The controlled quantity (q.i) is data, the perceptual signal (p) is theory

[Rick Marken 2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:Â

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective. I think this is the only way to see things if one is involved in doing research on PCT. Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me. Â

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p, from the observer’s perspective is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT: that it is a theory that explains the fact of control as it is seen in the behavior of living systems.Â

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*. Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance, are the data on which we base our conclusion that the behavior we see involves control. As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.Â

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*. The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.Â

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances) that motivates the theory that says that this observation is a result of control of perception. The controlled variable, q.i, is data. PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled. The perceptual signal, p, is theory.Â

RM: So the controlled variable, q.i, is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory). The idea that this is not the case is a misunderstanding of PCT so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.Â

BestÂ

Rick

···


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Fred Nickols (2018.05.29.0635 ET)]

Rick:

I understand that p is an analog of q.i. To control p we have to affect q.i. in ways such that p = r. An observer can see me affecting q.i. and say that I am controlling q.i. I can see me affecting q.i. and believe I am controlling q.i. Regardless of whose perspective you take, q.i. is not p and p is not q.i. p is an analog of q.i. q.i. is often a physical variable and p is a mental/biophysical signal. How can p = q.i.?

Fred

···

From: Richard Marken (rsmarken@gmail.com via csgnet Mailing List) csgnet@lists.illinois.edu
Sent: Tuesday, May 29, 2018 12:56 AM
To: csgnet@lists.illinois.edu
Cc: Richard Marken rsmarken@gmail.com
Subject: The controlled quantity (q.i) is data, the perceptual signal § is theory

[Rick Marken 2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective. I think this is the only way to see things if one is involved in doing research on PCT. Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me.

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p, from the observer’s perspective is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT: that it is a theory that explains the fact of control as it is seen in the behavior of living systems.

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*. Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance, are the data on which we base our conclusion that the behavior we see involves control. As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*. The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances) that motivates the theory that says that this observation is a result of control of perception. The controlled variable, q.i, is data. PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled. The perceptual signal, p, is theory.

RM: So the controlled variable, q.i, is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory). The idea that this is not the case is a misunderstanding of PCT so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.

Best

Rick

Richard S. Marken

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
–Antoine de Saint-Exupery

[Martin Taylor 2018.05.29.09.09]

A single variable such as q.i can't mean two different things at the

same time in normal science. You have to choose whether to use it
for one or the other of the two different things. In the case of
q.i, some use it as a variable in a control loop. But in this quote
(and frequently elsewhere) Rick uses it to refer to a perception in
a person who hypothesises that it is a perception of some external
variable that is controlled. These two are not the same variable,
and have the same value only by coincidence.
We are again dealing with the uncertain relationship between
perception and real reality, but that problem can be finessed by
making the assumption that the organism doing the controlling “C”
exists, the environment visible to C exists, and that the
observer/experimenter/theorist “T” can generate perceptions of the
environment visible to T. C has a particular view of the
environment, and T has made an effort to align T’s view with C’s
view, so let’s make the idealistic assumption (which can never be
realized in practice) that T’s view is identical to C’s view.
Now T performs a “Test for the Controlled Variable”, and
hypothesizes a range of possible variables C might be controlling.
Some of them clearly fail the test, while others may succeed, some
better, some not quite so well. If T’s hypotheses really and truly
include the variable C is controlling, then that one will be the one
that best matches the test criteria. Call it X. But there will be an
infinite set of other possibilities that vary in ways highly
correlated with X, such as Y, and if one of those had been included
in the set of T’s hypothesized possibilities while X was excluded, T
would see that Y was being well controlled. For example, X might be
a+b, whereas T had included only unit
power values and had tested Y, which was a+b. X might be “the red
car’s front bumper location” while Y was “the red car’s rear wheel
location proportionate to the distances of the car from the viewing
frame”.
No controller can control perfectly, so even if T had tested X, the
idealized requirements of the test such as the trio of correlations
among the disturbance influence, the output influence and the
perceptual input variable would not have been matched exactly. If T
incorrectly hypothesizes Y as a possibly controlled variable, T
finds the same result, good but not perfect control. How can T
distinguish whether the observed imperfection of control exists
because C is not controlling perfectly or because C is controlling a
variable correlated with, but not identical to the one T has found
to be best controlled among the ones theorized?
There are two quite distinct variables here, even if both are in the
environment common to C and to T. One is the variable T has
determined using the best available testing methods to be the one C
appears to be controlling, and the other is the one C is actually
controlling. T can never be sure that they are the same variable,
even under these idealized conditions that I set up to give T every
possible advantage. So which of these two variables should we call q.i? The Powers
diagrams and discussions say that q.i is the variable C controls
(theoretically by keeping a perception of it as near its reference
variable as C is able to do). Rick says it is the variable T has
determined by testing various hypotheses to be the external variable
C is most likely to be controlling. They aren’t the same variable,
so we ought to agree on which one to call q.i in our discussions. Is
it the Powers version or the Marken version that should prevail, and
what should we call the other?
Martin

···

On 2018/05/29 12:56 AM, Richard Marken
( via csgnet Mailing List) wrote:

rsmarken@gmail.com

          [Rick Marken

2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

          RM...

there seems to be general agreement that I am completely
wrong to think that q.i is the controlled perception, p,
from the observer’s perspective. I think this is the only
way to see things if one is involved in doing research on
PCT. Since most everyone on CSGNet is not involved in
doing PCT research I suppose the only problem with not
seeing that q.i as p from the observer’s perspective is
that you’ll keep getting into useless verbal arguments
about it with me.

        RM: Upon

reflection I realized that failure to understand that the
controlled quantity, q.i, is the controlled perception, p,
from the observer’s perspective is not just a problem for
researchers. If you don’t understand this, then you don’t
understand what is fundamental about PCT: that it is a
theory that explains the fact of control as it is seen in
the behavior of living systems.

0.951.05

[Fred Nickols (2018.05.29.1025 ET)]

I get the distinction you are driving at, Rick. I think “theoretical� vs “empirical� is a better way of labeling it than “theory� vs “data.�

Fred Nickols

···

From: Richard Marken (rsmarken@gmail.com via csgnet Mailing List) csgnet@lists.illinois.edu
Sent: Tuesday, May 29, 2018 12:56 AM
To: csgnet@lists.illinois.edu
Cc: Richard Marken rsmarken@gmail.com
Subject: The controlled quantity (q.i) is data, the perceptual signal § is theory

[Rick Marken 2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective. I think this is the only way to see things if one is involved in doing research on PCT. Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me.

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p, from the observer’s perspective is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT: that it is a theory that explains the fact of control as it is seen in the behavior of living systems.

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*. Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance, are the data on which we base our conclusion that the behavior we see involves control. As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*. The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances) that motivates the theory that says that this observation is a result of control of perception. The controlled variable, q.i, is data. PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled. The perceptual signal, p, is theory.

RM: So the controlled variable, q.i, is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory). The idea that this is not the case is a misunderstanding of PCT so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.

Best

Rick

Richard S. Marken

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
–Antoine de Saint-Exupery

[From Rupert Young (2018.05.29 15.50)]

What makes you think people (on csgnet) don't understand this?

Regards,
Rupert

···

( Rick Marken
2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

          RM...

there seems to be general agreement that I am completely
wrong to think that q.i is the controlled perception, p,
from the observer’s perspective. I think this is the only
way to see things if one is involved in doing research on
PCT. Since most everyone on CSGNet is not involved in
doing PCT research I suppose the only problem with not
seeing that q.i as p from the observer’s perspective is
that you’ll keep getting into useless verbal arguments
about it with me.

        RM: Upon

reflection I realized that failure to understand that the
controlled quantity, q.i, is the controlled perception, p,
from the observer’s perspective is not just a problem for
researchers. If you don’t understand this, then you don’t
understand what is fundamental about PCT: that it is a
theory that explains the fact of control as it is seen in
the behavior of living systems.

Rick what is too much is too much. You are seriously damaging PCT.

You don’t understand it and beside that you are making your own constructs which has no connection to PCT.

image002109.jpg

image002103.png

···

From: Richard Marken (rsmarken@gmail.com via csgnet Mailing List) csgnet@lists.illinois.edu
Sent: Tuesday, May 29, 2018 6:56 AM
To: csgnet@lists.illinois.edu
Cc: Richard Marken rsmarken@gmail.com
Subject: The controlled quantity (q.i) is data, the perceptual signal § is theory

[Rick Marken 2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective.

HB : Right. You are wrong. We proved to you with Bills’ diagrams (generic and “Rubber band”). In both cases “q.i.” is just “input qauntity” which consist of added effects of “q.o”. and “d” and enters “sensor function”. Do you agree with Bills’ diagrams or not ? Because it is obviously that you don’t agree with Bill as you are making your own theory RCT.

RM : I think this is the only way to see things if one is involved in doing research on PCT.

HB : First of all you are not doing any PCT research, You are constructing “tests” and “demos” which suit your imagination. With “tests” and “demos” you confirm what you want to confirm. It has nothing to do with scientific research. You are living in your RCT world Rick. You are no scientific researcher. You are just a dreamer.

RM : Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me.

HB : We are involved in PCT research but you are not. From any point of observing “q.i.” is “input quantity” to any LCS which is transformed into perceptual signal.

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p,

HB : There is no “controlled perception” in PCT. I have to ask you once again. Do you agree with Bills’ PCT ???

"Input quantity is q.i., controlled quantity is ….
<

Bill P (B:CP) : Consider once again the meaning of the term controlled quantity. A controlled quantity is controlled only because it is detected by a control system, compared with a reference, and affected by outputs based on the error thus detected. The controlled quantity is defined strictly by the behaving system’s perceptual computers; it may or may not be identifiable as an objective (need I put in quotes?) property of, or entity in, the physical environment. In general an observer will not, therefore, be able to see what a control system is controlling

HB : The controlled quantity is defined strictly by the behaving system’s perceptual computers; Â Do you understand this language ? Bill wrote it. Your teacher. “Q.I.” is not defined “strictly by perceptual computers”. Perception is. “Q.I.” is outside organism (see diagram). Can somebody read to Rick so that he will understand what is written ?

cid:image002.png@01D3F75A.9DF6C970

HB : Are you blind Rick ? You wear thick spectacles ? “Q.I.” is not and can’t be “controlled quantity” in PCT. I’ll ask you again. Do you agree with Bills definition of “Controlled quantiy”. We aew talking about PCT. Why don’t you use Bills’ tools for promoting PCT. You are promoting RCT (Ricks’ Control Theory).

It’s just input to “input function” which produces “perceptual signal” the “controlled quantity”. We read “q.i. diagram” and in your head there is no perceptual signal with any trace of “q.i. diagram”. Nothing. Empty ???

RM : …from the observer’s perspecttive is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT:

HB : You don’t understand anything anyway when you are in RCT mode.

RM : ……that it is a theory that explains the fact of control as it is seen in the behavior of living systems.

HB : PCT (Perceptual Control Theory) is general theory about how organisms function. You have to research such a phenomenon from tha aspect of many sciences and not from chair in your home behind computer and playing with joystick.

Bill P (B:CP):

CONTROL : Achievement and maintenance of a preselected state in the controlling system, through actions on the environment that also cancel the effects of disturbances.

And just in case I’ll add also definition of PCT control from Tim Carey :

TC (2014) :

According to PCT, control is a process of acting to bring a perceived aspect of the world into a match with a mental specification for the state of that perception

But RCT (Ricks’ Control Theory) explains control as it is seen in the behavior of living systems.

RCT (Ricks Control Theory) definition of control loop

CONTROL : Keeping of some »aspect of outer environment« in reference state, protected (defended) from disturbances.

HB : Do you understand the difference ???

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*.

HB : With what are you keeping “controlled quantity” or q.i. in reference state ??? With “Control of behavior” ??? Or you use Telekinesis ? How do you see that when person is sleeping or observing rock ?

Again, why don’t you use Bill’s definition. Why inventing your own wrong ?

Bill P (B:CP) : Consider once again the meaning of the term controlled quantity. A controlled quantity is controlled only because it is detected by a control system, compared with a reference, and affected by outputs based on the error thus detected. The controlled quantity is defined strictly by the behaving system’s perceptual computers; it may or may not be identifiable as an objective (need I put in quotes?) property of, or entity in, the physical environment. In general an observer will not, therefore, be able to see what a control system is controlling

RM : Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance,

HB : Ha,ha,ha… q.ii. is “protected from disturbances”. Then nothing influence organism. It’s protected against disturbances. What if somebody is shooting on you ? You are probably not afraid as you are “protected from disturbances”. That’s what control is for you. Protection from dosturbances.

