Concepts of control (RE: Lies (was On "variables" (was Re: Do we control "environmental variables"?)))

[EetuP]

(Sorry, I write too quickly, but have not much time for this. And I am afraid I have said this before.)

For me one problem seems to be different concepts of control. I can think at least three versions:

···

1. Keeping a variable in a certain (possibly changing) value (against possible disturbances).

2. Keeping a variable in a certain
   preset (possibly changing) value (against possible disturbances).

3. Keeping a variable in a certain (possibly changing) value (against possible disturbances) by keeping another variable in a certain
   preset (possibly changing) value (against possible disturbances).

The first one is Rick’s “scientific/engineering� or perhaps empiriristic/behavioristic (in a methodological, not theoretical way) version. An empirical “fact� which
is then explained with the second one.

The second one is a colloquial or naïve way to think – or a shortenedd way to talk also in PCT discussions. Its problem is that it assumes that we could set goals or
reference values straight in the environment. Or rather this version has two variant according to whether the variable is in the environment or inside the subject.

The third is the PCT enlightened version where the first variable is in the environment and the second inside the subject.

Does that make any sense?

Eetu

• Please, regard all my statements as questions,

no matter how they are formulated.

From: Martin Taylor csgnet@lists.illinois.edu
Sent: Saturday, May 19, 2018 11:46 PM
To: csgnet@lists.illinois.edu
Subject: Re: Lies (was On “variables” (was Re: Do we control “environmental variables”?))

[Martin Taylor 2018.05.19.16.38]

[From Bruce Abbott (2018.05.19.1620 EST)]

Bruce,

My comment was entirely on my impression that “control” means something different to you and to Boris. Your example was a good explanation of why the environmental variable is controlled as you understand control, but I thought that to Boris it would miss
the point entirely.

My explanation of a possible difference in interpretation was apparently wrong, at least in your case, and quite possibly in his. It’s just the way I understood your interpretations of the word “control” from the writings of both of you through the ages,
and I thought that if I were anywhere near correct, perhaps I could forestall a prolonged fruitless argument.

Let me ask you a question. I assumed that you took “the fact of control” as something that can be determined by observation of a variable subject to influences you can produce or observe. Specifically, you need not concern yourself with anything other than
the fluctuations of the magnitude of the variable and of the direct influences on it in order to perceive that it is or is probably controlled. This assumption as to your interpretation of the word lay behind my comment.

My question is: was I even close to being correct?

Martin

[Martin Taylor 2018.05.19.14.17]

[From Bruce Abbott (2018.19.1400 EDT)]

From: “Boris Hartman” (boris.hartman@masicom.net via csgnet Mailing List)
csgnet@lists.illinois.edu
Sent: Friday, May 18, 2018 10:42 AM
To: csgnet@lists.illinois.edu
Subject: FW: Lies (was On “variables” (was Re: Do we control “environmental variables”?))

[From Bruce Abbott (2018.05.15.1845 EDT)]

…

RY earlier :
Sure, a perceptual signal (q.i*g) may correspond to, or be a function of, variable aspects of the environment (q.i) but it is the perceptual signal that is controlled not the variable aspects of the environment

HB : It’s a perfect statement. When I saw it first time I wasn’t sure who wrote it. Bill or Rupert. I now know it was Rupert. It’s a PCT flower. Rupert has some really great statements on CSGnet.

I can support this statement with any means that is possible. With PCT means, biological, physiological, name it. But I can’t
think of any scientific mean that could support Bruce’s way of proving how “environmental variable and it’s analog perceptual signal” are both controlled. Maybe there is one “vague” way, but it got some other name, not control.

Let’s see if I can clarify it for you, Boris. … [explanation omitted]… Thus if p is controlled, then q.i. is also controlled.

I think Bruce and Boris are talking at cross-purposes. The way I read it, Bruce is taking an abstract viewpoint that control can be observed wherever it may occur, regardless of the mechanism by which it is achieved,
,

How did you reach that conclusion? My entire discussion centered on a specific example of a control system, with all the usual parts: input quantity, perceptual signal, reference (“set point�), comparator, error signal, output
function (drives the car’s throttle setting as a function of the error), feedback (throttle setting determines power, determines force exerted by the driving wheels on the road), feeding back onto the input quantity, the car’s speed.

whereas Boris is taking a mechanistic viewpoint in which the direction of causality matters. Both are using the same control diagram as a basis for their thinking.

