (position
of cursor, distance from cursor to target) these variables seem
more "real" (more "environmental") than variables like "self-esteem"
or "honesty". Of course, as you [Bill Powers (980307.1759 MST)]
note later in your post, variables like "position of cursor" and
"distance from cursor to target" are just as much perceptions
(functions of environmental v's) as are variables like "self-esteem"
and "honesty". I think the problem relates to something you
mentioned in B':CP (I think); lower level perceptions (sensations,
configurations, etc) seem more "real" (really "out there") than
higher level perceptions (principles like "honesty", system
concepts like "Christianity").

Anyway, one thing that is needed, I think, is a control demo where
the controlled variable is far less obvious than the disturbance
to that variable. This is more like the situation in real life
(and real psychological experiments) where we can see disturbances, d,
(verbal questions, light levels, etc) and outputs, o, (reaction time,
ratings, etc) very clearly but not the controlled variable, p'.
This is quite different from the situation in most of our control
demos, where we can see the controlled variable (distance from
cursor to target, say) and output (handle movements) quite easily,
but not the disturbance.

I will try to develop such a demo. I hope Bruce Abbott doesn't beat
me to it;-)

Best

Rick

···

--

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

[Bruce_Gregory]

[From Bruce Gregory (980308.1504 EST)]

Rick Marken (980308.1030)

Bill Powers (980307.1759 MST) --

The problem that is creating all our difficulties is this

variable

we're calling qi. That variable is really a fiction, one we use
because we can't observe anyone else's perceptual signals.

...So where, starting with this model, does the idea of the input
quantity come from, an input quantity that is controlled instead
of a collection of v's? It comes from the observer. The observer
receives inputs from the same or a similar collection of v's, and
the observer's perceptual input function combines them to

generate

a perceptual signal. It is this perceptual signal that the

observer

calls "qi" and projects, conceptually, into the environment to

take

the place of the collection of v's.

This is an excellent example of the value of our net discussions.
This is exactly the point I was trying to get at in an earlier

post:

Rick Marken (80306.1400) --

"Controlled variable" referes to the observer's _perception_ of
the variable that is being controlled by the controller (as a
controlled variable).

This too is the point I was aiming at talking about different
perceptual worlds.

But I couldn't quite get to the best part. Bill did get to it
when he notes that the variable qi, which we call the "controlled
variable" in our PCT diagrams and represent as though it actually
_existed_ as a variable in the environment (we write in on the
"environment" side of the disgram), is really a FICTION!

Yes, but only if you add that the contents of _all_ perceptual
worlds are fictions. What we often lose sight of is that the
diagrams we draw are representations of _our_ worlds, not
representations of _the_ world. (A similar situation arises in
physics where we talk about forces that arise only in the observer's
frame of reference (centrifugal force, for example) as being
"fictitious". A colleague of mine is want to say, "hell, they're
_all_ fictitious.)

Bruce

[Bruce_Abbott]

[From Bruce Abbott (980308.1505 EST)]

Rick Marken (980308.1030) --

I will try to develop such a demo. I hope Bruce Abbott doesn't beat
me to it;-)

Not to worry: I'm MUCH too busy attempting to defend my outmoded beliefs in
S-R theory and conventional psychology for that . . . ;->

Regards,

Ptolemy

[Bill_Powers1]

[From Bill Powers (980308.1620 MST)]

Rick Marken (980308.1030)--

Anyway, one thing that is needed, I think, is a control demo where
the controlled variable is far less obvious than the disturbance
to that variable. This is more like the situation in real life
(and real psychological experiments) where we can see disturbances, d,
(verbal questions, light levels, etc) and outputs, o, (reaction time,
ratings, etc) very clearly but not the controlled variable, p'.
This is quite different from the situation in most of our control
demos, where we can see the controlled variable (distance from
cursor to target, say) and output (handle movements) quite easily,
but not the disturbance.

Here is another cut at the diagram we're working on. It is a description of
a tracking experiment slightly more complex than that ones we've done. The
mouse affects both the target and the cursor, and there are disturbances
applied to the target and cursor as well. This gives us two environmental
feedback functions and two disturbance functions to play with. There is no
"CEV" or "qi" in the environment. Maybe, if you're going to work up an
experiment anyway, you might like to organize it around this diagram.