RM : ….are the data on which we base our conclusion that the behavior we see involves control.

HB : Because there is no such thing as “Behavior imvolve control” in PCT we would need some evidences that we could see how “behavior involve control”.Maybe you could offer some physiological evidences like Bill did that “Behavior is not control” and that it doesn’t involve any control ???

You are against Bills’ PCT. You are damage maker. Show us where you thnik Bill was wrong if you think that you have to make your own theory.

RM : As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.

HB : So what ? How is this connected to “behavior involve control” ? The heart of the problem to which PCT “contol theory” adress is the problem of keeping “intrinsic variables” in reference states. And behavior is just support. That’s how organism survive.

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*.

HB : Where do you this in PCT ?

Bill P (B:CP):

CONTROL : Achievement and maintenance of a preselected state in the controlling system, through actions on the environment that also cancel the effects of disturbances.

And just in case also definition of PCT control from Tim Carey :

TC (2014) :

According to PCT, control is a process of acting to bring a perceived aspect of the world into a match with a mental specification for the state of that perception

And certain Marken wrote this (2001) :

For example, catching the fly ball seems to involve moving so as to keep a perception of the optical projection of the ball fairly stationary.

HB : This was probably PCT Marken (not RCT) as that is far away in the past (2001) when certain Marken understood PCT.

In these days ceratin Marken is daily changing his oppinion. It’s the fastest changing I ever saw…

RM : …q.i. is contrrolled by S and, thus, corresponds to the perception S is controlling.

HB : Once “q.i.” is contolled by S and few hours before “q.i.” was controlled by “E”.

RM : ….the difference between q.i and p is that q.i is an observation in E (or a surrogate of E) while p is a theoretical neural signal in S.

HB : But if in both seems to be neural currents just “theoretical” so I’m asking myself whether they are alive or not ?

You are confussion maker Rick.

RM : The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.

HB : Go read again Bills’ literature. Even if it would be exact analog, it is not controlled and kept in reference state. Perceptual signal is kept in reference state. You wrote it for yourself :

RM (2001) : For example, catching the fly ball seems to involve moving so as to keep a perception of the optical projection of the ball fairly stationary

HB : Obviously in the time when you understood PCT you use similar terms for outside events as Bill and Kent : stability.

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances)

HB : This is RCT (Rikcs’ Control Thoery) which affirms that “controlled variable” (q.i.) in environment is kept in reference state. So it is “controlled”. This is the state today. It is not PCT Rick.

RCT (Ricks Control Theory) definition of control loop

CONTROL : Keeping of some »aspect of outer environment« in reference state, protected (defended) from disturbances.

OUTPUT FUNCTION : controlled effects (control of behavior) to outer environment so to keep some »controlled variable« in reference state

HB : And how many times do we have to tell you that nothing is “protected from disturbances”. To Humans is usually enough to repeat things once or twice. But repeating somebody the sam thing 30 x someone must ask himself whether he is repeating it to human being. Are you maybe Alien Rick from another Planet ?

Disturbances mostly affect organism and perception so that organism can counteract. If organism is protected from disturbances how can counteract disturbances. How can you influence people if they are protected from disturbances ?

RM : ….that motivates the theory that says that thiis observation is a result of control of perception. The controlled variable, q.i, is data.

HB : Data to whom ? It sems that there is only one person that can see data from real reality. All mighty Rick (superman).

RM : PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled.

HB : Where PCT says that ?

RM : The perceptual signal, p, is theory.

HB : So in your nervous system there is no real neural signals.just theoretical So we could say that you are theoretical construction. I assumed that you must be from another Planet. My assumption seems to be right.

HB : And how will you explain what is happening in nervous system which can be observed by physiologist, biologist etc. ? What is fluctuating in your nervous system… Theoretical impulses which are part of theoretical organism. Rick look yourself in mirror. Do you really exist or you are just theorethical construct ???

RM: So the controlled variable, q.i,

HB : “Q.i.” is not controlled variable. “P” is. Perceptual signal is controlled in comparator not “q.i.”. See the diagram.

cid:image001.jpg@01D37ABE.36063DF0

RM : ….is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory).

HB : “Q.I.” is necesary part of any control loop of any LCS. Even Bacteria. See Bills diagram and stop living in your imagination. Open your eyes to see what is really happening in reality.

RM : The idea that this is not the case is a misunderstanding of PCT…

HB : Rick stop confussing and misleading people. Look what you’ve done to Warren. Does he really deserve to share your halucinations ? Stop being RCT Rick. Start being PCT Rick. At least for Warrens credibility if you don’t care for yourself. And there are Powers ladies. Do you think on anybody else but yourself.

The idea that you think that “q.i.” is just experimenters (E) perceptual signal is wrong and shows misunderstading of PCT. You are the one that don’t understand PCT. You are thinking in RCT.

RM : ….so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.

HB : Right. RCT (Ricks’ Control Theory) is perfectly useless for understanding the behavior of living organisms…and for understanding how organisms function.

PCT is theory about how orgsnisms function and the function of “behavior” is just supportive :

Bill P. at all (50th Anniversary, 2011) :

Perceptual Control Theory (PCT) provides a general theory of functioning for organisms. At the conceptual core of the theory is the observation that living things control the perceived environment by means of their behavior. Consequently, the phenomenon of control takes center stage in PCT, with observable behavior playing an important but supporting role.

Boris

Best

Rick

Richard S. Marken

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
–Antoine de Saint-Exupery

Cheers Fred.

···

From: “Fred Nickols” (fred@nickols.us via csgnet Mailing List) csgnet@lists.illinois.edu
Sent: Tuesday, May 29, 2018 12:47 PM
To: csgnet@lists.illinois.edu
Subject: RE: The controlled quantity (q.i) is data, the perceptual signal § is theory

[Fred Nickols (2018.05.29.0635 ET)]

Rick:

I understand that p is an analog of q.i. To control p we have to affect q.i. in ways such that p = r. An observer can see me affecting q.i. and say that I am controlling q.i. I can see me affecting q.i. and believe I am controlling q.i. Regardless of whose perspective you take, q.i. is not p and p is not q.i. p is an analog of q.i. q.i. is often a physical variable and p is a mental/biophysical signal. How can p = q.i.?

HB : Very good question Fred… :blush:

Boris

Fred

From: Richard Marken (rsmarken@gmail.com via csgnet Mailing List) csgnet@lists.illinois.edu
Sent: Tuesday, May 29, 2018 12:56 AM
To: csgnet@lists.illinois.edu
Cc: Richard Marken rsmarken@gmail.com
Subject: The controlled quantity (q.i) is data, the perceptual signal § is theory

[Rick Marken 2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective. I think this is the only way to see things if one is involved in doing research on PCT. Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me.

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p, from the observer’s perspective is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT: that it is a theory that explains the fact of control as it is seen in the behavior of living systems.

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*. Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance, are the data on which we base our conclusion that the behavior we see involves control. As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*. The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances) that motivates the theory that says that this observation is a result of control of perception. The controlled variable, q.i, is data. PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled. The perceptual signal, p, is theory.

RM: So the controlled variable, q.i, is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory). The idea that this is not the case is a misunderstanding of PCT so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.

Best

Rick

Richard S. Marken

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
–Antoine de Saint-Exupery

[Bruce Abbott (2018.05.29.1245 EDT)]

[Rick Marken 2018-05-28_21:46:30]

RM: In my reply to Rupert today I said:

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective. I think this is the only way to see things if one is involved in doing research on PCT. Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me.

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p, from the observer’s perspective is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT: that it is a theory that explains the fact of control as it is seen in the behavior of living systems.

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*. Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance, are the data on which we base our conclusion that the behavior we see involves control. As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*. The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances) that motivates the theory that says that this observation is a result of control of perception. The controlled variable, q.i, is data. PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled. The perceptual signal, p, is theory.

RM: So the controlled variable, q.i, is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory). The idea that this is not the case is a misunderstanding of PCT so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.

Let’s say that the participant is controlling perceived sound intensity, p, which for him is a logarithmic function of the physical sound intensity, q.i. Unbeknownst to us, the experimenter is an intelligent alien species from one of the planets orbiting Alpha Centuri. Although she possesses a sense of hearing, her auditory input function makes p directly proportional to q.i. So when the physical intensity of the sound increases, it increases logarithmically for the participant but linearly for the experimenter.

The experimenter conducts the Test for the controlled variable and observes that her perception of q.i. is being held relatively constant as q.i. is being subject to a time-varying disturbance. (The participant rotates a knob to “correct� perceived deviations of the sound’s intensity.)

Theoretically there is a physical quantity “out there� in the shared environment of the participant and the experimenter, of which the perceptions of both are a specific (though different) function. Two questions:

  1. Is q.i. the physical quantity (sound-wave amplitude) that constitutes the controlled quantity, or the experimenter’s perception of it?

  2. Is there any way for the experimenter to determine what the participant’s input function for sound intensity actually is, or must the experimenter simply assume that it is similar to her own?

Bruce

[Bruce Nevin 2018-05-29_14:00:09 ET]

Rick Marken 2018-05-28_21:46:30 –

Fundamental, and clear.

The theory posits that p is a quantity (a rate of firing), so q.i must also be a quantity because (according to theory) p is an exact analog of q.i; and for that reason, we have called that in the environment which we observe to be defended from disturbances the input quantity q.i even when we are unable to quantify our observation. The term ‘controlled environment variable’ or CEV is appealing when q.i is not quantified (or not quantifiable). That’s the only difference between them that I can see.

Fred Nickols (2018.05.29.0635 ET)–

Regardless of whose perspective you take, q.i. is not p and p is not q.i. p is an analog of q.i. q.i. is often a physical variable and p is a mental/biophysical signal. How can p = q.i.?

p is an exact analog of q.i. The perceptual input function regularly transforms the value of q.i to the value of p.

Martin Taylor 2018.05.29.09.09 –

A single variable such as q.i can’t mean two different things at the same time in normal science. You have to choose whether to use it for one or the other of the two different things. In the case of q.i, some use it as a variable in a control loop. But in this quote (and frequently elsewhere) Rick uses it to refer to a perception in a person who hypothesises that it is a perception of some external variable that is controlled. These two are not the same variable, and have the same value only by coincidence.

[…]

The Powers diagrams and discussions say that q.i is the variable C controls (theoretically by keeping a perception of it as near its reference variable as C is able to do). Rick says it is the variable T has determined by testing various hypotheses to be the external variable C is most likely to be controlling. They aren’t the same variable, so we ought to agree on which one to call q.i in our discussions. Is it the Powers version or the Marken version that should prevail, and what should we call the other?

Martin, I fear that you’re being unnecessarily obtuse. The experimenter hypothesizes the value q.i as input to the subject on the basis of a value that is input to the experimenter. How else? Then the experimenter must test this hypothesis by building a model, or two or three models with different hypotheses about q.i. By this process of hypothesis and test aims to know what the really real value q.i is in the really real environment that both the investigator and the subject are severally perceiving (and controlling). Of course the hypothesized q.i in the mind of the investigator is different from the q.i that is perceived and controlled by the subject (and by the investigator as well). What links them and brings them as close as possible to unison is called science.

Bruce Abbott (2018.05.29.1245 EDT)–

Let’s say that the participant is controlling perceived sound intensity, p, which for him is a logarithmic function of the physical sound intensity, q.i. Unbeknownst to us, the experimenter is an intelligent alien species from one of the planets orbiting Alpha Centuri. Although she possesses a sense of hearing, her auditory input function makes p directly proportional to q.i. So when the physical intensity of the sound increases, it increases logarithmically for the participant but linearly for the experimenter.

Your visiting scientist is foolishly making a fundamental assumption that our human scientists are justified in making but she is not, and that is the assumption that the human scientist and the human subject are both structured in the same way. A human scientist should check this assumption, but how often is that done? So easily does it go by the board that you have forgotten it here.