The direction of causality matters in my description of the operation of the cruise control; how is this less mechanistic than the viewpoint that you presume Boris is taking?

Causality flows one way around a control loop. The difference between reference and perception causes the error, and the error causes the output that causes influence upon the thing being perceived. Therefore control of perception causes the stabilization of
the environmental variable (that an external observer sees in the abstract as control of the environmental variable).

I would say that control of perception does more than “stabilize� the environmental variable whose value determines the value of the perception. When the perception is simply the sensed value of the environmental value (e.g.,
car’s speed), and the latter determines the former, then the environmental variable is not just stabilized, it is controlled. (By the way, “stabilized� is a poor choice of words unless one is referring to regulators (systems with a fixed reference value).
When the reference varies (as in servo-mechanisms), both p and q.i. vary over time as p tracks the changes of reference.

Is the purpose of cruise control to control a perception of speed, or the car’s actual speed? Is it the purpose of power steering to control the sensed angle of the steering wheels, or their actual angle? Control of the latter
in each case is achieved by means of control of the former, but if both aren’t being controlled then I hope I’m not driving that car at the time!

Failure of the environmental to produce the desired value of the perception then finishes the causal loop, but the only forceful influence in the loop is that of the output on the environmental variable, compared to which the causal effect of the environmental
variable on the perception is a feather-touch.

Say again? I can’t make sense of the first part of the above sentence. As for the rest of it, true but irrelevant to the argument I have been making that control systems like cruise control to not just control their perceptions,
they control the environmental quantities of which those perceptions are functions.

That difference in force is what might make the difference between thinking of the perceptual variable as being controlled and the environmental variable just following along, as opposed to both being controlled because they vary in coordination with each other
and look statistically the same to an outside Analyst who can see both.

I would reverse that: the outside Analyst might see the environmental variable as being controlled (it, after all, is being directly influenced by the control system’s output) and the perceptual variable just being the way that
the control system represents that variable internally.

The wordings are in conflict, but are the ideas that lead to the words? The words are useful in keeping an argument going for a long time, but the ideas are useful to understanding.

Perhaps I misunderstand Boris, but I take him as saying that only perceptual signals are controlled, and my example was intended to clarify how the environmental variable from which p is derived may also be controlled when p is
controlled. In the simplest case, if p = q.i., and p is controlled, then q.i. is controlled.

Of course, I may have misunderstood both of you, but then that’s just words. Or is it?

That’s the real source of the problem: words. I understand that control systems control their perceptions. Somehow this has been taken to mean that they do
not control that which is being perceived (upon which the perception is based)!

Bruce

[EetuP]

Sorry, I must correct and add something. Delete the previous message.

For me one problem seems to be different concepts of control. I can think at least three versions:

···

1. Keeping a variable in a
   certain (possibly changing) value (against possible disturbances).

2. Keeping a variable in a certain
   preset (possibly changing) value (against possible disturbances).

3. Keeping a variable in a
   certain (possibly changing) value (against possible disturbances) by keeping another variable in a certain
   preset (possibly changing) value (against possible disturbances).

The first one is Rick’s “scientific/engineering� or perhaps empiriristic/behavioristic (in a methodological, not theoretical way) version. An empirical “fact� which
is then explained with the second version.

The second version has perhaps three sub-variants depending on what or where the variable is:

1. If the variable is thought to be in the environment then this is a colloquial or naïve way to understand control
   or

2. it can also be a shortened way to talk about the third version.

3. If the variable is inside the subject then this is the way of Boris and Rupert: The reference can be set only inside the subject and thus only variables inside the subject can be controlled – no matter how
   much some variables in tthe environment are affected according to the first version.

The third version (which should possibly be an official version in PCT) is perhaps the way of Martin and Bruce A (and is some statements of Rick) in these discussions.
The problem here can be that control seems to be understood as a double-pointed arrow or fight with two frontiers so that the variable in the environment and the variable inside the subject are both controlled equally. But the positions of the variables are
not equal, they have kind of a relation similar to that between means and ends. And so here we have still the initial problem and at least three alternatives:

1. Both variables are controlled (problem here is that “control� has two meanings!)

2. The environmental variable is controlled by the means of comparing the internal variable to the internal reference.

3. The internal variable (perception of course) is controlled (to the reference) by means of affecting the environmental variable.