                              > r = 0
                              >
                     ------->COMP --------> e
                    > >
                    > p = t - c | [CONTROL SYSTEM]
                    > >
   - - - - - - Fi - - - - - - - Fo = leaky integral - -
                  / \ |
                 / \ |
                > > >
          Fd1 | | Fe1 |
     d1-------> t <--- | -------------- |
             TAR>GET | \ |
                > > -m- MOUSE [ENVIRONMENT]
          Fd2 | | Fe2 /
     d2 ------- | ---> c <-------------
                > CUR>SOR
                > >
                 \ /
                  \ /
   - - - - - -Fi - - - - - - - - - - - -
                   >
                   > qi = t - c [OBSERVER]
                   V

Obviously, the reference signal could be set to some nonzero value (keep
the cursor an inch to the right of the target). Fe1, Fe2, Fd1, and Fd2 can
be four constants like 1,2,3 and 4. d1 and d2 are generated independently
of each other.

We know this will work, of course, but it's always best to do the
experiment and show that theory predicts the data once again.

For slow disturbances, we expect (with the functions standing for constant
multipliers) that

Fd1*d1 + Fe1*m - Fd2*d2 - Fe2*m = 0 (i.e., p = 0),

and therefore,

m = (Fd1*d1 - Fd2*d2)/(Fe2 - Fe1)

One detail: if Fe1 > Fe2, then we have to make m be the negative of the
integral to keep the feedback negative.

The net disturbing effect or disturbing signal is

DS = Fd1*d1 - Fd2*d2.

Best,

Bill P.

[bnhpct]

[From

Bill Powers (980308.1620 MST)--

Here is another cut at the diagram we're working on. [...]

                             > r = 0
                             >
                    ------->COMP --------> e
                   > >
                   > p = t - c | [CONTROL SYSTEM]
                   > >
  - - - - - - Fi - - - - - - - Fo = leaky integral - -
                 / \ |
                / \ |
               > > >
         Fd1 | | Fe1 |
    d1-------> t <--- | -------------- |
            TAR>GET | \ |
               > > -m- MOUSE [ENVIRONMENT]
         Fd2 | | Fe2 /
    d2 ------- | ---> c <-------------
               > CUR>SOR
               > >
                \ /
                 \ /
  - - - - - -Fi - - - - - - - - - - - -
                  >
                  > qi = t - c [OBSERVER]
                  V

Ah.

We were leaving out the observer.

So not only is qi or cv or CCEV not in the environment, qi is the
observer's notion of what p is inside the observed control system.

qi is a controlled perception p' inside the observer, as Rick said.

So we've been talking of empathy all along, haven't we. "Mind Readings".

And whatever is going on in the environment is free to be whatever it is,
uncluttered by complex variables mirroring the structures of our perceptual
hierarchies.

No wonder control of perceptual input is disturbed.

Ah.

···

At 05:07 PM 3/8/98 -0700, Bill Powers wrote:

[R_Marken]

[From Rick Marken (980308.2230)]

Me:

I will try to develop such a demo. I hope Bruce Abbott doesn't beat
me to it;-)

Bruce Abbott (980308.1505 EST) --

Not to worry: I'm MUCH too busy attempting to defend my outmoded
beliefs in S-R theory and conventional psychology for that . . . ;->

I'm afraid I was trying goad you into applying your considerable
programming skills to the development of modern psychology (PCT).
I think a great way to learn about and contribute to PCT is by
building programs that demonstrate basic principles of control.
But if you want to spend your time defending S-R psychology (and
it appears that you do) there's nothing I can do about it.

Me:

My point is that the term "controlled variable" (cv) represents the
_observer's_ perception of the perceptual variable (p) that the actor
is controlling. Both the cv and p are perceptual variables.

Bruce Abbott (980308.2235 EST) --

Fine. Why are you belaboring the obvious?

I don't believe I was "belaboring" anything. The fact that qi is not
really an "environmental variable" but a convenient fiction was not
that obvious to me. Was it obvious to you?

It's true; so what?

I think recognition of the fact that qi is not a a "real" environmental
variable helps focus the discussion of the relationship between
conventional and PCT approaches to the study of behavior. In
particular, I think it shows why it makes no sense to write a formula
like qi = o + d and then imagine that the relationships between d's
and o's that are observed in conventional research tell anything about
the variables (qi's) controlled in these experiments. The formula
implies that there is some environmental variable (qi) "out there"
that is a physical function of o and d and that you can "solve
for" qi once you know o and d.

But now we know that qi is a fiction; there are no qi's "out there"
in the environment. qi is just a stand-in for the mapping of
environmental variables, v, into the perceptual variable, p, that
is being controlled by the actor. There is simply no way to know
what function of v's corresponds to the controlled perceptual
variable by observing random relationships between d's and o's. It
can only be done by systematically manipulating many different d's,
which affect different v's, to determine which mapping of v's is
protected from the effects of these disturbances.