···

On Tue, May 29, 2018 at 12:45 PM, “Bruce Abbott” csgnet@lists.illinois.edu wrote:

[Bruce Abbott (2018.05.29.1245 EDT)]

Â

[Rick Marken 2018-05-28_21:46:30]

Â

RM: In my reply to Rupert today I said:Â

Â

RM… there seems to be general agreement that I am completely wrong to think that q.i is the controlled perception, p, from the observer’s perspective. I think this is the only way to see things if one is involved in doing research on PCT. Since most everyone on CSGNet is not involved in doing PCT research I suppose the only problem with not seeing that q.i as p from the observer’s perspective is that you’ll keep getting into useless verbal arguments about it with me. Â

Â

RM: Upon reflection I realized that failure to understand that the controlled quantity, q.i, is the controlled perception, p, from the observer’s perspective is not just a problem for researchers. If you don’t understand this, then you don’t understand what is fundamental about PCT: that it is a theory that explains the fact of control as it is seen in the behavior of living systems.Â

Â

RM: The controlled quantity is a variable that we see being kept in a reference state, often symbolized q.i*. Thus, variations (or lack thereof) in the controlled quantity, q.i, and the fact that q.i is being kept in a reference state, q.i*, protected from disturbance, are the data on which we base our conclusion that the behavior we see involves control. As Powers says on p. 175 of LCS “In these reference states we have the heart of the problem to which control theory is addressed”.Â

Â

RM: So PCT (which was then just called control theory because that’s what it is) is the theory that accounts for the observed fact that organisms keep certain variables, q.i, in reference states, q.i*. The theory accounts for this fact by assuming that the organism controls a perceptual signal, p, that is an exact analog of q.i.Â

Â

RM: It’s the fact of control (the fact that q.i is observed to be maintained in a reference state, q.i*, protected from normal disturbances) that motivates the theory that says that this observation is a result of control of perception. The controlled variable, q.i, is data. PCT says that this data can be accounted for by a theory that says that a perceptual signal that is an exact analog of of the controlled variable is being controlled. The perceptual signal, p, is theory.Â

Â

RM: So the controlled variable, q.i, is unquestionably the observer’s perspective on what the controller is presumed to be perceiving (in theory). The idea that this is not the case is a misunderstanding of PCT so profound as to make even a detailed knowledge of the theory perfectly useless for understanding the behavior of living organisms.Â

Â

Let’s say that the participant is controlling perceived sound intensity, p, which for him is a logarithmic function of the physical sound intensity, q.i. Unbeknownst to us, the experimenter is an intelligent alien species from one of the planets orbiting Alpha Centuri. Although she possesses a sense of hearing, her auditory input function makes p directly proportional to q.i. So when the physical intensity of the sound increases, it increases logarithmically for the participant but linearly for the experimenter.

Â

The experimenter conducts the Test for the controlled variable and observes that her perception of q.i. is being held relatively constant as q.i. is being subject to a time-varying disturbance. (The participant rotates a knob to “correct� perceived deviations of the sound’s intensity.)

Â

Theoretically there is a physical quantity “out there� in the shared environment of the participant and the experimenter, of which the perceptions of both are a specific (though different) function. Two questions:

Â

  1. Is q.i. the physical quantity (sound-wave amplitude) that constitutes the controlled quantity, or the experimenter’s perception of it?
  2. Is there any way for the experimenter to determine what the participant’s input function for sound intensity actually is, or must the experimenter simply assume that it is similar to her own?
    Â

Bruce

Â

[Bruce Abbott (2018.05.29.1520 EDT)]

[Bruce Nevin 2018-05-29_14:00:09 ET]

Bruce Abbott (2018.05.29.1245 EDT)–

Let’s say that the participant is controlling perceived sound intensity, p, which for him is a logarithmic function of the physical sound intensity, q.i. Unbeknownst to us, the experimenter is an intelligent alien species from one of the planets orbiting Alpha Centuri. Although she possesses a sense of hearing, her auditory input function makes p directly proportional to q.i. So when the physical intensity of the sound increases, it increases logarithmically for the participant but linearly for the experimenter.

Your visiting scientist is foolishly making a fundamental assumption that our human scientists are justified in making but she is not, and that is the assumption that the human scientist and the human subject are both structured in the same way. A human scientist should check this assumption, but how often is that done? So easily does it go by the board that you have forgotten it here.

I have not forgotten it here, Bruce. In fact, by bringing in my alien experimenter, I am highlighting it.

Rick has stated that q.i. is the experimenter’s perception. I prefer to interpret q.i. as the environmental quantity or quantities of which the participant’s perception is a function – real quantities out there inn the real world, not an experimenter’s perception of them. To highlight the difference, I brought in the example of the alien experimenter, whose perceptual input function differs significantly from the participant’s. The only significance I can see for bringing in the experimenter’s perception of q.i. is the assumption that what the experimenter perceives is what the participant perceives, and basing hypotheses on that assumption. An appropriately instrumented human experimenter might measure the output of the speaker that is producing q.i. in my example and discover that her impression of sound intensity is varying with the logarithm of the measured energy of the sound wave, and then might reasonably assume that the same is true of the human participant. My alien experimenter would find a proportional relationship between q.i. and her p and might wrongly hypothesize that the same is true of the human participant.

If q.i. is the experimenter’s p, I wonder how the experimenter’s p can affect the participant’s p, and what would happen if the experimenter were absent. Would q.i. cease to exist? (If a tree falls in the woods and there is no one to hear it . . .?). It makes a lot more sense to me to define q.i. in terms of real quantities and not experimenter perceptions, and leave it to experimental testing to discover their properties independent of any individual observer’s impressions of them. Following that tactic, both the human and alien observer should be able to deduce how the physical representation of q.i. relates to both the experimenter/observer’s and the participant’s p.

Care to take a stab at answering my two questions?

···

The experimenter conducts the Test for the controlled variable and observes that her perception of q.i. is being held relatively constant as q.i. is being subject to a time-varying disturbance. (The participant rotates a knob to “correct� perceived deviations of the sound’s intensity.)

Theoretically there is a physical quantity “out there� in the shared environment of the participant and the experimenter, of which the perceptions of both are a specific (though different) function. Two questions:

  1.   Is q.i. the physical quantity (sound-wave amplitude) that constitutes the controlled quantity, or the experimenter’s perception of it?
    
  1.   Is there any way for the experimenter to determine what the participant’s input function for sound intensity actually is, or must the experimenter simply assume that it is similar to her own?
    

Bruce

[Martin Taylor 2018.05.29.15.42]

[Bruce Nevin 2018-05-29_14:00:09 ET]

Martin Taylor 2018.05.29.09.09 –

        A single variable such as q.i can't mean two different

things at the same time in normal science. You have to
choose whether to use it for one or the other of the two
different things. In the case of q.i, some use it as a
variable in a control loop. But in this quote (and
frequently elsewhere) Rick uses it to refer to a perception
in a person who hypothesises that it is a perception of some
external variable that is controlled. These two are not the
same variable, and have the same value only by coincidence.

[…]

        The Powers diagrams and discussions say that q.i is the

variable C controls (theoretically by keeping a perception
of it as near its reference variable as C is able to do).
Rick says it is the variable T has determined by testing
various hypotheses to be the external variable C is most
likely to be controlling. They aren’t the same variable, so
we ought to agree on which one to call q.i in our
discussions. Is it the Powers version or the Marken version
that should prevail, and what should we call the other?

Martin, I fear that you’re being unnecessarily obtuse.

I'm not sure what you intend by "obtuse", which the OED and Random

House Unabridged define mainly in connection with angles, with a
metaphoric extension to dullness of the senses, which I suppose you
extend further to dullness of intellect. But if that is what you
intend, the word “unnecessarily” cannot apply, so all I can guess is
that you intend a pejorative comment with respect to my analysis.

My analysis provided just one among the variety of reasons provided

by myself and others why it is incorrect to say that q.i as a
perception in an observer/experimenter is (or even could be) the
same as a perception another person is controlling. Here’s the gist
of my argument. The rest of the “unnecessary” text was an effort to
be precise rather than applying to intuition on the part of the
reader.

Point 1. I allowed several impossibilities to show that even with

the observer/experimenter permitted to detect with the same values
everything that the subject perceives about the environment.

Point 2. No control can be perfect, so th observer-experimenter will

always see that control of ANY variable in the environment by the
subject is imperfect.

Point 3. The experimenter has no privileged access to the subject's

perceptual functions that are theorized to exist, so cannot know
just what functions of other environmental variables the subject may
use to create controlled perceptions.

Point 4. For any function of environmental variables there are

myriads of others that would seem to be controlled almost as well as
the one the subject is actually controlling.

Conclusion 1. The experimenter cannot know whether the function the

subject is controlling is among the hypotheses the experimenter is
comparing.

Conclusion 2. The experimenter cannot know whether the observed

imperfection of control for a given hypothesis about what the
subject is controlling is due to the hypothesis being wrong or to
inherent limits on the ability of the subject to control the
correctly hypothesized variable.

Final conclusion. The q.i that leads to a perception in the

experimenter is a different variable than the q.i that leads to a
perception being controlled by the subject.

···

On Tue, May 29, 2018 at 12:45 PM,
“Bruce Abbott” csgnet@lists.illinois.edu
wrote:

                [Bruce Abbott (2018.05.29.1245

EDT)]

Â

                        [Rick Marken 2018-05-28_21:46:30]

Â

                    RM:

In my reply to Rupert today I said:Â

Â

                      RM...

there seems to be general agreement that I am
completely wrong to think that q.i is the
controlled perception, p, from the observer’s
perspective. I think this is the only way to
see things if one is involved in doing
research on PCT. Since most everyone on
CSGNet is not involved in doing PCT research I
suppose the only problem with not seeing that
q.i as p from the observer’s perspective is
that you’ll keep getting into useless verbal
arguments about it with me. Â

Â

                      RM:

Upon reflection I realized that failure to
understand that the controlled quantity, q.i,
is the controlled perception, p, from the
observer’s perspective is not just a problem
for researchers. If you don’t understand this,
then you don’t understand what is fundamental
about PCT: that it is a theory that explains
the fact of control as it is seen in the
behavior of living systems.Â

Â

                      RM:

The controlled quantity is a variable that we
see being kept in a reference state, often
symbolized q.i*. Thus, variations (or lack
thereof) in the controlled quantity, q.i, and
the fact that q.i is being kept in a reference
state, q.i*, protected from disturbance, are
the data on which we base our conclusion that
the behavior we see involves control. As
Powers says on p. 175 of LCS “In these
reference states we have the heart of the
problem to which control theory is
addressed”.Â

Â

                      RM:

So PCT (which was then just called control
theory because that’s what it is) is the
theory that accounts for the observed fact
that organisms keep certain variables, q.i, in
reference states, q.i*. The theory accounts
for this fact by assuming that the organism
controls a perceptual signal, p, that is an
exact analog of q.i.Â

Â

                      RM:

It’s the fact of control (the fact that q.i is
observed to be maintained in a reference
state, q.i*, protected from normal
disturbances) that motivates the theory that
says that this observation is a result of
control of perception. The controlled
variable, q.i, is data. PCT says that this
data can be accounted for by a theory that
says that a perceptual signal that is an exact
analog of of the controlled variable is being
controlled. The perceptual signal, p, is
theory.Â

Â

                      RM:

So the controlled variable, q.i, is
unquestionably the observer’s perspective on
what the controller is presumed to be
perceiving (in theory). The idea that this is
not the case is a misunderstanding of PCT so
profound as to make even a detailed knowledge
of the theory perfectly useless for
understanding the behavior of living
organisms.Â

Â

                    Let’s say that the

participant is controlling perceived sound
intensity, p, which for him is a logarithmic
function of the physical sound intensity, q.i.Â
Unbeknownst to us, the experimenter is an
intelligent alien species from one of the
planets orbiting Alpha Centuri. Although she
possesses a sense of hearing, her auditory input
function makes p directly proportional to q.i.Â
So when the physical intensity of the sound
increases, it increases logarithmically for the
participant but linearly for the experimenter.

Â

                    The experimenter conducts the

Test for the controlled variable and observes
that her perception of q.i. is being held
relatively constant as q.i. is being subject to
a time-varying disturbance. (The participant
rotates a knob to “correct� perceived deviations
of the sound’s intensity.)

Â

                    Theoretically there is a

physical quantity “out there� in the shared
environment of the participant and the
experimenter, of which the perceptions of both
are a specific (though different) function. Two
questions:

Â

  1.                       Is q.i. the physical
    

quantity (sound-wave amplitude) that
constitutes the controlled quantity, or the
experimenter’s perception of it?