Then I must still mention on possibility - along Rick’s (mathematical/theoretical?) assumption - that the environmental and internal (perceptual) variables were equal
because the latter is a function of the previous.

Does all that make any sense?

(In the field of education I am inured that people mean different things with same terms and same things with different terms… Not very easy situation.)

Eetu

• Please, regard all my statements as questions,

no matter how they are formulated.

From: Martin Taylor csgnet@lists.illinois.edu
Sent: Saturday, May 19, 2018 11:46 PM
To: csgnet@lists.illinois.edu
Subject: Re: Lies (was On “variables” (was Re: Do we control “environmental variables”?))

[Martin Taylor 2018.05.19.16.38]

[From Bruce Abbott (2018.05.19.1620 EST)]

Bruce,

My comment was entirely on my impression that “control” means something different to you and to Boris. Your example was a good explanation of why the environmental variable is controlled as you understand control, but I thought that to Boris it would miss
the point entirely.

My explanation of a possible difference in interpretation was apparently wrong, at least in your case, and quite possibly in his. It’s just the way I understood your interpretations of the word “control” from the writings of both of you through the ages,
and I thought that if I were anywhere near correct, perhaps I could forestall a prolonged fruitless argument.

Let me ask you a question. I assumed that you took “the fact of control” as something that can be determined by observation of a variable subject to influences you can produce or observe. Specifically, you need not concern yourself with anything other than
the fluctuations of the magnitude of the variable and of the direct influences on it in order to perceive that it is or is probably controlled. This assumption as to your interpretation of the word lay behind my comment.

My question is: was I even close to being correct?

Martin

[Martin Taylor 2018.05.19.14.17]

[From Bruce Abbott (2018.19.1400 EDT)]

From: “Boris Hartman” (boris.hartman@masicom.net via csgnet Mailing List)
csgnet@lists.illinois.edu
Sent: Friday, May 18, 2018 10:42 AM
To: csgnet@lists.illinois.edu
Subject: FW: Lies (was On “variables” (was Re: Do we control “environmental variables”?))

[From Bruce Abbott (2018.05.15.1845 EDT)]

…

RY earlier :
Sure, a perceptual signal (q.i*g) may correspond to, or be a function of, variable aspects of the environment (q.i) but it is the perceptual signal that is controlled not the variable aspects of the environment

HB : It’s a perfect statement. When I saw it first time I wasn’t sure who wrote it. Bill or Rupert. I now know it was Rupert. It’s a PCT flower. Rupert has some really great statements on CSGnet.

I can support this statement with any means that is possible. With PCT means, biological, physiological, name it. But I can’t
think of any scientific mean that could support Bruce’s way of proving how “environmental variable and it’s analog perceptual signal” are both controlled. Maybe there is one “vague” way, but it got some other name, not control.

Let’s see if I can clarify it for you, Boris. … [explanation omitted]… Thus if p is controlled, then q.i. is also controlled.

I think Bruce and Boris are talking at cross-purposes. The way I read it, Bruce is taking an abstract viewpoint that control can be observed wherever it may occur, regardless of the mechanism by which it is achieved,
,

How did you reach that conclusion? My entire discussion centered on a specific example of a control system, with all the usual parts: input quantity, perceptual signal, reference (“set point�), comparator, error signal, output
function (drives the car’s throttle setting as a function of the error), feedback (throttle setting determines power, determines force exerted by the driving wheels on the road), feeding back onto the input quantity, the car’s speed.

whereas Boris is taking a mechanistic viewpoint in which the direction of causality matters. Both are using the same control diagram as a basis for their thinking.

The direction of causality matters in my description of the operation of the cruise control; how is this less mechanistic than the viewpoint that you presume Boris is taking?

Causality flows one way around a control loop. The difference between reference and perception causes the error, and the error causes the output that causes influence upon the thing being perceived. Therefore control of perception causes the stabilization of
the environmental variable (that an external observer sees in the abstract as control of the environmental variable).

I would say that control of perception does more than “stabilize� the environmental variable whose value determines the value of the perception. When the perception is simply the sensed value of the environmental value (e.g.,
car’s speed), and the latter determines the former, then the environmental variable is not just stabilized, it is controlled. (By the way, “stabilized� is a poor choice of words unless one is referring to regulators (systems with a fixed reference value).
When the reference varies (as in servo-mechanisms), both p and q.i. vary over time as p tracks the changes of reference.