Are you arguing for solipsism?

Not quite. It's all perception but it is easy to see that there
are regular constraints on the way we can _control_ those
percpetions. We can still have science though we might have to
change our concept of what science is about (it's about developing
"realities" (models) to explain our perceptions rather than about
inventing perceptions (models) to explain reality).

···

--

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

[Jeff_Vancouver2]

[from Jeff Vancouver 980309.1045 EST]

[From Bill Powers (980308.0133 MST)]

Rick Marken (980307.1250)--

The problem here is that you and Jeff are using "disturbance"
differently.

That is surely an aspect of it. But it still doesn't explain
statements like:

But then what is the physical system that allows o to have an
impact on d?

Jeff may be thinking of d as the perturbation of qi away from its reference
level. With that idea in mind, it makes at least some sense to speak of the
impact of the output on this perturbation (working against it). It's not
quite the right way to analyze the system -- the actual perturbation
reaches its maximum, normally, at about the same time that the output
produces the maximum opposing effect, so the cancellation of disturbance
effects has already happened.

Maybe that's not what Jeff thinks. Perhaps we should ask.

This is what I was thinking, but the lack of clarity was my fault. I see
how Rick came to the conclusion that I did not know what I was talking
about. Furthermore, all the discussion about what I am supposedly
controlling has made me realize that the number of perceptions involved are
too numerous to easily analyze (much less control). However, one
perception that I can control for is "not participating in this thread." I
do believe it is important, but I am not controlling any of those other
perceptions, so another time maybe.

Jeff

A great many people think they are thinking when they are merely
rearranging their prejudices.
                -- William James

[BUSINESS_GAVIN_HERMA]

( Gavin
Ritz 2011.08.26.14.55NZT)

Is mathematics a controlled variable?

Is it the purest of all controlled variables?

Is it the interval (as in spacetime) between the physical world
and our inner mental world controlled variable?

If it is then it’s subject to itself
as a subobject classifier of category theory.

Which means that it’s an
open-paradigm? The subobject classifier is the subobject of a set (in this case
mathematics) which is often called the truth value object.

One can cut the parts of the whole anyway
one chooses and create new ones, mentally that is.

Looks like maybe Gödel incompleteness theorem
holds for PCT too.

See “Why math Works, Scientific
American August 2011”

···

[rsmarken]

[From Rick Marken (2011.08.26.0920)]

Gavin
Ritz (2011.08.26.14.55NZT)–

Is mathematics a controlled variable?

It could be. First you have to define the variable; the word “mathematics” could refer to several variables, for example, subject matter (the values of this variable being things like “number theory”, “calculus” algebra", “applied”, “topology”, etc), type (discrete, vs continuous variable), etc. Once you have defined the variable you can test to see if people control for that variable. I think there is at least anecdotal evidence that people do control for the mathematical subject matter in which they specialize; I know people who have worked to become “applied” mathematicians and resisted disturbances in the form of attempts by other faculty to seduce them into other fields.

Is it the purest of all controlled variables?

Is it the interval (as in spacetime) between the physical world
and our inner mental world controlled variable?

You clearly have no idea what a controlled variable is.

RSM

···

If it is then it’s subject to itself
as a subobject classifier of category theory.

Which means that it’s an
open-paradigm? The subobject classifier is the subobject of a set (in this case
mathematics) which is often called the truth value object.

One can cut the parts of the whole anyway
one chooses and create new ones, mentally that is.

Looks like maybe Gödel incompleteness theorem
holds for PCT too.

See “Why math Works, Scientific
American August 2011”

–
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[BUSINESS_GAVIN_HERMA]

[From Rick Marken (2011.08.26.0920)]

Gavin
Ritz (2011.08.26.14.55NZT)–

Is mathematics a controlled variable?

It could be. First you have to define the variable; the word “mathematics” could refer to several variables, for example, subject matter (the values of this variable being things like “number theory”, “calculus” algebra", “applied”, “topology”, etc), type (discrete, vs continuous variable), etc. Once you have defined the variable you can test to see if people control for that variable. I think there is at least anecdotal evidence that people do control for the mathematical subject matter in which they specialize; I know people who have worked to become “applied” mathematicians and resisted disturbances in the form of attempts by other faculty to seduce them into other fields.

Is it the purest of all controlled variables?