  1.                       Is there any way for
    

the experimenter to determine what the
participant’s input function for sound
intensity actually is, or must the
experimenter simply assume that it is similar
to her own?
Â

Bruce

Â

[Rick Marken 2018-05-30_20:25:27]

[Fred Nickols (2018.05.29.0635 ET)]

Â

FN: ... How can p = q.i.?

RM: When I say that p = q.i I don't mean that they are the same physical entities; I mean that p is the same function of environmental variables as q.i. For example, in the "What is size" demo (<http://www.mindreadings.com/ControlDemo/Size.html>http://www.mindreadings.com/ControlDemo/Size.html) if an observer (the computer, in this case) sees that the subject is controlling the area of the rectangle then q.i = height X width and it is assumed that p is also = height X width. So q.i = p = height X width. That is, q.i is a multiplicative function of the environmental variables height and width and so p is assumed to be the same multiplicative function of these variables. That is how p = q.i.
RM: A non-mathematical example can be seen in the "coin game" when E sees that S is controlling for a "Z" pattern of coins. In this case E can see that the controlled quantity, q.i, is the pattern of the coins, which is one function (or aspect) of the coins, as opposed to a different aspect of the coins such as their relative size. So it is assumed that S is controlling a perceptual signal, p, that is an analog of q.i, which are the variations in the aspect of the coins (the pattern) that E sees being controlled.
RM: This can be understood in terms of the "canonical" PCT diagram of the behavior of a living control system by noting that three variables in that diagram -- d, q.i, and q.o -- are on the "environment" side of the diagram. That is because the disturbance (d), controlled quantity (q.i) and output (q.o) variables can all be seen, sometimes with the necessary aid of instruments, by an observer of behaving system. For example, in the compensatory tracking task that we use to demonstrate the phenomenon of control (<http://www.mindreadings.com/ControlDemo/BasicTrack.html>http://www.mindreadings.com/ControlDemo/BasicTrack.html) the computer generated disturbance (d), controlled quantity (q.i, distance from cursor to target) and output (q.o, mouse movement) are all visible and plotted out as data at the end of the experiment (a model that assumes that the perceptual signal is also the distance from cursor to target -- that q.i = p -- accounts for the data nearly perfectly, suggesting that p does indeed = q.i).Â
RM: In many examples of behavior, however, d and q.o are detectable only with special instruments. For example, when a person moves their finger in a circular trajectory, q.i is the observed movement of the finger, d is the varying effect of gravity on the arm and q.o are the varying muscle forces that move the finger in the circular trajectory while, at the same time, opposing the varying gravitational forces that would prevent the circular trajectory from being produced. In this case only q.i is visible without instrumentation, making it hard to tell that the movement trajectory is a controlled quantity. Instrumentation would be needed to measure the varying effects of gravity and muscle forces on the arm movement. If such measurement were made the strong negative correlation between disturbance and output (as in the tracking task results) would make it clear that the trajectory of the finger is, indeed, a controlled quantity.Â
RM: In fact, in much is not most of what we see as behavior -- such as walking, opening doors, lifting suitcases -- its hard to tell that what we are seeing are controlled quantities, <http://q.is>q.i's, being kept in variable reference states because disturbances to these variables and the outputs that nearly perfectly oppose those disturbances are invisible without instrumentation. This is why no one before Powers' had managed to come to the realization that all behavior involves control. Powers was able to do it because he was trained as a physicist and he realized that the consistent results that we see people producing are constantly being affected by continuously varying disturbances, invisible to the observer, that should make such consistency extremely unlikely. So these disturbances must be being precisely countered by continuously varying outputs, also invisible to the observer, that are preventing these disturbances from having any effect on these results. So what Powers was able to see was that the consistent results that we see people producing -- the walking, opening doors, and lifting suitcases that seem to simply be "emitted" by the organism -- are controlled results of the organisms outputs: controlled quantities or q.i. PCT was developed to explain how organisms are able to do this.Â
BestÂ
Rick

Â

···

--
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Martin Taylor 2018.05.30.23.34]

That's rather different from saying that q.i is only a perception in

the mind of an observer, or that q.i is the same variable as p,
which are a couple of the claims against which some of us have been
arguing. If all the inputs to the perceptual function are based on
immediate sensory input, I come close to agreeing with this if you
substitute f(q.i) for q.i, allowing for possibilities such as that p
= log(q.i).

The only caveat is a question of the precision of the equality,

since p is some exact value and q.i is some exact value, but the
sensory equipment has limited resolution, which means that for a
particular precise value of q.i, p has some precise value from a
probability distribution around the average vale of p for that value
of q.i. So I would say that p has a slightly fuzzy mapping to q.i
(not is approximately equal, because it could approximate any
single-valued function of q.i).

Martin
···

[Rick Marken 2018-05-30_20:25:27]

                [Fred Nickols (2018.05.29.0635

ET)]

FN: … How can p = q.i.?

          RM: When I say that p = q.i I don't mean that they are

the same physical entities; I mean that p is the * same
function of environmental variables* as q.i.

[Bruce Nevin 2018-05-31_13:11:07 ET]

Bruce Abbott (2018.05.29.1520 EDT) –

Care to take a stab at answering my two questions?

 1.    Is q.i. the physical quantity (sound-wave amplitude) that constitutes the controlled quantity, or the experimenter’s perception of it?

In a sense, both: q.i is the physical quantity of input to the subject’s perceptual input function as represented by the experimenter’s perceptions thereof, formalized as a quantitative measurement–which of course is a perception controlled by the experimenter using measuring instruments appropriate to the sensory modality being investigated, as for example instruments for measuring the intensity of acoustic energy (pulsation of air pressure due to movement of molecules in atmospheric gases) at different levels of the acoustic spectrum. As such q.i is a scientifically convenient representation of the effect of certain aspects of the environment (as perceived by the experimenter) on the particular perceptual input function in the subject that is involved in the subject’s control loop (as perceived and experimentally verified by the experimenter). It is not ‘the thing in the environment’, but rather it is a numerical representation of that portion of the environmental influence of that thing which gives rise to the subject’s controlled perception, a critical datum in a model of the control loop which is the inferred object under investigation.

2.    Is there any way for the experimenter to determine what the participant’s input function for sound intensity actually is, or must the experimenter simply assume that it is similar to her own?Â

One way is by finding experimentally that the only way to get the damned computer simulation of the individual subject’s control to work was by defining the input function in that way. I hope she wouldn’t undertake invasive anatomical and functional investigations such as human scientists have done with nonhuman animals, but that kind of approach to psychophysics is another possibility. And no, the experimenter is not obligated to assume that she is structured the same way. I have presumed that your non-human visitor is a competent scientist and therefore would not make such an assumption about an alien species.

Martin (
2018.05.29.15.42) invokes differences in input functions due to different histories of reorganization. The difference between logarithmic and proportional transformations of acoustic energy to perceptual signals is fairly assumed to have been established by evolution rather than by learning. But your visitor by definition could not assume common evolutionary origin (barring chariots of the gods scenarios).

Martin Taylor 2018.05.29.15.42 –

I’m not sure what you intend by “obtuse”, which the OED and Random House Unabridged define mainly in connection with angles, with a metaphoric extension to dullness of the senses, which I suppose you extend further to dullness of intellect. But if that is what you intend, the word “unnecessarily” cannot apply, so all I can guess is that you intend a pejorative comment with respect to my analysis.

I’m sorry that the way I said that was abrasive. In American usage, at least, the meaning of the phrase “you’re being obtuse” is shifting toward “you’re missing something obvious”. (There’s more detail about that in a note at the end of this post.) Maybe I was wrong to say that you were missing something obvious. It may be that you are just drilling down on a different aspect of the problem, and what I see as obvious is simply out of scope for your exercise. By your four points you have concluded that perceptual input to the experimenter is necessarily different from perceptual input to the subject, and therefore both cannot be labeled q.i.Â

You’ve framed this in terms of labels in our control diagram. You object to representing more than one thing by a single label, q.i.Â

What are those several things? We have two perceptions of something, the observer’s and the subject’s, and the results of the Test warrant the claim that these are perceptions of the same thing in the environment. Intervening between the thing and the two perceptions, we have the perceptual input from that thing that gives rise to those two perceptions–effects described by physics as photons entering the eye, changes in air pressure entering the ears, and the like.

Where are the labels? We have a model of control and a diagram of that model. But the diagram only represents one of the two perceptions and only one of the two perceptual inputs. Our diagram has no place for perceptual input to the observer or for the observer’s perception arising from that perceptual input.

In (Bruce Nevin 20170321.19:45) I offered this diagram of the interaction of the observer and the subject. (I cannibalized a diagram of yours for the graphic elements.)

TCV1.so.jpg

The quantum q.i is the amount of perceptual input reaching those sensors and perceptual input functions that participate in the control loop that is represented by the upper part of the diagram. In this diagram, q.i for the subject is the observer’s quantification of the influence labeled S.s (uppercase S for ‘sensory input’). The corresponding input for the observer is labeled S.o, and as the observer is necessarily limited to measuring and quantifying S.o the equivalence of S.o to S.s is of critical importance.

The strictures of your four points, by the way, would seem logically to make it impossible to do PCT research. Even with (1) perfect consonance of point of view (2) the subject’s control is unpredictably imperfect, (3) her perceptual input functions may unpredictably differ from those of the experimenter, and (4) ‘myriad’ functions of atomic environmental variables could be controlled as well as that particular selection and combination that the subject is controlling.Â

But we know by numerous examples of meticulous and highly successful PCT research that this is possible. The experimental methodology of PCT enables the experimenter to control for (1)Â differences in point of view and (3) possible differences in perceptual input functions, and by process of elimination (4) find a best-fit specification of the perceptual input function, and (2) produce a generative model that is unpredictably imperfect in performance in the same way and to the same degree as the subject is. This is the fruit not of theory but rather of empirical trial and error. By current PCT standards, a model of motor control is acceptable which demonstrates 95% to 99% fidelity to the subject’s behavior. So yes, 100% fidelity is probably unattainable for the reasons that you indicate. But PCT research is demonstrably–demonstratedly–not impossible.

What is the relationship between S.s and S.o, and are we warranted to say that they both are what is usually meant by q.i? (Meaning, really, that if q.i is a quantification of S.o it is equally a quantification of S.i, within the abovementioned tolerance.) A result of the Test plus the creation and refinement of an accurate computer simulation is the equivalence of S.s and S.o within that 95% tolerance. If S.o does not equal S.s, then the model predicated upon S.o falls short of 95% fidelity because the observer has not entirely succeeded in (1) taking the point of view of the subject, (3) considering all possible perceptual input functions, and/or (4) considering all possible sources of environmental input to the controlled perception. But if the model is successful to that degree, then S.o does equal S.s to that degree, and the quantity represented by the label q.i is in fact (to that degree of accuracy) not only the quantity of input to the subject, S.s, but also the quantity measured by the experimenter and used to in the implementation of the simulation, S.o.

As I noted a year ago last March when I submitted this diagram, the perceptual signals p.s and p.o are unlikely to be identical because it is very improbable that the respective perceptual input functions would be identical (your point 3). The respective reference signals and error signals likewise need not be and probably are not identical.Â

  • The reference level at which q.i is controlled, however, is a datum that the observer measures in the environment, and then represents indirectly by the inferred value of the reference signal in the computer simulation.Â
  • In the model, the reference level that is observed in the environment is taken to be the value of perceptual input S.s which is transformed by the subject’s perceptual input function to the perceptual signal p.s which the subject is controlling.Â
  • But necessarily the reference level that is observed in the environment is the value of perceptual input S.o which is transformed by the observer’s perceptual input function to the perceptual signal p.o which the observer is perceiving and (when introducing disturbances) is controlling.Â

The observer not only perceives the reference level, the disturbances, and the subject’s control outputs as
 phenomena in the environment, the observer also measures and quantifies them. The measurements and quantities are additional perceptions that the observer is certainly controlling. Specifically, they are the output quantity q.o, the disturbance quantity q.d, and … here we go … the input quantity q.i.Â

The observer then asserts that the input quantity–a numerical value–is what is entering the subject’s perceptual input function, where it is transformed to another number, p.s, and so on. This is a projection of the model, as imagined by the observer, onto what is going on. In the real world, there are no numbers in transit. The numbers are in the model.