Is the purpose of cruise control to control a perception of speed, or the car’s actual speed? Is it the purpose of power steering to control the sensed angle of the steering wheels, or their actual angle? Control of the latter
in each case is achieved by means of control of the former, but if both aren’t being controlled then I hope I’m not driving that car at the time!

Failure of the environmental to produce the desired value of the perception then finishes the causal loop, but the only forceful influence in the loop is that of the output on the environmental variable, compared to which the causal effect of the environmental
variable on the perception is a feather-touch.

Say again? I can’t make sense of the first part of the above sentence. As for the rest of it, true but irrelevant to the argument I have been making that control systems like cruise control to not just control their perceptions,
they control the environmental quantities of which those perceptions are functions.

That difference in force is what might make the difference between thinking of the perceptual variable as being controlled and the environmental variable just following along, as opposed to both being controlled because they vary in coordination with each other
and look statistically the same to an outside Analyst who can see both.

I would reverse that: the outside Analyst might see the environmental variable as being controlled (it, after all, is being directly influenced by the control system’s output) and the perceptual variable just being the way that
the control system represents that variable internally.

The wordings are in conflict, but are the ideas that lead to the words? The words are useful in keeping an argument going for a long time, but the ideas are useful to understanding.

Perhaps I misunderstand Boris, but I take him as saying that only perceptual signals are controlled, and my example was intended to clarify how the environmental variable from which p is derived may also be controlled when p is
controlled. In the simplest case, if p = q.i., and p is controlled, then q.i. is controlled.

Of course, I may have misunderstood both of you, but then that’s just words. Or is it?

That’s the real source of the problem: words. I understand that control systems control their perceptions. Somehow this has been taken to mean that they do
not control that which is being perceived (upon which the perception is based)!

Bruce

[MartinT]

[Martin Taylor 2018.05.21.17.04]

Yes, but there's another way to look at the problem. This breakdown

is based on process. You use the word “keeping” in all the types.
You could also use a breakdown based on observation and theory,
using a form such as “observed (or theorized) that a variable is
kept”. Such a breakdown is related to yours, but is not identical.
Let’s assume, with Bill P and Rick, that “the fact of control” can
be observed 4. A variable is observed to be controlled (type 1 or 2 above) in
its own right.
     4a. Control is an all-or-none concept. A variable either is or
is not controlled.
     4b. Control is a concept that varies from exact control (not
realizable in practice) to no control.
5. A perceptual variable is theorized to exist and control of it is
theorized to be responsible for the variations observed in 1 or 2.
    5a. The observed variable is said to be controlled, and the
theorized perception is also controlled.
    5b. The observed variable is said NOT to be controlled, because
the theorized perceptual variable is controlled, and two variables
in the same loop cannot be controlled at the same time.
6. If a variable X is linked causally to a variable Y so that
changes in X are reflected in variable Y and X is controlled, then
    6a. Y is also said to be controlled.
    6b. Y is said NOT to be controlled.
I agree with you that how the word is used depends both on personal
preference and on context. Once upon Yes.
No, this is physically not the case. We conveniently use a “unit
transform” in the simplest kind of analysis of control systems, but
at best that is an approximation. At the very least, the actual
value of the environmental variable does not have a one-to-one
correspondence with the corresponding perceptual variable because
the perceptual resolution is finite. Slightly different values of
the environmental variable cannot be perceptually distinguished, and
because of inevitable noise effects (neural firings, for example),
the same environmental variable may produce different perceptual
values at different times. I learned this 50 years ago when a
colleague where I was on Sabbatical demonstrated that a short line
at a fixed slope flashed for a short time was perceived as a line
with a definite slope, but that perceived slope might be anywhere
within a large angle (30 to 60 degrees – I don’t remember the
detail). All perceptions have limited resolution, so even if we
ignore the qualitative difference between an environmental variable
such as a height and an internal representation such as a neural
firing rate, the two values cannot be equal except by chance or as a
rough approximation. That’s quite apart from the likelihood that the
perceptual value is a non-linear function, such as the logarithm, of
the environmental value.
Definitely yes.
No, and I think you hit on an importantpoint about the various
meaning people intend by the word “control” and the different
understandings others have when they read it or hear it.
Martin

···

On 2018/05/21 2:41 AM, Eetu Pikkarainen
( via csgnet Mailing List) wrote:

eetu.pikkarainen@oulu.fi

        Sorry, I must correct and add something. Delete

the previous message.