Is it the interval (as in spacetime) between the physical world
and our inner mental world controlled variable?

You clearly have no idea what a controlled variable is.

You are absolutely correct. I have to agree totally with you on this.

RSM

···

If it is then it’s subject to itself
as a subobject classifier of category theory.

Which means that it’s an
open-paradigm? The subobject classifier is the subobject of a set (in this case
mathematics) which is often called the truth value object.

One can cut the parts of the whole anyway
one chooses and create new ones, mentally that is.

Looks like maybe Gödel incompleteness theorem
holds for PCT too.

See “Why math Works, Scientific
American August 2011�

–
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[Bruce_Buchanan1]

[From Bruce Buchanan 940904. 21:25]

Rick Marken (940904.1045) writes:

. . . a "controlling variable" is just a figment of the imagination of
fevered S-R psychology brains. Variables don't control but some
systems control some variables; variables that are controlled are called
controllED variables. . . .

Ultimately, what people control (according to PCT) are perceptual
signals. . . .

Bill Powers (940904.0840 MDT) writes:

A controllING variable is a variable that
controls something else. As variables generally aren't control systems,
what this usually means is a variable that DETERMINES some other
variable, as through a physical connection. A controllED variable, on
the other hand, is a variable that is controlled by some control system;
its state is determined by something other than itself. . . .

. . . In a control system, the only controllING variable is the reference

signal; it controls in the sense that it specifies the value to which the
control system will bring the perceptual signal.

As a relative newcomer to this field I realize that I have a lot to learn,
and I much appreciate the feedback and clarification of possible
mistunderstandings.
However in several previous postings where I used the term ControllING
Variable what I had in mind were reference signals at the highest
hierarchical levels which, as I conceive them, are the main determinants
that set the ControllED Variables in terms of which we assess specific
perceptions e.g. whether we are keeping the car on the road.

I further understand that the ControllING Variables are developed/evolve to
provide guidance as reference signals, i.e. to modulate the appropriateness
of (perhaps more than one) lower level ControllED Variable.

If this is not correct, somebody might let me know - perhaps by email, so
that not everyone has to endure repeated elementary discussions!

In any case it does seem to me that something corresponding to the higher
controlling reference signals, which I also think of as guiding values, are
crucial determinants of control as well as highly problematic in our lives.

Rick writes:

What is the difference between a "real problem" and the problem faced
by a hierarchical control system - which is to get it's perceptual signals
matching the references for those signals[?]

I take the point being made, which is perfectly valid as far as it goes.
However it is also my understanding that the perceptual signals are aspects
which we somehow abstract from the larger "real" situation which,
admittedly unknowable to us in itself (Ding an Sich), nevertheless affects
us.

Another consideration is that part of the problem of the hierarchical
control system is also to adjust the higher level reference signals - or
Controlling Variables (as I have been thinking of them anyway - see above)
- is a continously improving approximation to reality whatever it is.

Anyway, for what its worth, and I'm quite sure its not the last word, those
are my thoughts on the point you make.

On Rick's riddle (of the Sphinx):

Why are socio-economic theories (capitalism, communism, socialism, etc) of
how individuals should organize themselves to produce the "best" social
results like inverse kinematic models of how inidvidual agents produce
intended results?

Perhaps because they tend to (1) misunderstand the essential differences
between needs and relationships at the level of _society_ with those at the
level of _individuals_, (2) get things backwards, starting with objectives
and methods based upon what they already know from the past, and being slow
and inadequate in meeting the many unexpected contingencies which so
importantly go the shape the future.

Needless to say, I am sure there is no "right" answer; at least I hope that
no Sphinx will punish wrong answers! But I look forward to Rick's answer.

Cheers!

Bruce B.

[Abbott_Bruce]

[From Bruce Abbott (981014.1650 EST)]

Rick Marken (981014.1340) --

Bruce Abbott seems to disagree. He seems to think that conventional
psychologists are aware of the existence of controlled variables
(though under another name) and that they deal with this phenomenon
regularly in their research.

Richard, I have not made such a claim. You and Bill seem to be getting your
kicks lately out of attributing to me words and ideas I have neither
expressed nor implied. I don't even _recognize_ these beliefs and
misconceptions I am supposed to hold.

I've asserted that it is possible that at least _some_ psychologists and
biologists understand control theory quite well, but express their knowledge
in words defined differently from their definitions under PCT.

My actual reasoning:
        Assertion: Some understand control theory but express that
                    knowledge using words defined differently.
        Assertion: Ramashandran and Blakeslee use words to describe control
                    that are consistent with those definitions.
        Conclusion: Ramashandran and Blakeslee may understand control theory.

Implication falsely drawn by RM and attributed to me as a belief:
        Assertion: Ramashandran and Blakeslee may understand control theory
                    but express that knowledge using words defined differently.
        Assertion: Ramachandran and Blakeslee are psychologists.
        Conclusion: Psychologists understand control theory but just use
                    different words to express that knowledge.

I've never said or implied anything remotely like the latter. It is not an
implication of my reasoning.

I would just like Bruce A. to point me
to the places in the psychological literature -- preferably the
behavioral research literature -- that deal with controlled variables
(by whatever name they are known).

I _have_ stated that _some_ psychologists do understand control theory quite
well. (This is quite another matter from saying that "conventional
psychologists" -- which you take to be all psychologists save yourself and
perhaps two others -- do.) I don't have the references here with me at the
moment, but I'll supply a few examples when I get home. Meanwhile, do you
remember Ettinger and Staddon (1983)? This is the study in which the
evidence was said to support the view that rats control the rate of
reinforcement on FR schedules. It's a perfectly good control-system
analysis. The only problem was that, as my reanalysis and replication
showed, rats don't do anything of the sort.

Regards,

Bruce

[R_Marken]

[From Rick Marken (981014.1530)]

Me:

Bruce Abbott seems to disagree. He seems to think that conventional
psychologists are aware of the existence of controlled variables

Bruce Abbott (981014.1650 EST)

Richard, I have not made such a claim.

Ok. I hope this means that you agree that conventional psychologists
are _not_ aware of the existence of controlled variables?

I _have_ stated that _some_ psychologists do understand control
theory quite well.

This has nothing to do with whether or not psychologists are aware of
the
existence of controlled variables. I have known a number of
psychologists
who understood control theory in great detail but couldn't tell a
controlled variable from "stimulus" variable if they tried.

Meanwhile, do you remember Ettinger and Staddon (1983)? This is the study
in which the evidence was said to support the view that rats control the
rate of reinforcement on FR schedules. It's a perfectly good control-
system analysis.

Yes. This is an example where psychologists were able to _guess_ that
a variable (rate of reinforcement) might be under control. That one
out of 1,000,000 studies.

The only problem was that, as my reanalysis and replication showed, rats
don't do anything of the sort.

Same thing with the McBeath study (so now it's 2 out of 1,000,000);
they guessed that LOT was under control; so they were able to
hypothesize a controlled variable. Unfortunately, like Ettinger and
Staddon, they didn't really understand the concept of controlled
variable so they were unable to do the kinds of Tests that would
show that LOT is _not_ under control and that some other variable
_is_.

Your analysis of the Ettinger and Staddon data showed that rate of
reinforcement is not under control. What your analysis didn't show is
what _is_ under control.

Best

Rick

···

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

[Abbott_Bruce]

[From Bruce Abbott (981014.1330 EST)]

Rick Marken (981014.1530) --

Bruce Abbott seems to disagree. He seems to think that conventional
psychologists are aware of the existence of controlled variables

Bruce Abbott (981014.1650 EST)

Richard, I have not made such a claim.

Ok. I hope this means that you agree that conventional psychologists
are _not_ aware of the existence of controlled variables?

I _have_ stated that _some_ psychologists do understand control
theory quite well.

Meanwhile, do you remember Ettinger and Staddon (1983)? This is the study
in which the evidence was said to support the view that rats control the
rate of reinforcement on FR schedules. It's a perfectly good control-
system analysis.

Yes. This is an example where psychologists were able to _guess_ that
a variable (rate of reinforcement) might be under control. That one
out of 1,000,000 studies.

Same thing with the McBeath study (so now it's 2 out of 1,000,000);
they guessed that LOT was under control; so they were able to
hypothesize a controlled variable.

I only intended to show that, here and there, some behavioral and biological
scientists have developed elegant control-system models and have conducted
empirical studies to evaluate the models and guide their modification.
That's all I need in order to disprove your assertion that nobody outside
this group has done so. I am aware of a fair number of them. So that I
know what to shoot for, just how many studies will I have to list before you
stop saying that "conventional psychologists" do not understand control? I
thought your assertion was that none do.

Your analysis of the Ettinger and Staddon data showed that rate of
reinforcement is not under control. What your analysis didn't show is
what _is_ under control.