If what is going on is control of perception of sound (Bruce Abbott’s example), molecules are in motion with the aggregate effect that air pressure is varying (with probably complex changes in rate may be teased out into component sine waves by Fourier analysis) over a range (amplitude) which also may be varying. But in the experimental work that I have in mind what the observer measures is bands of acoustic energy, each roughly 400 Hz wide and attenuating at the upper and lower margins, at time t centered at roughly 290, 1900, and 2500 Hz., and then gradually changing over time until at time t’ the three bands are centered at roughly 450, 1550, and 2500 Hz. The subject (using means that we’ll ignore here) resists this change and restores the centers of acoustic energy to approximately their original values centered at 290, 1900, and 2500 Hz. The observer is monitoring and recording the numerical values. The subject is hearing the the sound of her own voice in headphones producing the vowel /i/ (as in “bid”) and gradually shifting to the vowel /É›/ (as in “bed”), and is acting in such a way as to hear the original sound of the vowel in “bid” in her headphones. The observer, while monitoring and recording the numerical values, is concurrently hearing the subject’s voice in the open environment first producing the /i/ of “bid” and then gradually shifting toward the vowel /æ/ in “bad”, and sees the centers or acoustic energy at about 450, 1550, and 2300 Hz.

The disturbance shifting the centers of energy in the acoustic spectrum affects only the sound heard in the headphones. The observer can also listen to the sound in the headphones, but for purposes of documenting results and creating a simulation the numbers are what matter.

Whenever the experimenter has the matching headphones on, both S.s and S.o may be represented by those numbers, labeled q.i. But q.i is a numerical representation of the perceptual input S.s (and sometimes S.o).Â

Here is an image of the sound spectrograms (with the formants for “bead” in there on the left side for good measure):

formants.jpg

···

On Tue, May 29, 2018 at 4:28 PM, Martin Taylor csgnet@lists.illinois.edu wrote:

[Martin Taylor 2018.05.29.15.42]

[Bruce Nevin 2018-05-29_14:00:09 ET]

Martin Taylor 2018.05.29.09.09 –

        A single variable such as q.i can't mean two different

things at the same time in normal science. You have to
choose whether to use it for one or the other of the two
different things. In the case of q.i, some use it as a
variable in a control loop. But in this quote (and
frequently elsewhere) Rick uses it to refer to a perception
in a person who hypothesises that it is a perception of some
external variable that is controlled. These two are not the
same variable, and have the same value only by coincidence.

[…]

        The Powers diagrams and discussions say that q.i is the

variable C controls (theoretically by keeping a perception
of it as near its reference variable as C is able to do).
Rick says it is the variable T has determined by testing
various hypotheses to be the external variable C is most
likely to be controlling. They aren’t the same variable, so
we ought to agree on which one to call q.i in our
discussions. Is it the Powers version or the Marken version
that should prevail, and what should we call the other?

Martin, I fear that you’re being unnecessarily obtuse.

I'm not sure what you intend by "obtuse", which the OED and Random

House Unabridged define mainly in connection with angles, with a
metaphoric extension to dullness of the senses, which I suppose you
extend further to dullness of intellect. But if that is what you
intend, the word “unnecessarily” cannot apply, so all I can guess is
that you intend a pejorative comment with respect to my analysis.

My analysis provided just one among the variety of reasons provided

by myself and others why it is incorrect to say that q.i as a
perception in an observer/experimenter is (or even could be) the
same as a perception another person is controlling. Here’s the gist
of my argument. The rest of the “unnecessary” text was an effort to
be precise rather than applying to intuition on the part of the
reader.

Point 1. I allowed several impossibilities to show that even with

the observer/experimenter permitted to detect with the same values
everything that the subject perceives about the environment.

Point 2. No control can be perfect, so th observer-experimenter will

always see that control of ANY variable in the environment by the
subject is imperfect.

Point 3. The experimenter has no privileged access to the subject's

perceptual functions that are theorized to exist, so cannot know
just what functions of other environmental variables the subject may
use to create controlled perceptions.

Point 4. For any function of environmental variables there are

myriads of others that would seem to be controlled almost as well as
the one the subject is actually controlling.

Conclusion 1. The experimenter cannot know whether the function the

subject is controlling is among the hypotheses the experimenter is
comparing.

Conclusion 2. The experimenter cannot know whether the observed

imperfection of control for a given hypothesis about what the
subject is controlling is due to the hypothesis being wrong or to
inherent limits on the ability of the subject to control the
correctly hypothesized variable.

Final conclusion. The q.i that leads to a perception in the

experimenter is a different variable than the q.i that leads to a
perception being controlled by the subject.

=============



That argument gives the experimenter magical powers, which are still

insufficiently powerful to allow the experimenter’s perceived q.i to
be identified with the subject’s q.i or the subject’s perception of
q.i. If you reduce the experimenters access to magical powers, you
don’t even need the argument above, as various people have pointed
out:

(1) The subject's p may be a different function of q.i than the

experimenter’s.

(2) The experimenter cannot have the identical values of all

environmental variables as does the subject, and therefore is not
necessarily able to form a perception that is the same as the
subject’s for any aspect of the environment.

(3) The experimenter and the subject have had different life

experiences, and are unlikely to have reorganized to have the same
set of perceptual functions other than those that are genetically
determined by genes they have in common.

(4 related to 3) The experimenter and the subject have perceptual

experiences immediately before the study that are different, and

(5) Rupert's "If a tree falls in the forest when nobody is

listening" question: If there is no observer or experimenter, can
the subject keep controlling? If yes, then q.i cannot be a
perception in the experimenter, but must be something that the
subject perceives to be in the environment.

Any one of these would be sufficient to demonstrate that the q.i in

the environment that corresponds to a perception the subject is
controlling is a different variable than the one the experimenter
perceives and is testing. Your following comment goes further to
provide yet another argument for the same conclusion.

      The experimenter hypothesizes the value q.i as input to the

subject on the basis of a value that is input to the
experimenter. How else? Then the experimenter must test this
hypothesis by building a model, or two or three models with
different hypotheses about q.i.

When the possibilities are essentially uncountable in number, two or

three isn’t going to get you very far. You are talking about
optimizing one function so that it approaches a match to another
function. But the first is a map of the second, and the map gets
more and more complicated the higher the level in the hierarchy.
Most of us go along with Korzybski “The map is not the territory”.
The only point at issue is Rick’s repeated assertion that the map IS
the territory.

      By this process of hypothesis and test aims to know what

the really real value q.i is in the really real environment
that both the investigator and the subject are severally
perceiving (and controlling). Of course the hypothesized q.i
in the mind of the investigator is different from the q.i that
is perceived and controlled by the subject (and by the
investigator as well). What links them and brings them as
close as possible to unison is called science.

Apart from the parenthetical assertion that the investigator

controls q.i (presumably during the TCV), I have no disagreement
with this paragraph at all. The “of course” sentence indicates that
you do agree with the others who have commented on this issue – the
investigator’s q.i is not the subject’s q.i, and the two of them
should have different names in our discussions.

Bruce Abbott (2018.05.29.1245 EDT)–

        Let’s say that the participant is controlling perceived

sound intensity, p, which for him is a logarithmic function
of the physical sound intensity, q.i. Unbeknownst to us,
the experimenter is an intelligent alien species from one of
the planets orbiting Alpha Centuri. Although she possesses
a sense of hearing, her auditory input function makes p
directly proportional to q.i. So when the physical
intensity of the sound increases, it increases
logarithmically for the participant but linearly for the
experimenter.

      Your visiting scientist is foolishly making a fundamental

assumption that our human scientists are justified in making
but she is not, and that is the assumption that the human
scientist and the human subject are both structured in the
same way.

Surely a PCT theorist who believes in any form of learning or

reorganization would not make that assumption, would they? The
probable failure of the assumption is part of what makes the
experimenter’s function-optimization task so difficult.

      A human scientist should check this assumption, but how

often is that done?

Quite a lot, by conventional psychophysical researchers. Perhaps not

by the PCT-aware researchers who really ought to me even more
sensitive to the possibility that the assumption fails.

Martin
        On Tue, May 29, 2018 at 12:45 PM,

“Bruce Abbott” csgnet@lists.illinois.edu
wrote:

                [Bruce Abbott (2018.05.29.1245

EDT)]

Â

                        [Rick Marken 2018-05-28_21:46:30]

Â

                    RM:

In my reply to Rupert today I said:Â

Â

                      RM...

there seems to be general agreement that I am
completely wrong to think that q.i is the
controlled perception, p, from the observer’s
perspective. I think this is the only way to
see things if one is involved in doing
research on PCT. Since most everyone on
CSGNet is not involved in doing PCT research I
suppose the only problem with not seeing that
q.i as p from the observer’s perspective is
that you’ll keep getting into useless verbal
arguments about it with me. Â

Â

                      RM:

Upon reflection I realized that failure to
understand that the controlled quantity, q.i,
is the controlled perception, p, from the
observer’s perspective is not just a problem
for researchers. If you don’t understand this,
then you don’t understand what is fundamental
about PCT: that it is a theory that explains
the fact of control as it is seen in the
behavior of living systems.Â

Â

                      RM:

The controlled quantity is a variable that we
see being kept in a reference state, often
symbolized q.i*. Thus, variations (or lack
thereof) in the controlled quantity, q.i, and
the fact that q.i is being kept in a reference
state, q.i*, protected from disturbance, are
the data on which we base our conclusion that
the behavior we see involves control. As
Powers says on p. 175 of LCS “In these
reference states we have the heart of the
problem to which control theory is
addressed”.Â

Â

                      RM:

So PCT (which was then just called control
theory because that’s what it is) is the
theory that accounts for the observed fact
that organisms keep certain variables, q.i, in
reference states, q.i*. The theory accounts
for this fact by assuming that the organism
controls a perceptual signal, p, that is an
exact analog of q.i.Â

Â

                      RM:

It’s the fact of control (the fact that q.i is
observed to be maintained in a reference
state, q.i*, protected from normal
disturbances) that motivates the theory that
says that this observation is a result of
control of perception. The controlled
variable, q.i, is data. PCT says that this
data can be accounted for by a theory that
says that a perceptual signal that is an exact
analog of of the controlled variable is being
controlled. The perceptual signal, p, is
theory.Â

Â

                      RM:

So the controlled variable, q.i, is
unquestionably the observer’s perspective on
what the controller is presumed to be
perceiving (in theory). The idea that this is
not the case is a misunderstanding of PCT so
profound as to make even a detailed knowledge
of the theory perfectly useless for
understanding the behavior of living
organisms.Â

Â

                    Let’s say that the

participant is controlling perceived sound
intensity, p, which for him is a logarithmic
function of the physical sound intensity, q.i.Â
Unbeknownst to us, the experimenter is an
intelligent alien species from one of the
planets orbiting Alpha Centuri. Although she
possesses a sense of hearing, her auditory input
function makes p directly proportional to q.i.Â
So when the physical intensity of the sound
increases, it increases logarithmically for the
participant but linearly for the experimenter.

Â

                    The experimenter conducts the

Test for the controlled variable and observes
that her perception of q.i. is being held
relatively constant as q.i. is being subject to
a time-varying disturbance. (The participant
rotates a knob to “correct� perceived deviations
of the sound’s intensity.)

Â

                    Theoretically there is a

physical quantity “out there� in the shared
environment of the participant and the
experimenter, of which the perceptions of both
are a specific (though different) function. Two
questions:

Â

  1.                       Is q.i. the physical
    

quantity (sound-wave amplitude) that
constitutes the controlled quantity, or the
experimenter’s perception of it?

  1.                       Is there any way for
    

the experimenter to determine what the
participant’s input function for sound
intensity actually is, or must the
experimenter simply assume that it is similar
to her own?
Â

Bruce

Â

[Martin Taylor 2018.05.31.13.38]

[Bruce Nevin 2018-05-31_13:11:07 ET]

      Martin (
        2018.05.29.15.42)

invokes differences in input functions due to different
histories of reorganization. The difference between
logarithmic and proportional transformations of acoustic
energy to perceptual signals is fairly assumed to have been
established by evolution rather than by learning. But your
visitor by definition could not assume common evolutionary
origin (barring chariots of the gods scenarios).