Â

        For me one problem seems to be different

concepts of control. I can think at least three versions:

1.           Keeping a variable in a
   

certain (possibly changing) value (against possible
disturbances).

1.           Keeping a variable in a certain
   

preset (possibly changing) value (against possible
disturbances).

1.           Keeping a variable in a
   

certain (possibly changing) value (against possible
disturbances) by keeping another variable in a
certain
preset (possibly changing) value (against possible
disturbances).

Â

        The first one is Rick’s

“scientific/engineering� or perhaps
empiriristic/behavioristic (in a methodological, not
theoretical way) version. An empirical “fact� which is then
explained with the second version.

Â

        The second version has perhaps three

sub-variants depending on what or where the variable is:

1.           If the
   

variable is thought to be in the environment then this is
a colloquial or naïve way to understand control
or

1.           it can
   

also be a shortened way to talk about the third version.

1.           If the
   

variable is inside the subject then this is the way of
Boris and Rupert: The reference can be set only inside the
subject and thus only variables inside the subject can be
controlled – no matter how much some variables in the
environment are affected according to the first version.

Â

        The third version (which should possibly be an

official version in PCT) is perhaps the way of Martin and
Bruce A (and is some statements of Rick) in these
discussions. The problem here can be that control seems to
be understood as a double-pointed arrow or fight with two
frontiers so that the variable in the environment and the
variable inside the subject are both controlled equally. But
the positions of the variables are not equal, they have kind
of a relation similar to that between means and ends. And so
here we have still the initial problem and at least three
alternatives:

1.           Both
   

variables are controlled (problem here is that “control�
has two meanings!)

1.           The
   

environmental variable is controlled by the means of
comparing the internal variable to the internal reference.

1.           The
   

internal variable (perception of course) is controlled (to
the reference) by means of affecting the environmental
variable.

Â

        Then I must still mention on possibility -

along Rick’s (mathematical/theoretical?) assumption - that
the environmental and internal (perceptual) variables were
equal because the latter is a function of the previous.

Â

Does all that make any sense?

Â

        (In the field of education I am inured that

people mean different things with same terms and same things
with different terms… Not very easy situation.)

Â

Eetu

          - Please, regard all my statements as

questions,

   no matter how they are formulated.

Â

Â

From: Martin Taylor <csgnet@lists.illinois.edu >
Sent: Saturday, May 19, 2018 11:46 PM
To: csgnet@lists.illinois.edu
Subject: Re: Lies (was On “variables” (was Re: Do
we control “environmental variables”?))

Â

[Martin Taylor 2018.05.19.16.38]

          [From Bruce Abbott

(2018.05.19.1620 EST)]

Bruce,

      My comment was entirely on my impression that "control" means

something different to you and to Boris. Your example was a
good explanation of why the environmental variable is
controlled as you understand control, but I thought that to
Boris it would miss the point entirely.

      My explanation of a possible difference in interpretation was

apparently wrong, at least in your case, and quite possibly in
his. It’s just the way I understood your interpretations of
the word “control” from the writings of both of you through
the ages, and I thought that if I were anywhere near correct,
perhaps I could forestall a prolonged fruitless argument.

      Let me ask you a question. I assumed that you took "the fact

of control" as something that can be determined by observation
of a variable subject to influences you can produce or
observe. Specifically, you need not concern yourself with
anything other than the fluctuations of the magnitude of the
variable and of the direct influences on it in order to
perceive that it is or is probably controlled. This assumption
as to your interpretation of the word lay behind my comment.

My question is: was I even close to being correct?

Martin

Â

Â

Â

        [Martin

Taylor 2018.05.19.14.17]

          [From Bruce

Abbott (2018.19.1400 EDT)]

Â

From:
“Boris Hartman” (boris.hartman@masicom.net
via csgnet Mailing List)
csgnet@lists.illinois.edu

              **Sent:** Friday, May 18, 2018 10:42 AM
              **To:** csgnet@lists.illinois.edu
              **Subject:** FW: Lies (was On "variables" (was Re:

Do we control “environmental variables”?))