Maybe not, but the data did show that the rat would act to replace a food
pellet when one was not present in the chamber. If the variable,
food-status (which can be present or absent) is disturbed to the absent
state, the rat acts promptly to restore it to the present state, so long as
the rat is hungry and is allowed to eat the pellets so produced. If the rat
had recently consumed a large number of pellets, then it stopped controlling
this variable.

Regards,

Bruce

[Abbott_Bruce]

[From Bruce Abbott (981015.1840 EST)]

Rick Marken (981015.1600) --

My questions to you have nothing to do with whether "some
behavioral and biological scientists have developed elegant
control-system models and have conducted empirical studies
to evaluate the models". I know plenty of behavioral and
biological scientists who can build control models just fine.

Wow, this is news to me -- I thought you were claiming that these folks
don't know anything at all about the phenomenon of control. I thought you
said that (all) "conventional" psychologists and biologists were clueless.

What I have not seen is any evidence that behavioral and biological
scientists have any awareness of the fact that organisms are
controlling perceptual representations of their environment. That
is, they seem to have no awareness of the existence of _controlled
variables_.

Come off it, Rick. You can't understand how control systems operate without
knowing that.

Regards,

Bruce

[R_Marken]

[From Rick Marken (981015.1600)]

Me:

Same thing with the McBeath study (so now it's 2 out of 1,000,000);

Bruce Abbott (981014.1330 EST) --

I only intended to show that, here and there, some behavioral
and biological scientists have developed elegant control-system
models and have conducted empirical studies to evaluate the
models and guide their modification.

McBeath did not have an "elegant control-system model"; he had
no model at all. He did not conduct his study to evaluate a model;
he conducted it to see what the retinal path of the ball would
look like given the assumption that optical acceleration was
controlled. But he had no model of optical acceleration control
so he didn't know what the pattern of movement of the ball
would look like if subjects were controlling optical acceleration.
If he had had such a model, he would have seen that his results
were as consistent with a model that controls optical acceleration
as they are with a model that controls LOT.

My questions to you have nothing to do with whether "some
behavioral and biological scientists have developed elegant
control-system models and have conducted empirical studies
to evaluate the models". I know plenty of behavioral and
biological scientists who can build control models just fine.
What I have not seen is any evidence that behavioral and biological
scientists have any awareness of the fact that organisms are
controlling perceptual representations of their environment. That
is, they seem to have no awareness of the existence of _controlled
variables_.

So let's try again:

1. Do you agree that controlled variables (PCT sense) are
real (observable) phenomena?

2. If so, then what are these variables called in conventional
psychology? In particular, what are these variables called in
textbooks on behavioral research (such as yours)?

Best

Rick

···

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

[R_Marken]

[From Rick Marken (981015.1812)]

Me:

My questions to you have nothing to do with whether "some
behavioral and biological scientists have developed elegant
control-system models and have conducted empirical studies
to evaluate the models". I know plenty of behavioral and
>biological scientists who can build control models just fine.

Bruce Abbott (981015.1840 EST) --

Wow, this is news to me -- I thought you were claiming that
these folks don't know anything at all about the phenomenon
of control.

I am. I have found that knowing how to build a model of control
has nothing to do with whether or not a person understands the
phenomenon of control. My opinion about thia should come as no
surprise to you since I have acknowledged many times on the net
the fact that many people in my field (human factors) pioneered
the application of control models to behavior; they just didn't
apply the models properly so they had no idea that the distance
from cursor to target was a controlled variable -- controlled
relative to a potentially variable reference specification
inside the controller.

Me:

What I have not seen is any evidence that behavioral and biological
scientists have any awareness of the fact that organisms are
controlling perceptual representations of their environment. That
is, they seem to have no awareness of the existence of _controlled
variables_.

Bruce:

Come off it, Rick. You can't understand how control systems
operate without knowing that.

I'm not trying to be difficult, Bruce. As I said, I'm writing a
paper about controlled variables for an audience of conventional
psychologists. I want to know what my audience already knows
about controlled variables. So I'm trying to use you as my
informant. In order to be able to do this I have to know whether
or not you have the same understanding of the term "controlled
vriable" as I do. Thus I want to know:

1. Do you agree that controlled variables (PCT sense) are
real (observable) phenomena?

If you agree that controlled variables are observable variables
in the environment then you can be a useful informant because
you know what I know -- that there are controlled variables to
be observed. Then you can tell me what I really want to know:

2.... what are these variables called in conventional psychology?
In particular, what are these variables called in textbooks on
behavioral research (such as yours)?

Won't you please answer these simple questions?

Thanks

Rick

···

--

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