Martin Taylor 2018.05.29.15.42 –

          I'm not sure what you

intend by “obtuse”, which the OED and Random House
Unabridged define mainly in connection with angles, with a
metaphoric extension to dullness of the senses, which I
suppose you extend further to dullness of intellect. But
if that is what you intend, the word “unnecessarily”
cannot apply, so all I can guess is that you intend a
pejorative comment with respect to my analysis.

      I'm sorry that the way I said that was abrasive. In

American usage, at least, the meaning of the phrase “you’re
being obtuse” is shifting toward “you’re missing something
obvious”. (There’s more detail about that in a note at the end
of this post.) Maybe I was wrong to say that you were missing
something obvious. It may be that you are just drilling down
on a different aspect of the problem, and what I see as
obvious is simply out of scope for your exercise. By your four
points you have concluded that perceptual input to the
experimenter is necessarily different from perceptual input to
the subject, and therefore both cannot be labeled q.i.

Exactly. Your discussion is fine, (at least I have no problem with

it), but it simply depends on asserting that “approximately equal”
is identically equivalent to “exactly equal”, and that if two
variables are with high probability approximately equal, then they
are the same variable. They are not.

        The

strictures of your four points, by the way, would seem
logically to make it impossible to do PCT research. Even
with (1) perfect consonance of point of view (2) the
subject’s control is unpredictably imperfect, (3) her
perceptual input functions may unpredictably differ from
those of the experimenter, and (4) ‘myriad’ functions of
atomic environmental variables could be controlled as well
as that particular selection and combination that the
subject is controlling.

How do you draw that conclusion? In every science, measures of

variables have a limited resolution. My only point was that it is
improper to say that the map is the territory. The measure (the
experimenter’s estimation/perception) is not the thing measured (the
value of the input to the subject, for example, or the value of the
physical variable that is assumed to be identically the
input to the subject).

An inherent inability to make the map BE the territory, by your

logic, would make all science impossible, not just PCT science. I
think there is some fault in that logic, since I think that much
science has done and is doing a creditable job of providing a
workable construct of how the world works. I include PCT science in
that class, and I reject the logic that says that if a measure isn’t
exact, and if a measure is actually of a property whose values are
correlated with some other variable (e.g. the reading of a voltmeter
as compared to the actual voltage) then that measure totally fails
as a measure of the target variable (i.e. I reject the idea that
reading the voltmeter says nothing about the actual voltage).

That's enough for now. I have to shut my computer down to prepare

for leaving on vacation, so I expect that this response is full of
typos.

Martin

[Rick Marken 2018-05-31_14:40:49]

[Martin Taylor 2018.05.30.23.34]

RM: When I say that p = q.i I don't mean that they are the same physical entities; I mean that p is the same function of environmental variables as q.i.

MT: That's rather different from saying that q.i is only a perception in the mind of an observer,

 RM: No, it's the same thing. In PCT a perception is a function of environmental variables.

MT: or that q.i is the same variable as p,

 RM: Actually, q.i is the same variable as p in the sense that both are the same function of environmental variables.

MT:Â If all the inputs to the perceptual function are based on immediate sensory input, I come close to agreeing with this if you substitute f(q.i) for q.i, allowing for possibilities such as that p = log(q.i).

RM: This implies that q.i is a variable in the environment. It is not. It is a function of environmental variables.Â
Â

MT: The only caveat is a question of the precision of the equality, since p is some exact value and q.i is some exact value

RM: In PCT research the question (posed by the test for the controlled variable) is whether an observed function of environmental variables corresponds to the actual function of environmental variables that is under controlled. Precision of control, in terms of how well p corresponds to q.i, is never a question in PCT research because p is a theoretical variable that always corresponds exactly to what the researcher determines the controlled quantity (q.i) to be. To the extent that one wants to develop a model that includes some neural noise, this noise would typically be introduced to the perceptual signal after it has been computed by the perceptual function that corresponds to the function that defines q.i. So if q.i is found to be the area of a rectangle so that q.i = height * width then p in the model of control of area is defined as p = height * width. If you want to add some noise into the perceptual signal in order to to account for the lack of precision of control then p = height * width + noise. But noise plays very little part in control, often accounting for less than .5% of the variance in q.i and probably less than 1/2 of the noise is in the perceptual signal. The accuracy of perception is mainly a question in conventional psychophysical research where the interest is in determining how precisely an organism can perceive a variable of interest to the psychophysicist, but not necessarily of any interest to the organism.
Â

MT: but the sensory equipment has limited resolution, which means that for a particular precise value of q.i, p has some precise value from a probability distribution around the average vale of p for that value of q.i. So I would say that p has a slightly fuzzy mapping to q.i (not is approximately equal, because it could approximate any single-valued function of q.i).

RM: Since p is a theoretical variable that is always defined by the function that defines q.i, I think it's better to think of p as an exact mapping of q.i to p with some added noise that makes variations in p not exactly equal to variations in q.i. The important thing to keep in mind is that controlled quantities, q.i's, are not variables out in the environment; they are not variables "out there" to be perceived "through a glass darkly. Controlled quantities are aspects of the environment that are defined by functions of variables that are in the environment. The main goal of PCT research is to determine the functions that define the controlled quantities that organisms control and the relationship between those controlled quantities.Â
Best
Rick

···

--
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Bruce Nevin 2018-05-31_18:12:04 ET]

From here, this disagreement looks like a divergence of reference:

Rick: p is a numerical variable in a simulation which the theoretical model attributes to the organism. It is computed as a function of environmental variables, the same function of variables which specifies q.i. Therefore p = q.i by definition, because they are both defined as the same function of the same variables. Rick is referring to a generative simulation which models the organism’s control.

Martin: p. is a signal within the organism which is computed by sensors and neural structures in the organism. The theoretical model represents these sensors and neural structures as the perceptual input function in the control loop. q.i is the net effect of certain environmental variables impinging upon the organism’s sensors and being transformed by neural structures so as to constitute the signal p within the organism. q.i and p are necessarily different. The difference is inherent in the transformation from physical phenomena to neural rates of firing which is accomplished by the sensors and neural structures that constitute the perceptual input function. Martin is referring to physical and psychophysical events, processes, and relationships which can be modeled by a generative simulation.

Martin will not be able to correct my representation of his views until sometime next week. I think the basis for how he and Rick are talking past each other will turn out to be a difference of reference along these lines.

···

On Thu, May 31, 2018 at 5:40 PM, Richard Marken csgnet@lists.illinois.edu wrote:

[Rick Marken 2018-05-31_14:40:49]

[Martin Taylor 2018.05.30.23.34]

MT: That's rather different from saying that q.i is only a perception in

the mind of an observer,

 RM: No, it’s the same thing. In PCT a perception is a function of environmental variables.

MT: or that q.i is the same variable as p,

 RM: Actually, q.i is the same variable as p in the sense that both are the same function of environmental variables.

MT:Â  If all the inputs to the perceptual function are based on

immediate sensory input, I come close to agreeing with this if you
substitute f(q.i) for q.i, allowing for possibilities such as that p
= log(q.i).

RM: This implies that q.i is a variable in the environment. It is not. It is a function of environmental variables.Â

Â

MT: The only caveat is a question of the precision of the equality,

since p is some exact value and q.i is some exact value

RM: In PCT research the question (posed by the test for the controlled variable) is whether an observed function of environmental variables corresponds to the actual function of environmental variables that is under controlled. Precision of control, in terms of how well p corresponds to q.i, is never a question in PCT research because p is a theoretical variable that always corresponds exactly to what the researcher determines the controlled quantity (q.i) to be. To the extent that one wants to develop a model that includes some neural noise, this noise would typically be introduced to the perceptual signal after it has been computed by the perceptual function that corresponds to the function that defines q.i. So if q.i is found to be the area of a rectangle so that q.i = height * width then p in the model of control of area is defined as p = height * width. If you want to add some noise into the perceptual signal in order to to account for the lack of precision of control then p = height * width + noise. But noise plays very little part in control, often accounting for less than .5% of the variance in q.i and probably less than 1/2 of the noise is in the perceptual signal. The accuracy of perception is mainly a question in conventional psychophysical research where the interest is in determining how precisely an organism can perceive a variable of interest to the psychophysicist, but not necessarily of any interest to the organism.

Â

MT: but the

sensory equipment has limited resolution, which means that for a
particular precise value of q.i, p has some precise value from a
probability distribution around the average vale of p for that value
of q.i. So I would say that p has a slightly fuzzy mapping to q.i
(not is approximately equal, because it could approximate any
single-valued function of q.i).

RM: Since p is a theoretical variable that is always defined by the function that defines q.i, I think it’s better to think of p as an exact mapping of q.i to p with some added noise that makes variations in p not exactly equal to variations in q.i. The important thing to keep in mind is that controlled quantities*, q.i’s, are not variables out in the environment*; they are not variables “out there” to be perceived "through a glass darkly. Controlled quantities are aspects of the environment that are defined by functions of variables that are in the environment. The main goal of PCT research is to determine the functions that define the controlled quantities that organisms control and the relationship between those controlled quantities.Â

Best

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

          RM: When I say that p = q.i I don't mean that they are

the same physical entities; I mean that p is the * same
function of environmental variables* as q.i.

[Bruce Nevin 2018-06-01_07:00:44 ET]

In summary, and again referring to the two-person diagram in (Bruce Nevin  2018-05-31_13:11:07), the observer/investigator derives a measured value (a perception that is controlled by carrying out measuring and quantifying procedures) from his own S.o input and assigns that value to q.i in a diagram and a corresponding model/simulation of the subject’s control loop, but in the diagram and model q.i refers to the subject’s input S.s. That diagram conventionally has omitted the observer/investigator and S.o. This omission obscures the origin and character of q.i. This obscurity has contributed to a long-running dispute about the existential status of q.i.

S.o is input for the observer’s perception of the subject’s environmental input S.s; that is, it is input for the observer’s perception of the influence from the relevant aspects of the environment upon the subject’s relevant perceptual input function(s). q.i is the observer/experimenter’s quantification of S.o.Â

The output variable q.o and the disturbance variable q.d (or simply d) originate in the same way as quantification perceptions controlled by the observer, but have not been subject to the same kind of dispute. This is because q.i is sometimes identified with that which is perceived, the really real whatsis in the environment, whereas q.o and q.d are clearly understood as (measures of) environmental influences affecting the value of q.i. But actually q.i is also (a measure of) an environmental influence; it is (ideally) a quantitative measure of the influence of that really real whatever-it-is upon the perceptual input function(s) of the subject, as perceived and measured by the observer/experimenter.

I say ‘ideally’ because in many cases we don’t or can’t quantify that influence, but we conventionally use the term q.i for ‘input quantity’ anyway. This usage follows naturally from construing these three variables as influences in the environment rather than as controlled perceptions (quantitative measures) of those influences that are really real in the environment. This informality of usage adds to the confusion if it supports the misconstrual of q.i as that-which-is-perceived, forgetting that q.i is a quantitative measure by the observer of observed environmental influence from that-which-is-perceived upon the subject’s sensors.

You will note that that-which-is-perceived is represented in the diagram by a small green rectangle that has no label. This rectangle might be understood as those aspects of that-which-is-perceived which are influenced by q.o and q.d with a resulting effect on q.i. So understood, that rectangle might merit the label CEV (controlled environmental variable), but the control model doesn’t need a variable for what is really real in the environment, so all that the diagram needs is those three environmental influences around a black box (or here a green box). Or four influences, counting S.o.

formants.jpg

TCV1.so.jpg

···

On Thu, May 31, 2018 at 1:26 PM, Bruce Nevin bnhpct@gmail.com wrote:

[Bruce Nevin 2018-05-31_13:11:07 ET]

Bruce Abbott (2018.05.29.1520 EDT) –

Care to take a stab at answering my two questions?

 1.    Is q.i. the physical quantity (sound-wave amplitude) that constitutes the controlled quantity, or the experimenter’s perception of it?

In a sense, both: q.i is the physical quantity of input to the subject’s perceptual input function as represented by the experimenter’s perceptions thereof, formalized as a quantitative measurement–which of course is a perception controlled by the experimenter using measuring instruments appropriate to the sensory modality being investigated, as for example instruments for measuring the intensity of acoustic energy (pulsation of air pressure due to movement of molecules in atmospheric gases) at different levels of the acoustic spectrum. As such q.i is a scientifically convenient representation of the effect of certain aspects of the environment (as perceived by the experimenter) on the particular perceptual input function in the subject that is involved in the subject’s control loop (as perceived and experimentally verified by the experimenter). It is not ‘the thing in the environment’, but rather it is a numerical representation of that portion of the environmental influence of that thing which gives rise to the subject’s controlled perception, a critical datum in a model of the control loop which is the inferred object under investigation.

