[From Bill Powers (981013.1004 MDT)]
Hank Folson 981013.0800--
What I think Carver calls 'behavior' is the output in a PCT model. If so,
Carver says the output of a PCT loop is the 'goal' of the system. Anyone
care to explain how that works?
I'm eagerly awaiting any "responses" to Hank's question-stimulus.
Best,
Bill P.
[From Bruce Gregory (981013.1215 EDT)]
Bill Powers (981013.1004 MDT)
Hank Folson 981013.0800--
>What I think Carver calls 'behavior' is the output in a PCT model. If so,
>Carver says the output of a PCT loop is the 'goal' of the system. Anyone
>care to explain how that works?I'm eagerly awaiting any "responses" to Hank's question-stimulus.
Imagine you are about to serve a tennis ball. Your goal is to (1) toss the
ball into air (behavior) to a predetermined height, and (2) hit the tennis
ball with your racket (behavior) causing it to land in bounds near the spot
you have selected. The mechanism by which this sequence of actions occurs
can be modeled using PCT. It is not always a bad idea to distinguish between
_what_ you want to do and _how_ you are able to accomplish this. I predict
with a high degree of confidence that my answer will not satisfy Rick, since
it does not contain the words "controlled variable."
Bruce Gregory
[From Rick Marken (981013.0930)]
Hank Folson (981013.0800) --
What I think Carver calls 'behavior' is the output in a PCT model.
If so, Carver says the output of a PCT loop is the 'goal' of the
system. Anyone care to explain how that works?
Bill Powers (981013.1004 MDT) --
I'm eagerly awaiting any "responses" to Hank's question-stimulus.
Bruce Abbott (981012.1240 EST) already answered Hank's question.
What Carver calls "behavior" is what we call the "controlled
variable" in PCT. That's why Carver et al. are correct when they
say that "behavior patterns" are controlled.
Hank is just trying to do what I did: make a "big deal" about the
difference between what Carver and other conventional psychologists
say and PCT's "control of perception". As Bruce explained to me,
such distinctions are just "much ado about nothing".
What I am eagerly awaiting is Bruce Abbott's response to my
[Rick Marken (981012.1950)] request for "references to _controlled
variables_ in the psychological literature".
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[from Jeff Vancouver 981013.1200 EST]
[From Hank Folson 981013.0800]
What I think Carver calls 'behavior' is the output in a PCT model. If so,
Carver says the output of a PCT loop is the 'goal' of the system. Anyone
care to explain how that works?
I think that Bruce A's post on the engineering termology of control systems
is very instructive there.
I should say that my reading of Carver is that he has the basic control
process down, but that he differs on some details and language. The
details and language get him into big trouble here on this net, but that is
an issue of missing the forest for the trees. Hence, I took issue with
three of the things Carver said at his APA talk. I argued with him
afterwards about one of them (I did not want to make a pest of myself) and
got him to see it my way (at least that is how I perceived it). But I
think the differences should be conceived of as differing hypotheses
regarding specific phenomenon (e.g., are systems that are off, off because
gain is 0, output is fed directly to input, or perception fed to goal).
Sincerely,
Jeff
[From Hank Folson 981013.0800]
The Carver thread brings back warm memories of my short-lived "Behavior:
the Phlogiston of PCT" thread.
From Carver's web page:
"This new book presents a thorough overview of a model of human
functioning based on the idea that behavior is goal-directed and
regulated by feedback control processes."
My reading is that the 'behavior' referred to here is the observable
'behavior' of classic psychology, not Bill Power's "control of
perception" definition of 'behavior'. Yes? No?
What I think Carver calls 'behavior' is the output in a PCT model. If so,
Carver says the output of a PCT loop is the 'goal' of the system. Anyone
care to explain how that works?
Sincerely,
Hank Folson
704 ELVIRA AVE. REDONDO BEACH CA 90277
Phone: 310-540-1552 Fax: 310-361-8202 Web Site: www.henryjames.com
[From Rick Marken (981013.1015)]
Bruce Gregory (981013.1215 EDT) --
I predict with a high degree of confidence that my answer will
not satisfy Rick, since it does not contain the words "controlled
variable."
Your prediction is partly correct; your answer did not satisfy
me much. But that's not because it did not contain the words
"controlled variable". It's because it was sloppy and vague.
The main problem was that you used the word "behavior" to describe
both actions (toss) and the intended results of those actions
(ball at certain height above the shoulder). Actually, you did
what Carver and Abbott seem to be doing; trying to make believe
you are talking about control when you are not.