Â

                      [From

Bruce Abbott (2018.05.15.1845 EDT)]

…

RY earlier :
Sure,
a perceptual signal (q.i*g) may correspond to, or be
a function of, variable aspects of the environment
(q.i) but it is the perceptual signal that is
controlled not the variable aspects of the
environment

Â

                HB : It's a perfect statement. When I saw

it first time I wasn’t sure who wrote it. Bill or
Rupert. I now know it was Rupert. It’s a PCT flower.
Rupert has some really great statements on CSGnet.

Â

I can support this statement with any means that is possible. With PCT
means, biological, physiological, name it. Â But I
can’t think of any scientific mean that could
support Bruce’s way of proving how “environmental
variable and it’s analog perceptual signal” are both
controlled. Maybe there is one “vague” way, but it
got some other name, not control.

Â

                Let's see if I can clarify it for you,

Boris. … [explanation omitted]…Â Thus if p is
controlled, then q.i. is also controlled.

Â

Â

        I

think Bruce and Boris are talking at cross-purposes. The way
I read it, Bruce is taking an abstract viewpoint that
control can be observed wherever it may occur, regardless of
the mechanism by which it is achieved,
,

          How did you reach that

conclusion? My entire discussion centered on a specific
example of a control system, with all the usual parts:
input quantity, perceptual signal, reference (“set
point�), comparator, error signal, output function (drives
the car’s throttle setting as a function of the error),
feedback (throttle setting determines power, determines
force exerted by the driving wheels on the road), feeding
back onto the input quantity, the car’s speed.

        whereas

Boris is taking a mechanistic viewpoint in which the
direction of causality matters. Both are using the same
control diagram as a basis for their thinking.

          The direction of causality

matters in my description of the operation of the cruise
control; how is this less mechanistic than the viewpoint
that you presume Boris is taking?

        Causality

flows one way around a control loop. The difference between
reference and perception causes the error, and the error
causes the output that causes influence upon the thing being
perceived. Therefore control of perception causes the
stabilization of the environmental variable (that an
external observer sees in the abstract as control of the
environmental variable).

          I would say that control of

perception does more than “stabilize� the environmental
variable whose value determines the value of the
perception. When the perception is simply the sensed
value of the environmental value (e.g., car’s speed), and
the latter determines the former, then the environmental
variable is not just stabilized, it is controlled. (By
the way, “stabilized� is a poor choice of words unless one
is referring to regulators (systems with a fixed reference
value). When the reference varies (as in
servo-mechanisms), both p and q.i. vary over time as p
tracks the changes of reference.

          Is the purpose of cruise control

to control a perception of speed, or the car’s actual
speed? Is it the purpose of power steering to control the
sensed angle of the steering wheels, or their actual
angle? Control of the latter in each case is achieved by
means of control of the former, but if both aren’t being
controlled then I hope I’m not driving that car at the
time!

        Failure of the environmental to produce the desired value of

the perception then finishes the causal loop, but the only
forceful influence in the loop is that of the output on the
environmental variable, compared to which the causal effect
of the environmental variable on the perception is a
feather-touch.

          Say again?  I can’t make sense of

the first part of the above sentence. As for the rest of
it, true but irrelevant to the argument I have been making
that control systems like cruise control to not just
control their perceptions, they control the environmental
quantities of which those perceptions are functions.

Â

        That

difference in force is what might make the difference
between thinking of the perceptual variable as being
controlled and the environmental variable just following
along, as opposed to both being controlled because they vary
in coordination with each other and look statistically the
same to an outside Analyst who can see both.

          I would reverse that: the outside

Analyst might see the environmental variable as being
controlled (it, after all, is being directly influenced by
the control system’s output) and the perceptual variable
just being the way that the control system represents that
variable internally.

        The wordings are in conflict, but are the ideas that lead to

the words? The words are useful in keeping an argument going
for a long time, but the ideas are useful to understanding.

          Perhaps I misunderstand Boris,

but I take him as saying that only perceptual signals are
controlled, and my example was intended to clarify how the
environmental variable from which p is derived may also be
controlled when p is controlled. In the simplest case, if
p = q.i., and p is controlled, then q.i. is controlled.

        Of course, I may have misunderstood both of you, but then

that’s just words. Or is it?

          That’s the real source of the

problem: words. I understand that control systems control
their perceptions. Somehow this has been taken to mean
that they do
not control that which is being perceived (upon
which the perception is based)!

Bruce

Â

[rsmarken]

[Rick Marken 2018-05-21_18:34:49]

···

On Sun, May 20, 2018 at 10:30 PM, Eetu Pikkarainen csgnet@lists.illinois.edu wrote:

EP: For me one problem seems to be different concepts of control. I can think at least three versions:

1. Keeping a variable in a certain (possibly changing) value (against possible disturbances).
2. Keeping a variable in a certain
   preset (possibly changing) value (against possible disturbances).
3. Keeping a variable in a certain (possibly changing) value (against possible disturbances) by keeping another variable in a certain
   preset (possibly changing) value (against possible disturbances).
   EP: The first one is Rick’s “scientific/engineeringâ€? or perhaps empiriristic/behavioristic (in a methodological, not theoretical way) version. An empirical “factâ€? which
   is then explained with the second one.

Â

EP: The second one is a colloquial or naïve way to think – or a shhortened way to talk also in PCT discussions. Its problem is that it assumes that we could set goals or
reference values straight in the environment. Or rather this version has two variant according to whether the variable is in the environment or inside the subject.

Â

EP: The third is the PCT enlightened version where the first variable is in the environment and the second inside the subject.

Â

EP: Does that make any sense?

RM: I agree that it is difficult to describe control verbally. I have used both definitions 1 and 2 at various times. But I’m not completely satisfied with them. The only problem with 3 (for me) is that it’s what Lewis Carroll would have called a portmanteau sentence; it binds a definition and (partial) explanation of control into one package. The part of the sentence before “by” is a definition of the observable phenomenon of control and the part after the “by” is a partial theoretical explanation of this observation as the control of “another variable” (which, of course, is the perceptual signal). So sentence 3 is certainly the “PCT enlightened” description of control since it contains both a description of the phenomenon that PCT explains (control) as well as a tantalizing portion of the PCT explanation of the mechanism that produces that phenomenon.Â

RM: Actually, I see 3 as an expansion of the title of Bill’s book “Behavior: The control of perception”. The first part of 3 (“keeping variables in certain (possibly changing) states in the face of disturbances” corresponds to “Behavior” and the second part of 3 (“by keeping perceptual variables matching reference specifications”) corresponds to “The control of perception”. [Pardon my slight re-phrasings of your words but they sound better to my old PCT ears].Â

BestÂ

Rick

Â

Eetu

• Please, regard all my statements as questions,

   no matter how they are formulated.

Â

Â

From: Martin Taylor csgnet@lists.illinois.edu
Sent: Saturday, May 19, 2018 11:46 PM
To: csgnet@lists.illinois.edu
Subject: Re: Lies (was On “variables” (was Re: Do we control “environmental variables”?))

Â

[Martin Taylor 2018.05.19.16.38]

[From Bruce Abbott (2018.05.19.1620 EST)]

Bruce,

My comment was entirely on my impression that “control” means something different to you and to Boris. Your example was a good explanation of why the environmental variable is controlled as you understand control, but I thought that to Boris it would miss
the point entirely.

My explanation of a possible difference in interpretation was apparently wrong, at least in your case, and quite possibly in his. It’s just the way I understood your interpretations of the word “control” from the writings of both of you through the ages,
and I thought that if I were anywhere near correct, perhaps I could forestall a prolonged fruitless argument.

Let me ask you a question. I assumed that you took “the fact of control” as something that can be determined by observation of a variable subject to influences you can produce or observe. Specifically, you need not concern yourself with anything other than
the fluctuations of the magnitude of the variable and of the direct influences on it in order to perceive that it is or is probably controlled. This assumption as to your interpretation of the word lay behind my comment.

My question is: was I even close to being correct?

Martin

Â

Â

Â

[Martin Taylor 2018.05.19.14.17]

[From Bruce Abbott (2018.19.1400 EDT)]

Â

From: “Boris Hartman” (boris.hartman@masicom.net via csgnet Mailing List)
csgnet@lists.illinois.edu
Sent: Friday, May 18, 2018 10:42 AM
To: csgnet@lists.illinois.edu
Subject: FW: Lies (was On “variables” (was Re: Do we control “environmental variables”?))

Â

[From Bruce Abbott (2018.05.15.1845 EDT)]

…

RY earlier :
Sure, a perceptual signal (q.i*g) may correspond to, or be a function of, variable aspects of the environment (q.i) but it is the perceptual signal that is controlled not the variable aspects of the environment

Â

HB : It’s a perfect statement. When I saw it first time I wasn’t sure who wrote it. Bill or Rupert. I now know it was Rupert. It’s a PCT flower. Rupert has some really great statements on CSGnet.

Â

I can support this statement with any means that is possible. With PCT means, biological, physiological, name it. But I can’t
think of any scientific mean that could support Bruce’s way of proving how “environmental variable and it’s analog perceptual signal” are both controlled. Maybe there is one “vague” way, but it got some other name, not control.

Â

Let’s see if I can clarify it for you, Boris. … [explanation omitted]…Â Thus if p is controlled, then q.i. is also controlled.

Â

Â

I think Bruce and Boris are talking at cross-purposes. The way I read it, Bruce is taking an abstract viewpoint that control can be observed wherever it may occur, regardless of the mechanism by which it is achieved,
,

How did you reach that conclusion? My entire discussion centered on a specific example of a control system, with all the usual parts: input quantity, perceptual signal, reference (“set pointâ€?), comparator, error signal, output
function (drives the car’s throttle setting as a function of the error), feedback (throttle setting determines power, determines force exerted by the driving wheels on the road), feeding back onto the input quantity, the car’s speed.

whereas Boris is taking a mechanistic viewpoint in which the direction of causality matters. Both are using the same control diagram as a basis for their thinking.

The direction of causality matters in my description of the operation of the cruise control; how is this less mechanistic than the viewpoint that you presume Boris is taking?

Causality flows one way around a control loop. The difference between reference and perception causes the error, and the error causes the output that causes influence upon the thing being perceived. Therefore control of perception causes the stabilization of
the environmental variable (that an external observer sees in the abstract as control of the environmental variable).

I would say that control of perception does more than “stabilizeâ€? the environmental variable whose value determines the value of the perception. When the perception is simply the sensed value of the environmental value (e.g.,
car’s speed), and the latter determines the former, then the environmental variable is not just stabilized, it is controlled. (By the way, “stabilizedâ€? is a poor choice of words unless one is referring to regulators (systems with a fixed reference value).Â
When the reference varies (as in servo-mechanisms), both p and q.i. vary over time as p tracks the changes of reference.

Is the purpose of cruise control to control a perception of speed, or the car’s actual speed? Is it the purpose of power steering to control the sensed angle of the steering wheels, or their actual angle? Control of the latter
in each case is achieved by means of control of the former, but if both aren’t being controlled then I hope I’m not driving that car at the time!

Failure of the environmental to produce the desired value of the perception then finishes the causal loop, but the only forceful influence in the loop is that of the output on the environmental variable, compared to which the causal effect of the environmental
variable on the perception is a feather-touch.

Say again? I can’t make sense of the first part of the above sentence. As for the rest of it, true but irrelevant to the argument I have been making that control systems like cruise control to not just control their perceptions,
they control the environmental quantities of which those perceptions are functions.

Â

That difference in force is what might make the difference between thinking of the perceptual variable as being controlled and the environmental variable just following along, as opposed to both being controlled because they vary in coordination with each other
and look statistically the same to an outside Analyst who can see both.

I would reverse that: the outside Analyst might see the environmental variable as being controlled (it, after all, is being directly influenced by the control system’s output) and the perceptual variable just being the way that
the control system represents that variable internally.

The wordings are in conflict, but are the ideas that lead to the words? The words are useful in keeping an argument going for a long time, but the ideas are useful to understanding.

Perhaps I misunderstand Boris, but I take him as saying that only perceptual signals are controlled, and my example was intended to clarify how the environmental variable from which p is derived may also be controlled when p is
controlled. In the simplest case, if p = q.i., and p is controlled, then q.i. is controlled.

Of course, I may have misunderstood both of you, but then that’s just words. Or is it?

That’s the real source of the problem: words. I understand that control systems control their perceptions. Somehow this has been taken to mean that they do
not control that which is being perceived (upon which the perception is based)!

Bruce

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