2.    Is there any way for the experimenter to determine what the participant’s input function for sound intensity actually is, or must the experimenter simply assume that it is similar to her own?Â

One way is by finding experimentally that the only way to get the damned computer simulation of the individual subject’s control to work was by defining the input function in that way. I hope she wouldn’t undertake invasive anatomical and functional investigations such as human scientists have done with nonhuman animals, but that kind of approach to psychophysics is another possibility. And no, the experimenter is not obligated to assume that she is structured the same way. I have presumed that your non-human visitor is a competent scientist and therefore would not make such an assumption about an alien species.

Martin (
2018.05.29.15.42) invokes differences in input functions due to different histories of reorganization. The difference between logarithmic and proportional transformations of acoustic energy to perceptual signals is fairly assumed to have been established by evolution rather than by learning. But your visitor by definition could not assume common evolutionary origin (barring chariots of the gods scenarios).

Martin Taylor 2018.05.29.15.42 –

I’m not sure what you intend by “obtuse”, which the OED and Random House Unabridged define mainly in connection with angles, with a metaphoric extension to dullness of the senses, which I suppose you extend further to dullness of intellect. But if that is what you intend, the word “unnecessarily” cannot apply, so all I can guess is that you intend a pejorative comment with respect to my analysis.

I’m sorry that the way I said that was abrasive. In American usage, at least, the meaning of the phrase “you’re being obtuse” is shifting toward “you’re missing something obvious”. (There’s more detail about that in a note at the end of this post.) Maybe I was wrong to say that you were missing something obvious. It may be that you are just drilling down on a different aspect of the problem, and what I see as obvious is simply out of scope for your exercise. By your four points you have concluded that perceptual input to the experimenter is necessarily different from perceptual input to the subject, and therefore both cannot be labeled q.i.Â

You’ve framed this in terms of labels in our control diagram. You object to representing more than one thing by a single label, q.i.Â

What are those several things? We have two perceptions of something, the observer’s and the subject’s, and the results of the Test warrant the claim that these are perceptions of the same thing in the environment. Intervening between the thing and the two perceptions, we have the perceptual input from that thing that gives rise to those two perceptions–effects described by physics as photons entering the eye, changes in air pressure entering the ears, and the like.

Where are the labels? We have a model of control and a diagram of that model. But the diagram only represents one of the two perceptions and only one of the two perceptual inputs. Our diagram has no place for perceptual input to the observer or for the observer’s perception arising from that perceptual input.

In (Bruce Nevin 20170321.19:45) I offered this diagram of the interaction of the observer and the subject. (I cannibalized a diagram of yours for the graphic elements.)

The quantum q.i is the amount of perceptual input reaching those sensors and perceptual input functions that participate in the control loop that is represented by the upper part of the diagram. In this diagram, q.i for the subject is the observer’s quantification of the influence labeled S.s (uppercase S for ‘sensory input’). The corresponding input for the observer is labeled S.o, and as the observer is necessarily limited to measuring and quantifying S.o the equivalence of S.o to S.s is of critical importance.

The strictures of your four points, by the way, would seem logically to make it impossible to do PCT research. Even with (1) perfect consonance of point of view (2) the subject’s control is unpredictably imperfect, (3) her perceptual input functions may unpredictably differ from those of the experimenter, and (4) ‘myriad’ functions of atomic environmental variables could be controlled as well as that particular selection and combination that the subject is controlling.Â

But we know by numerous examples of meticulous and highly successful PCT research that this is possible. The experimental methodology of PCT enables the experimenter to control for (1)Â differences in point of view and (3) possible differences in perceptual input functions, and by process of elimination (4) find a best-fit specification of the perceptual input function, and (2) produce a generative model that is unpredictably imperfect in performance in the same way and to the same degree as the subject is. This is the fruit not of theory but rather of empirical trial and error. By current PCT standards, a model of motor control is acceptable which demonstrates 95% to 99% fidelity to the subject’s behavior. So yes, 100% fidelity is probably unattainable for the reasons that you indicate. But PCT research is demonstrably–demonstratedly–not impossible.

What is the relationship between S.s and S.o, and are we warranted to say that they both are what is usually meant by q.i? (Meaning, really, that if q.i is a quantification of S.o it is equally a quantification of S.i, within the abovementioned tolerance.) A result of the Test plus the creation and refinement of an accurate computer simulation is the equivalence of S.s and S.o within that 95% tolerance. If S.o does not equal S.s, then the model predicated upon S.o falls short of 95% fidelity because the observer has not entirely succeeded in (1) taking the point of view of the subject, (3) considering all possible perceptual input functions, and/or (4) considering all possible sources of environmental input to the controlled perception. But if the model is successful to that degree, then S.o does equal S.s to that degree, and the quantity represented by the label q.i is in fact (to that degree of accuracy) not only the quantity of input to the subject, S.s, but also the quantity measured by the experimenter and used to in the implementation of the simulation, S.o.

As I noted a year ago last March when I submitted this diagram, the perceptual signals p.s and p.o are unlikely to be identical because it is very improbable that the respective perceptual input functions would be identical (your point 3). The respective reference signals and error signals likewise need not be and probably are not identical.Â

  • The reference level at which q.i is controlled, however, is a datum that the observer measures in the environment, and then represents indirectly by the inferred value of the reference signal in the computer simulation.Â
  • In the model, the reference level that is observed in the environment is taken to be the value of perceptual input S.s which is transformed by the subject’s perceptual input function to the perceptual signal p.s which the subject is controlling.Â
  • But necessarily the reference level that is observed in the environment is the value of perceptual input S.o which is transformed by the observer’s perceptual input function to the perceptual signal p.o which the observer is perceiving and (when introducing disturbances) is controlling.Â

The observer not only perceives the reference level, the disturbances, and the subject’s control outputs as
 phenomena in the environment, the observer also measures and quantifies them. The measurements and quantities are additional perceptions that the observer is certainly controlling. Specifically, they are the output quantity q.o, the disturbance quantity q.d, and … here we go … the input quantity q.i.Â

The observer then asserts that the input quantity–a numerical value–is what is entering the subject’s perceptual input function, where it is transformed to another number, p.s, and so on. This is a projection of the model, as imagined by the observer, onto what is going on. In the real world, there are no numbers in transit. The numbers are in the model.

If what is going on is control of perception of sound (Bruce Abbott’s example), molecules are in motion with the aggregate effect that air pressure is varying (with probably complex changes in rate may be teased out into component sine waves by Fourier analysis) over a range (amplitude) which also may be varying. But in the experimental work that I have in mind what the observer measures is bands of acoustic energy, each roughly 400 Hz wide and attenuating at the upper and lower margins, at time t centered at roughly 290, 1900, and 2500 Hz., and then gradually changing over time until at time t’ the three bands are centered at roughly 450, 1550, and 2500 Hz. The subject (using means that we’ll ignore here) resists this change and restores the centers of acoustic energy to approximately their original values centered at 290, 1900, and 2500 Hz. The observer is monitoring and recording the numerical values. The subject is hearing the the sound of her own voice in headphones producing the vowel /i/ (as in “bid”) and gradually shifting to the vowel /É›/ (as in “bed”), and is acting in such a way as to hear the original sound of the vowel in “bid” in her headphones. The observer, while monitoring and recording the numerical values, is concurrently hearing the subject’s voice in the open environment first producing the /i/ of “bid” and then gradually shifting toward the vowel /æ/ in “bad”, and sees the centers or acoustic energy at about 450, 1550, and 2300 Hz.

The disturbance shifting the centers of energy in the acoustic spectrum affects only the sound heard in the headphones. The observer can also listen to the sound in the headphones, but for purposes of documenting results and creating a simulation the numbers are what matter.

Whenever the experimenter has the matching headphones on, both S.s and S.o may be represented by those numbers, labeled q.i. But q.i is a numerical representation of the perceptual input S.s (and sometimes S.o).Â

Here is an image of the sound spectrograms (with the formants for “bead” in there on the left side for good measure):


A note on “you’re being obtuse” meaning “you’re missing something obvious.” You’re right, of course, this refers to what you wrote and is not a comment on your intelligence! This somewhat shifted meaning appears to be gaining ground, at least in American usage. Googling “don’t be obtuse” I immediately found this corroboration:Â

“In American English, I think the meaning is shifting towards ‘missing the obvious’ which is part of the first definition at Dictionary.com.”

https://english.stackexchange.com/questions/37304/what-is-the-meaning-of-she-can-be-obtuse

It did seem to me that you were ignoring something obvious so I stated what I thought should be obvious. Maybe I am wrong, but that was my intent.

/Bruce

On Tue, May 29, 2018 at 4:28 PM, Martin Taylor csgnet@lists.illinois.edu wrote:

[Martin Taylor 2018.05.29.15.42]

[Bruce Nevin 2018-05-29_14:00:09 ET]

Martin Taylor 2018.05.29.09.09 –

        A single variable such as q.i can't mean two different

things at the same time in normal science. You have to
choose whether to use it for one or the other of the two
different things. In the case of q.i, some use it as a
variable in a control loop. But in this quote (and
frequently elsewhere) Rick uses it to refer to a perception
in a person who hypothesises that it is a perception of some
external variable that is controlled. These two are not the
same variable, and have the same value only by coincidence.

[…]

        The Powers diagrams and discussions say that q.i is the

variable C controls (theoretically by keeping a perception
of it as near its reference variable as C is able to do).
Rick says it is the variable T has determined by testing
various hypotheses to be the external variable C is most
likely to be controlling. They aren’t the same variable, so
we ought to agree on which one to call q.i in our
discussions. Is it the Powers version or the Marken version
that should prevail, and what should we call the other?

Martin, I fear that you’re being unnecessarily obtuse.

I'm not sure what you intend by "obtuse", which the OED and Random

House Unabridged define mainly in connection with angles, with a
metaphoric extension to dullness of the senses, which I suppose you
extend further to dullness of intellect. But if that is what you
intend, the word “unnecessarily” cannot apply, so all I can guess is
that you intend a pejorative comment with respect to my analysis.

My analysis provided just one among the variety of reasons provided

by myself and others why it is incorrect to say that q.i as a
perception in an observer/experimenter is (or even could be) the
same as a perception another person is controlling. Here’s the gist
of my argument. The rest of the “unnecessary” text was an effort to
be precise rather than applying to intuition on the part of the
reader.

Point 1. I allowed several impossibilities to show that even with

the observer/experimenter permitted to detect with the same values
everything that the subject perceives about the environment.

Point 2. No control can be perfect, so th observer-experimenter will

always see that control of ANY variable in the environment by the
subject is imperfect.

Point 3. The experimenter has no privileged access to the subject's

perceptual functions that are theorized to exist, so cannot know
just what functions of other environmental variables the subject may
use to create controlled perceptions.

Point 4. For any function of environmental variables there are

myriads of others that would seem to be controlled almost as well as
the one the subject is actually controlling.

Conclusion 1. The experimenter cannot know whether the function the

subject is controlling is among the hypotheses the experimenter is
comparing.

Conclusion 2. The experimenter cannot know whether the observed

imperfection of control for a given hypothesis about what the
subject is controlling is due to the hypothesis being wrong or to
inherent limits on the ability of the subject to control the
correctly hypothesized variable.

Final conclusion. The q.i that leads to a perception in the

experimenter is a different variable than the q.i that leads to a
perception being controlled by the subject.

=============



That argument gives the experimenter magical powers, which are still

insufficiently powerful to allow the experimenter’s perceived q.i to
be identified with the subject’s q.i or the subject’s perception of
q.i. If you reduce the experimenters access to magical powers, you
don’t even need the argument above, as various people have pointed
out:

(1) The subject's p may be a different function of q.i than the

experimenter’s.

(2) The experimenter cannot have the identical values of all

environmental variables as does the subject, and therefore is not
necessarily able to form a perception that is the same as the
subject’s for any aspect of the environment.

(3) The experimenter and the subject have had different life

experiences, and are unlikely to have reorganized to have the same
set of perceptual functions other than those that are genetically
determined by genes they have in common.