I think we could stop these silly arguments if people would
just say "controlled variable" when that's what they are
talking about. Why keep trying to read an understanding of
control into people's writings? Wouldn't it be easier to
just say that two of the variables people control when they
play tennis are the height of ball above their shoulder and
the location of the ball in the service court? And that
these variables are controlled by variable actions such as
the upward force exerted on the ball and the angle of the
the racquet face when hitting the ball? If (as you and
Bruce Abbott seem to think) psychologists like Carver know
all about control, then why don't they talk about it in nice
clear English instead of in (as Dag says) "mush".
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Bruce Gregory (981013.1337 EDT)]
Rick Marken (981013.1015)
Bruce Gregory (981013.1215 EDT) --
> I predict with a high degree of confidence that my answer will
> not satisfy Rick, since it does not contain the words "controlled
> variable."Your prediction is partly correct; your answer did not satisfy
me much. But that's not because it did not contain the words
"controlled variable". It's because it was sloppy and vague.
Flattery will get you nowhere.
The main problem was that you used the word "behavior" to describe
both actions (toss) and the intended results of those actions
(ball at certain height above the shoulder).
What's sloppy and vague about that? Behavior _does_ consist of actions and
achieving the intended results of those actions. At least everywhere else in
the world but on CSGnet. Doesn't "pitching" involve having the ball go where
you want it to? Ditto for "hitting"?
Actually, you did
what Carver and Abbott seem to be doing; trying to make believe
you are talking about control when you are not.
Good to know that I am not alone.
I think we could stop these silly arguments if people would
just say "controlled variable" when that's what they are
talking about. Why keep trying to read an understanding of
control into people's writings? Wouldn't it be easier to
just say that two of the variables people control when they
play tennis are the height of ball above their shoulder and
the location of the ball in the service court? And that
these variables are controlled by variable actions such as
the upward force exerted on the ball and the angle of the
the racquet face when hitting the ball? If (as you and
Bruce Abbott seem to think) psychologists like Carver know
all about control, then why don't they talk about it in nice
clear English instead of in (as Dag says) "mush".
Reminds me of the observation that if only the Jews and Arabs would put
aside their difference and get along like good Christians there would be no
problems in the Middle East. Saying that people "control variables" is
PCTspeak. If you speak the language, the expression conveys a great deal
with economy. If you don't it sounds like mush. Some people think they are
playing tennis, not controlling perceptual variables. The latter is part of
a model that attempts to explain how people are able to play tennis.
Bruce Gregory
[From Bill Powers (981013.1244 MDT)]
Bruce Gregory (981013.1215 EDT)--
Imagine you are about to serve a tennis ball. Your goal is to (1) toss the
ball into air (behavior) to a predetermined height, and (2) hit the tennis
ball with your racket (behavior) causing it to land in bounds near the spot
you have selected. The mechanism by which this sequence of actions occurs
can be modeled using PCT. It is not always a bad idea to distinguish between
_what_ you want to do and _how_ you are able to accomplish this. I predict
with a high degree of confidence that my answer will not satisfy Rick, since
it does not contain the words "controlled variable."
That's the conventional analysis which ignores the fact that you know about
these things only in terms of what your perceptions present to you.
I would say that you control for seeing the ball rise into the air by a
certain amount, and then for seeing and feeling your racket contacting the
ball at some perceived speed and in some perceived direction. Your means
for controlling these perceptions consists of creating a set of kinesthetic
perceptions representing efforts, joint angles, and skin pressures,
especially in the hand holding the racket but also in all the rest of your
body.
But then, maybe there's nothing to this PCT fantasy, and we can ignore it
whenever it's easier to talk the other way.
Best,
Bill P.
[From Bill Powers (981013.1253 MDT)]
Bruce Gregory (981013.1337 EDT)--
Rick:
The main problem was that you used the word "behavior" to describe
both actions (toss) and the intended results of those actions
(ball at certain height above the shoulder).
Bruce G:
What's sloppy and vague about that? Behavior _does_ consist of actions and
achieving the intended results of those actions. At least everywhere else in
the world but on CSGnet.
Do you call the act of putting food in your mouth and chewing it
"digestion?" Do you call the act of opening the car door "starting the
car?" Do you call buying popcorn "seeing a movie?" Do you call sharpening a
pencil "writing a letter?"
If you see generating the muscle forces that accelerate a ball upward as
being the same thing as seeing a ball describe a trajectory into the air,
then I suppose there's really no point in arguing that they're different.
Even if they are different you can't see the difference, so you wouldn't
understand the explanation. Apparently.
Doesn't "pitching" involve having the ball go where
you want it to? Ditto for "hitting"?
What does "involve" mean? Isn't this just a loose and sloppy term that lets
you say two things have something to do with each other without having to
understand WHAT?
Actually, you did
what Carver and Abbott seem to be doing; trying to make believe
you are talking about control when you are not.Good to know that I am not alone.
No, you should be ashamed. If you want to praise sloppy speaking and
thinking, why not do it where it will be appreciated?
Saying that people "control variables" is
PCTspeak. If you speak the language, the expression conveys a great deal
with economy. If you don't it sounds like mush. Some people think they are
playing tennis, not controlling perceptual variables. The latter is part of
a model that attempts to explain how people are able to play tennis.
So if you just want to play tennis, shut up and play. Stop trying to
explain it, because you're not doing very well at that part.
Best,
Bill P.
[From Rick Marken (981013.1215)]
Bruce Gregory (981013.1337 EDT) --
Saying that people "control variables" is PCTspeak.
No. It's ordinary English speak. How else would you say 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
[From Bruce Gregory (981013.1642 EDT)]
That's the conventional analysis which ignores the fact that you
know about
these things only in terms of what your perceptions present to you.
This is true of physics and astronomy as well, needless to say. Should they
be criticized for failing to talk about controlled variables when describing
an experiment or observation?
I would say that you control for seeing the ball rise into the air by a
certain amount, and then for seeing and feeling your racket contacting the
ball at some perceived speed and in some perceived direction. Your means
for controlling these perceptions consists of creating a set of
kinesthetic
perceptions representing efforts, joint angles, and skin pressures,
especially in the hand holding the racket but also in all the rest of your
body.
Sounds like the beginnings of a good model. It seems to me that you are on
the right track.
But then, maybe there's nothing to this PCT fantasy, and we can ignore it
whenever it's easier to talk the other way.
It _is_ a PCT fantasy, in the same sense that classical physics and quantum
mechanics are fantasies. (We normally refer to them as models, but I can
accept your terminology.)
Bruce Gregory
[From Bruce Gregory (981013.1646 EDT)]
Rick Marken (981013.1215)
Bruce Gregory (981013.1337 EDT) --
> Saying that people "control variables" is PCTspeak.
No. It's ordinary English speak. How else would you say it?
Could you direct me to some examples where people who are not familiar with
PCT use the term "controlled variables"? Thank you.
Bruce Gregory
[From Bruce Gregory 9981013.1655 EDT)]
Bill Powers (981013.1253 MDT)
No, you should be ashamed. If you want to praise sloppy speaking and
thinking, why not do it where it will be appreciated?
So if you just want to play tennis, shut up and play. Stop trying to
explain it, because you're not doing very well at that part.
Hard to miss your message. Just recall who started this exchange with a
request. I learned many years ago not to ask questions that I did not want
to hear the answers to.
Bruce Gregory
[From Bruce Abbott (981013.1705 EST)]
Rick Marken (981013.1430) --
But that's not the point. It's not frequent usage that makes saying
"people control variables" ordinary English. It's the fact that
the ordinary meaning of these words communicates a concept (of
a controlled variable) that other words (like "behavior" or "action"
or "dependent variable" or "behavior pattern") don't.
And the number 1 ordinary meaning of the word "control" is . . . .
to exercise restraint or direction over; dominate, command
[from the American College Dictionary]
Engineering definition of "control":
When the value of a dependent variable Y is determined uniquely by
an independent variable X, the behavior of Y may be said to be
"controlled" by X. [from McFarland, 1972]
One example: Under constant load and source power, the speed of a fan is
controlled (uniquely determined by) the setting of the fan-speed control.
(That's why it's called a "control.") The setting of the control exercises
restraint or direction over the speed of the fan.
Under this definition of "control" it is possible to distinguish two forms
of control: open-loop and closed-loop. "Control" as commonly used may refer
to either. A "controlled variable" would be one whose value (e.g., motor
speed) is uniquely determined by the value of another variable (fan speed
setting), and this meaning is used _all the time_ in behavioral science
(e.g., "stimulus control").
So much for ordinary meaning.
Regards,
Bruce
[From Rick Marken (981013.1430)]
Bruce Gregory (981013.1337 EDT) --
Saying that people "control variables" is PCTspeak.
Me:
No. It's ordinary English speak. How else would you say it?
Bruce Gregory (981013.1646 EDT)--
Could you direct me to some examples where people who are not
familiar with PCT use the term "controlled variables"? Thank you.
Take a look at p. 689 of a paper by Tresilian in Quarterly Journal
of Experimental Psychology, 1995, 48A, 688 - 715. So far, that's
the only use of the term that I've been able to find; and he used
the term correctly; he was talking about a variable (optical
acceleration) that is kept under control (maintained at some value).
But that's not the point. It's not frequent usage that makes saying
"people control variables" ordinary English. It's the fact that
the ordinary meaning of these words communicates a concept (of
a controlled variable) that other words (like "behavior" or "action"
or "dependent variable" or "behavior pattern") don't.
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Bruce Abbott (981013.1815 EST)]
Rick Marken (981013.1550) --
Bruce Abbott (981013.1705 EST)
And the number 1 ordinary meaning of the word "control" is . . . .
to exercise restraint or direction over; dominate, command
...
So much for ordinary meaning.
So what's your point?
You claimed that the ordinary, everyday meaning of "controlled variable" is
precisely the PCT meaning of the term. (Have you forgotten your own
argument?) My point is that the ordinary meaning of the term is looser than
the PCT meaning and would include the meaning as used by psychologists when
they say things like "the rate of responding is controlled by the
discriminative stimulus." In that sense, rate of responding is a controlled
variable whether or not rate of responding is controlled in the PCT sense.
Are you saying that, because the word
"control" has several different connotations, psychologists
(yourself included) have been wise to avoid talking about
controlled variables?
No. Are you? (I didn't say anything remotely like that. Where do you _get_
this stuff?)
Regards,
Bruce
[From Rick Marken (981013.1550)]
Bruce Abbott (981013.1705 EST)--
And the number 1 ordinary meaning of the word "control" is . . . .
to exercise restraint or direction over; dominate, command
...
So much for ordinary meaning.
So what's your point? Are you saying that, because the word
"control" has several different connotations, psychologists
(yourself included) have been wise to avoid talking about
controlled variables? Is this why you don't discuss controlled
variables in your text? What do you call controlled variables
in your text?
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken
[From Rick Marken (981013.1800)]
Bruce Abbott (981013.1815 EST)
You claimed that the ordinary, everyday meaning of "controlled
variable" is precisely the PCT meaning of the term.
OK. Let's just say that the term "controlled variable" is at least
as good a way of describing a variable that is under control
as describing it as "behavior" or a "behavior pattern".
My point is that the ordinary meaning of the term is looser than
the PCT meaning and would include the meaning as used by
psychologists when they say things like "the rate of responding
is controlled by the discriminative stimulus."
I find that hard to believe. Even if the ordinary meaning of
"control" is "determine", an environmental variable whose value
is controlled (determined) by the organism's responses can still
be distinguished (verbally) from a response variable (rate of
responding) that is controlled (determined) by a stimulus variable.
In that sense, rate of responding is a controlled variable whether
or not rate of responding is controlled in the PCT sense.
In PCT, the term "controlled variable" refers to a particular kind
of variable; one whose value is brought to and maintained in
specified states by the behaving organism. So the "PCT sense" of
the term "controlled variable" _is_ this phenomenon; "controlled
variable" points to these kinds of variables, just as the term
"independent variable" points to a variable you can manipulate
(control) and "dependent variable" points to a variable whose
value is presumably determined by variations in the independent
variable.
So controlled variables (PCT sense) are _real_ phenomena. Their
existence is easily demonstrated and of fundamental significance
to an understanding of behavior.
Do you agree that controlled variables (PCT sense) are real
phenomena?
If so, then what are these variables called in conventional
psychology? What are they called in your textbook on behavioral
reseach?
Best
Rick
--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/
[From Bill Powers (981013.2046 MDT)]
Bruce Abbott (981013.1705 EST)--
One example: Under constant load and source power, the speed of a fan is
controlled (uniquely determined by) the setting of the fan-speed control.
(That's why it's called a "control.") The setting of the control exercises
restraint or direction over the speed of the fan.
Under this definition of "control" it is possible to distinguish two forms
of control: open-loop and closed-loop. "Control" as commonly used may refer
to either. A "controlled variable" would be one whose value (e.g., motor
speed) is uniquely determined by the value of another variable (fan speed
setting), and this meaning is used _all the time_ in behavioral science
(e.g., "stimulus control").So much for ordinary meaning.
I think you're missing Rick's point, which is that "controlled variable",
in ordinary English, is a way of referring to a variable that is under
control by the system in question. But this term is extraordinarily sparse
in the psychological literature, _whatever_ meaning you want to give to the
term "control." In some branches of psychology, the reason for this
omission may well be a refusal to think that an organism actually controls
variables in its environment, rather than being controlled by them. In
others the reason may simply be lack of acquaintance with engineering
terminology. And in still others, it is failure to hear the difference
between "control variable" (a variable that controls something else) and
"controlled variable" ( a variable controlled BY something else). Rick and
I have both had reviewers misquote us as saying "control variable".
Unfortunately there is no way to appeal to an etymological defense of any
proposed meaning for "control." The term was borrowed from a French
bookkeeping term with only a vague connection to any actual control
phenomenon as we think of them in PCT. I notice that in your definition
above, you have to resort to the usual _ceteris paribus_ disclaimer: "Under
constant load and source power" the fan speed control regulates the speed
of the fan. This is what most people unacquainted with feedback systems say
when talking about control, because they can't conceive of any way for a
speed controller to control speed when _neither the load nor the source
power remains constant_. That kind of "control" remains a mystery to them.
PCT is about the kind of control that can go on working under a very
considerable relaxation of the "ceteris paribus" disclaimer. It's this
ability to maintain accurate control when disturbances are NOT prevented
that makes the PCT model unique among other models. The problem is that
when we PCTers refer to control, we always mean the kind that can work when
all else is NOT held equal, but since our listeners often use the word
control as you do above, they automatically assume that we're speaking of
ordinary causation.
We have made repeated efforts -- at least I have -- to find some way of
referring to PCT-type control that specifically excludes the other meanings
of the word. As long as people continue to hear PCT-control as if it refers
to simple lineal causation, they won't bother to check further and find
that we're referring to a phenomenon with which they are unfamiliar. Every
time I propose inventing a new word that can't possibly be confused with
the kind of control that is just lineal causation, I am reassured that we
don't need to do this, because people are capable of understanding the
difference. But the sad fact is that they are not. And this failure to
understand the difference is helped along when others insist that control
comes in two flavors: open-loop and closed-loop, so either way, we're
talking about control. Well, that may be so, but we're not talking about
maintaining variables in predetermined states when all else that can affect
those variables is NOT held constant. And that is the only kind of control
of any interest in PCT. It is also, I maintain, the only kind of control
that matches the way organisms actually work.
I am obviously in the minority when it comes to word usage, so my real
desire, which is to change what people mean when they say control and get
them to use other terms like "influence" and "determine" for other effects,
is doomed. Not wishing to go around again in this futile argument, I
therefore propose once again that we invent a new term, meaning "to bring a
variable to a predetermined state and maintain it there when other
influences on that variable are neither negligible nor held constant." If
we can find such a term that doesn't feel ugly to say or sound ugly and
contrived to hear, I will be perfectly willing to give up the use of the
term "control" and abdicate my position as a "control" theorist.
Stabilating? Regufying? Fixication? Purpofaction? Oh, God, here we go again.
Best,
Bill P.
[From Hank Folson (981013.2230)]
Bruce Gregory (981013.1215 EDT)
Imagine you are about to serve a tennis ball. Your goal is to (1) toss the
ball into air (behavior) to a predetermined height, and (2) hit the tennis
ball with your racket (behavior) causing it to land in bounds near the spot
you have selected. The mechanism by which this sequence of actions occurs
can be modeled using PCT.
Wouldn't my immediate goal (1) be to perceive the ball over my head
within some volume of space such that I know from experience that I can
hit it with my racket across the net? This PCT approved goal is internal
to my system, and has something to do with reference levels for where the
ball can be perceived as hittable. If so, tossing the ball in the air is
not a goal, it is just an _output_ intended to put the ball into a
position perceived as hittable. And, if so, what does this say about
Carver's idea?
As to goal (2), the position of the tossed ball generates an error signal
when it is outside of the hittable zone. When the ball falls into the
hittable zone, goal (1) has been achieved. Now goal (2) is that when the
ball is in a hittable position, I want it to begin flying towards
becoming an service ace or whatever. This PCT approved goal is internal
to my system. It certainly isn't an output, and it certainly isn't
'behavior'. And, if so, what does this say about Carver's idea?
It is not always a bad idea to distinguish between
_what_ you want to do and _how_ you are able to accomplish this.
Actually, it is essential if we are to understand PCT. One, or both of
us, is wrong about this tennis example :-).
PCT says our goals are internal, and we use _varying_ outputs (What
everyone else calls 'behaviors'.) to achieve our goals. Other theories
(including Carver?) say that our 'behaviors' are our goals. The question
that raises is this: If our outputs ('behaviors') are our goals, what
means can we possibly we have to achieve these goals???
Sincerely,
Hank Folson
704 ELVIRA AVE. REDONDO BEACH CA 90277
Phone: 310-540-1552 Fax: 310-361-8202 Web Site: www.henryjames.com