(4 related to 3) The experimenter and the subject have perceptual

experiences immediately before the study that are different, and

(5) Rupert's "If a tree falls in the forest when nobody is

listening" question: If there is no observer or experimenter, can
the subject keep controlling? If yes, then q.i cannot be a
perception in the experimenter, but must be something that the
subject perceives to be in the environment.

Any one of these would be sufficient to demonstrate that the q.i in

the environment that corresponds to a perception the subject is
controlling is a different variable than the one the experimenter
perceives and is testing. Your following comment goes further to
provide yet another argument for the same conclusion.

      The experimenter hypothesizes the value q.i as input to the

subject on the basis of a value that is input to the
experimenter. How else? Then the experimenter must test this
hypothesis by building a model, or two or three models with
different hypotheses about q.i.

When the possibilities are essentially uncountable in number, two or

three isn’t going to get you very far. You are talking about
optimizing one function so that it approaches a match to another
function. But the first is a map of the second, and the map gets
more and more complicated the higher the level in the hierarchy.
Most of us go along with Korzybski “The map is not the territory”.
The only point at issue is Rick’s repeated assertion that the map IS
the territory.

      By this process of hypothesis and test aims to know what

the really real value q.i is in the really real environment
that both the investigator and the subject are severally
perceiving (and controlling). Of course the hypothesized q.i
in the mind of the investigator is different from the q.i that
is perceived and controlled by the subject (and by the
investigator as well). What links them and brings them as
close as possible to unison is called science.

Apart from the parenthetical assertion that the investigator

controls q.i (presumably during the TCV), I have no disagreement
with this paragraph at all. The “of course” sentence indicates that
you do agree with the others who have commented on this issue – the
investigator’s q.i is not the subject’s q.i, and the two of them
should have different names in our discussions.

Bruce Abbott (2018.05.29.1245 EDT)–

        Let’s say that the participant is controlling perceived

sound intensity, p, which for him is a logarithmic function
of the physical sound intensity, q.i. Unbeknownst to us,
the experimenter is an intelligent alien species from one of
the planets orbiting Alpha Centuri. Although she possesses
a sense of hearing, her auditory input function makes p
directly proportional to q.i. So when the physical
intensity of the sound increases, it increases
logarithmically for the participant but linearly for the
experimenter.

      Your visiting scientist is foolishly making a fundamental

assumption that our human scientists are justified in making
but she is not, and that is the assumption that the human
scientist and the human subject are both structured in the
same way.

Surely a PCT theorist who believes in any form of learning or

reorganization would not make that assumption, would they? The
probable failure of the assumption is part of what makes the
experimenter’s function-optimization task so difficult.

      A human scientist should check this assumption, but how

often is that done?

Quite a lot, by conventional psychophysical researchers. Perhaps not

by the PCT-aware researchers who really ought to me even more
sensitive to the possibility that the assumption fails.

Martin
        On Tue, May 29, 2018 at 12:45 PM,

“Bruce Abbott” csgnet@lists.illinois.edu
wrote:

                [Bruce Abbott (2018.05.29.1245

EDT)]

Â

                        [Rick Marken 2018-05-28_21:46:30]

Â

                    RM:

In my reply to Rupert today I said:Â

Â

                      RM...

there seems to be general agreement that I am
completely wrong to think that q.i is the
controlled perception, p, from the observer’s
perspective. I think this is the only way to
see things if one is involved in doing
research on PCT. Since most everyone on
CSGNet is not involved in doing PCT research I
suppose the only problem with not seeing that
q.i as p from the observer’s perspective is
that you’ll keep getting into useless verbal
arguments about it with me. Â

Â

                      RM:

Upon reflection I realized that failure to
understand that the controlled quantity, q.i,
is the controlled perception, p, from the
observer’s perspective is not just a problem
for researchers. If you don’t understand this,
then you don’t understand what is fundamental
about PCT: that it is a theory that explains
the fact of control as it is seen in the
behavior of living systems.Â

Â

                      RM:

The controlled quantity is a variable that we
see being kept in a reference state, often
symbolized q.i*. Thus, variations (or lack
thereof) in the controlled quantity, q.i, and
the fact that q.i is being kept in a reference
state, q.i*, protected from disturbance, are
the data on which we base our conclusion that
the behavior we see involves control. As
Powers says on p. 175 of LCS “In these
reference states we have the heart of the
problem to which control theory is
addressed”.Â

Â

                      RM:

So PCT (which was then just called control
theory because that’s what it is) is the
theory that accounts for the observed fact
that organisms keep certain variables, q.i, in
reference states, q.i*. The theory accounts
for this fact by assuming that the organism
controls a perceptual signal, p, that is an
exact analog of q.i.Â

Â

                      RM:

It’s the fact of control (the fact that q.i is
observed to be maintained in a reference
state, q.i*, protected from normal
disturbances) that motivates the theory that
says that this observation is a result of
control of perception. The controlled
variable, q.i, is data. PCT says that this
data can be accounted for by a theory that
says that a perceptual signal that is an exact
analog of of the controlled variable is being
controlled. The perceptual signal, p, is
theory.Â

Â

                      RM:

So the controlled variable, q.i, is
unquestionably the observer’s perspective on
what the controller is presumed to be
perceiving (in theory). The idea that this is
not the case is a misunderstanding of PCT so
profound as to make even a detailed knowledge
of the theory perfectly useless for
understanding the behavior of living
organisms.Â

Â

                    Let’s say that the

participant is controlling perceived sound
intensity, p, which for him is a logarithmic
function of the physical sound intensity, q.i.Â
Unbeknownst to us, the experimenter is an
intelligent alien species from one of the
planets orbiting Alpha Centuri. Although she
possesses a sense of hearing, her auditory input
function makes p directly proportional to q.i.Â
So when the physical intensity of the sound
increases, it increases logarithmically for the
participant but linearly for the experimenter.

Â

                    The experimenter conducts the

Test for the controlled variable and observes
that her perception of q.i. is being held
relatively constant as q.i. is being subject to
a time-varying disturbance. (The participant
rotates a knob to “correct� perceived deviations
of the sound’s intensity.)

Â

                    Theoretically there is a

physical quantity “out there� in the shared
environment of the participant and the
experimenter, of which the perceptions of both
are a specific (though different) function. Two
questions:

Â

  1.                       Is q.i. the physical
    

quantity (sound-wave amplitude) that
constitutes the controlled quantity, or the
experimenter’s perception of it?

  1.                       Is there any way for
    

the experimenter to determine what the
participant’s input function for sound
intensity actually is, or must the
experimenter simply assume that it is similar
to her own?
Â

Bruce

Â

[Rick Marken 2018-06-02_22:20:45]

···

[Bruce Nevin 2018-05-31_18:12:04 ET]

BN: p is a numerical variable in a simulation which the theoretical model attributes to the organism. It is computed as a function of environmental variables, the same function of variables which specifies q.i. Therefore p = q.i by definition…

Â

BN: Martin: p. is a signal within the organism which is computed by sensors and neural structures in the organism…q.i is the net effect of certain environmental variables impinging upon the organism’s sensors and being transformed by neural structures so as to constitute the signal p within the organism. q.i and p are necessarily different…

BN: Martin will not be able to correct my representation of his views until sometime next week. I think the basis for how he and Rick are talking past each other will turn out to be a difference of reference along these lines.

RM: I don’t know if this captures it or not because I don’t understand what it is about q.i being “the net effect of certain environmental variables impinging upon the organisms sensors and being transformed by neural structures so as to constitute p within the organism” that makes q.i <> p? I’m having difficulty understanding what the “net effect” of environmental variables is.Â

RM: Also, could you send me the reference to the “formant control” study that you describe inÂ
[Bruce Nevin 2018-05-31_13:11:07 ET]?

Thanks

BestÂ

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Bruce Nevin 2018-06-03_08:59:53 ET]

Rick Marken 2018-06-02_22:20:45Â –

RM: I don’t know if this captures it or not because I don’t understand what it is about q.i being “the net effect of certain environmental variables impinging upon the organisms sensors and being transformed by neural structures so as to constitute p within the organism” that makes q.i <> p? I’m having difficulty understanding what the “net effect” of environmental variables is.Â

For the PCT experimenter (other than Henry Yin),
p is not directly observed, andÂ

q.i is a quantity established by controlling measure-and-quantify perceptions with regard to the relevant input to the subject’s sensors. From this point of view, their quantitative relationship is unknown.

For the modeler controlling perceptions of q.i as established by the observer, p is a quantity that is defined as equal to q.i for purposes of creating a generative simulation. Hence, from this point of view p = q.i with respect to the model.Â

For the neurophysiologist, for the psychophysicist, and for whoever is controlling for understanding q.i as being physically in the environment and p as being physically inside the organism, p is understood to be a rate of firing that is related to q.i in some regular way, but not identical to q.i. Being mindful of the complexity of firing rates in bundles of dendrites and axons, the distribution and branching of synapses, and differences in activation across the sensory endpoints affected by what is measured as q.i, a person taking this point of view sees the assertion that p=q.i as a postulate that properly serves the purposes of the modeler but cannot be literally true.

I proposed that a difference of role and point of view of this sort has been a source of sometimes rancorous disagreement.

RM: Also, could you send me the reference to the “formant control” study that you describe in  [Bruce Nevin 2018-05-31_13:11:07 ET]?

The references are Katseff 2010; Katseff & Houde 2008; Katseff, Houde, & Johnson 2008, 2010, as follows:

Katseff, Shira E. (2010). Linguistic constraints on compensation for altered auditory feedback. Ph.D. Dissertation, U.C. Berkeley.

http://linguistics.berkeley.edu/~shira/katseff_dissertation.pdf

Katseff, Shira & John F. Houde, (2008). Compensation ?=? Mental Representation. LabPhon11 abstracts, edited by Paul Warren, Wellington, NZ.

http://old.labphon.org/LabPhon11/publish/LP11%20abstracts/Katseff%20and%20Houde.pdf

Katseff, Shira, John F. Houde, & Keith Johnson, (2008). Partial compensation in speech adaptation. 2008 Annual Report of the UC Berkeley Phonology Lab, 444-461.

http://linguistics.berkeley.edu/phonlab/documents/2008/katseff_houde_annrpt08.pdf

Katseff, Shira, John F. Houde, & Keith Johnson, (2010). Auditory feedback shifts in one formant cause multi-formant responses. Journal of the Acoustical Society of America, 127.3:1955.

https://www.researchgate.net/publication/42439355_Auditory_feedback_shifts_in_one_formant_cause_multi-formant_responses

In this last reference, they talk about how disturbance of one formant is resisted by actions that have the measured effect of shifting another. It seems clear that for the subject this is the perceived effect of correcting the heard vowel when it doesn’t ‘sound right’. It may be that their borrowed equipment was only able to shift one frequency band at a time.

I have created a diagram of a model. One discussion is in my Stanford presentation

https://www.youtube.com/watch?v=nj8D9k7QCUU

If you don’t want to watch the whole thing, the relevant parts begin at about 5:30, introducing what Katseff and Houde saw as a puzzle, and again at about 28:50 modeling a PCT solution to the puzzle.

···

On Sun, Jun 3, 2018 at 1:21 AM, Richard Marken csgnet@lists.illinois.edu wrote:

[Rick Marken 2018-06-02_22:20:45]

[Bruce Nevin 2018-05-31_18:12:04 ET]

BN: p is a numerical variable in a simulation which the theoretical model attributes to the organism. It is computed as a function of environmental variables, the same function of variables which specifies q.i. Therefore p = q.i by definition…

Â

BN: Martin: p. is a signal within the organism which is computed by sensors and neural structures in the organism…q.i is the net effect of certain environmental variables impinging upon the organism’s sensors and being transformed by neural structures so as to constitute the signal p within the organism. q.i and p are necessarily different…

BN: Martin will not be able to correct my representation of his views until sometime next week. I think the basis for how he and Rick are talking past each other will turn out to be a difference of reference along these lines.

RM: I don’t know if this captures it or not because I don’t understand what it is about q.i being “the net effect of certain environmental variables impinging upon the organisms sensors and being transformed by neural structures so as to constitute p within the organism” that makes q.i <> p? I’m having difficulty understanding what the “net effect” of environmental variables is.Â

RM: Also, could you send me the reference to the “formant control” study that you describe inÂ
[Bruce Nevin 2018-05-31_13:11:07 ET]?

Thanks

BestÂ

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery