Elements in the PCT model

[Bj_rn_Simonsen]

[From Bjorn Simonsen (2006.11.14,19:45 EUST)]

From Rick Marken (2006.11.12.1100)

Maybe somebody’s comments after this mail
will explain the Real Real World better for me.

In many (most?) cases, disturbance variables are
not

perceived at all except via their effects on controlled variables.

This mail, Rick express uncertainty about what it is that
influence the disturbance variables so they become perceived and the
relationship between disturbances and perception.

Commenting your mail, Rick I stopped with this
sentence. And many questions turned up. And here they are. Later I will comment
your mail.

I will touch on two spheres and I will try to tell
when I am where.

First I will visit PCT and name an example. Remember
PCT is model of the brain’s internal organization.

When I stand with closed eyes and with my back to my
wife, I say to myself “I wish to look at her face”.

In the PCT model this wish is represented by a
reference value, let us say r = 5 signal units (impulses per second). This
value results in an output quantity. The output quantity is a physical unit and
in the PCT model this output quantity get its correct physical units by
multiplying with ko which is a constant with units “signal units per physical
unit” (qo = ko*(r – p)). I understand that qo is an expression that expresses
my muscular exercise.

This muscular exercise may result in unintended
effects and some actions. The actions have also their effects in the
environment. These effects may often be very extensive and other times they are
scarcely noticeable. The output effect must therefore be multiplied with a
Feedback Gain factor, just as Bill does it in his Live Block model (Feedback
effect :=FeedbackGain * OutputQuantity). I understand it as if the FeedbackGain
has a unit of “Input Quantity units per physical unit”.

Now I will say how I understand the Feedback Effect.
As the name expresses it is an effect. Remember the example I am talking about.
I started with closed eyes and wished to see the face of my wife. The output
quantity or my actions constitute the opening of my eyes, turning my face
toward my wife. If she for one or another reason will not let me see her face,
she starts leaving the room. Then I shout for her to stop and show me her face.
I open another door do come in front of her outside the house and so on. My
shout and my opening the other door may be the variation in the Feedback
Effect. The value of the Feedback Gain is great if my actions lead to great
effects and it is smaller if the effects are not that big.

Do I understand the Feedback Effect correct?

Now the Feedback Effect “has reached” the circle for Controlled
variable. This circle is a Simulation trick. It doesn’t exist in our
environment or outside our Sensors.

The face of my wife ( and her shoulders and her hat)
constitute the disturbance. Its units are special. Let us name them units of
disturbance. These units are very different from e.g. the references units.
Therefore it doesn’t tell us anything if the value of both the reference and
the disturbance is e.g. 5.

To get the value that the disturbance constitutes of
the Input quantity it must be multiplied by a constant with units like “Input
units per disturbance unit” (kd*d).

Now we have a value that reaches our sensors, the
Input quantity (qi = keqo + kdd)

This is a number that doesn’t represent any certain
thing in the environment. It represents the image I have in front of my eyes
before I have turned my face against my wife and it represents the changing
effect of this image when I turn around.

In the Input Function the Input quantity is reversed
to a perceptual signal by multiplying it with an Input Gain Factor. This is
done in Bill’s Live Block model. It is not done in B:CP. Here Bill say that the
value of the Input Gain Factor is 1 or he has implemented it in his ko value
(qo = ko*e).

Now we have a perceptual signal with units quite like
the reference units. The perceptual signal is different from the disturbance
units.

I think Bill lacks the Disturbance Gain Factor in his
Live Block Model. This results in a misunderstanding for my part. Let me
explain.

If I wish to see the face of my wife, I initiate with
a reference value like 5. I open my eyes and start turning around. Now the
disturbance has one or another value, but not 5 I think. Then there is an error
and an Output quantity or action that result in more turning around and a change
of the disturbance value.

When I have turned around and if I am lucky to see the
face of my wife, I thought that the disturbance got a value like 5.

The reason I thought like this, you see if you play
with Bill’s Live Block. Put the reference and the disturbance to the same value
5. Then you get a perceptual signal like 5, an error like zero and a Feedback
Effect like zero.

Then the Input quantity, qi = keqo + kdd also is 5.

The output quantity is zero and I don’t execute any
actions. I don’t need to execute any actions because I see what I wish to see.

Now back to my misunderstanding, because of what I
have written above. Reference = 5, disturbance = 5, error = zero, no actions
because I see what I wish to see, perceptual signal = 5 and qi = keqo + kdd
also is 5.

This tells me that I perceive the 5 disturbance units
(as the disturbance value), I perceive the Real Real World.

This is of course wrong. What I perceive is in reality
only the best I can perceive the Real Real World. It is not a one-to-one ”copy”.

If I play with Bill’s Live Block I can manipulate the
problem above by setting the disturbance = 6 (kd = 1.2 input quantities per
disturbance unit). Here I think the face of my wife is 6 disturbance units.

If I wished to see the face of my wife I initiated a
reference like 5. I turned around and saw the face of my wife, a disturbance
like 6 disturbance units.

If you play this Game on Bill’s Live Block, the error
becomes very near zero (-0.01). But the Output quantity (actions) exists with a
value like – 0.99 and a feedback effect like – 0.99. The perceptual signal is
quite near the reference value, but I can’t say that I perceive the face of my
wife, a disturbance like 6 disturbance units. Because my perception is
influenced by the feedback effect = -0.99. A feedback affect I don’t understand
what is.

Let me stop with the PCT model of the brain’s internal
organization and talk about neurology, the nerves and the environment.

How shall we understand the Output quantity (Actions)
and the Feedback Effect?

Is it correct to say that the body will move its
muscles as long as there is an error?

Is it correct to say that circumstances in the
environment may result in different Feedback Effects?

Is it correct to say that the Feedback Effect doesn’t
exist in the world of neurons and the Real Real World around?

Is it correct to say that the Feedback Effect is the
Effect that makes the disturbance most like the reference?

Is it correct to say that there is more balance
between the perception and the disturbance the less the Feedback Effect is?

bjorn

[Bill_Powers1]

When I stand with closed eyes
and with my back to my wife, I say to myself “I wish to look at her
face”. In the PCT model this wish is represented by a reference value,
let us say r = 5 signal units (impulses per second).
[From Bill Powers (2006.11.14.1457 MST)]

Bjorn Simonsen
(2006.11.14,19:45 EUST) –

I think your difficulties stem mainly from trying to use too simple a
model for the complex situation you have chosen as an example. The Live
Block Diagram represents a situation in which there are continuous
functions relating the output quantity, the disturbance, and the input
quantity over some range of contiguous values. In your example of turning
around to see your wife these relationships are not present.

When you begin, you say

Unfortunately,this system could not work. There is no perceptual signal
representing your wife when you open your eyes. You see some part of the
room, perhaps a wall and a window, but no wife. You require a
higher-order control system to find your wife before you can establish
the kind of simple control relationship shown in the Live Block Diagram.
For example, there might be a system for doing a scan, with the scan
being stopped when the image of your wife comes into the field of view.
Once that image is visible, its location can be perceived relative to the
center of vision, and a control system for reducing that distance to zero
can operate by rotating the eyes, head, and body. Of course that
position-control system doesn’t know that the object it is perceiving is
a “wife”. It’s just an object, defined by contrast with the
background. Only your higher-order systems know anything about this
object but its location in the visual field.

The Live Block Diagram illustrates some basic properties of a
continously-operating analog control system. To represent the complex
task of turning around to look at your wife, you would need several
levels of control, with multiple control systems at each level. A single
block diagram can’t even come close to modeling all that must happen. It
shows how a single dimension of control of a single variable happens. It
can’t even show how you would center your eyes on the wife-object both
horizontally and vertically at the same time.

You must also remember that this model can’t explain how the various
aspects of the scene come to be represented as perceptual signals. How
does the brain pick out an object from a scene full of visual
intensities? How is position represented, considering that the retinal
image is affected but the orientation of the eyes, head, torso, legs, and
feet? What coordinate system or systems does the brain use for
calculating distances and positions?

If you can define a control task so there is only a single
one-dimensional controlled variable (as in a tracking task) you can apply
the control system equations successfully and make good predictions. But
the examples you are using involve multiple dimensions of control and
multiple levels of control, so you can’t apply the control equations in
such a simple way.

Best,

Bill P.

[Martin_Taylor2]

[Martin Taylor 2006.11.14.17.30]

[From Bjorn Simonsen (2006.11.14,19:45 EUST)]
From Rick Marken (2006.11.12.1100)

Maybe somebody's comments after this mail will explain the Real Real World better for me.

>In many (most?) cases, disturbance variables are not

perceived at all except via their effects on controlled variables.

This mail, Rick express uncertainty about what it is that influence the disturbance variables so they become perceived and the relationship between disturbances and perception.

I think you misunderstand. What Rick is saying is that the ONLY thing represented by the perceptual signal of a control system is the summation of the output and the disturbance. The perceptual signal does not represent the output, and it does not represent the disturbance. There's no uncertainty about that simple fact. There is no "relationship" between perception and disturbance, other than that the disturbance contributes to the perception, as does the output.

You seem to have an unconventional view of what a "disturbance" is. You can observe all sorts of aspects of the outer world without controlling for them. If you aren't controlling a perception, all you have is input from the outer world. There is no "disturbance". A disturbance is some influence on a perceptual signal from the outer world that causes its state to deviate from the value for which you are controlling that perception. No control, no disturbance.

When I stand with closed eyes and with my back to my wife, I say to myself "I wish to look at her face".
In the PCT model this wish is represented by a reference value, let us say r = 5 signal units (impulses per second). This value results in an output quantity. The output quantity is a physical unit

Still "impulses per second", though.

and in the PCT model this output quantity get its correct physical units by multiplying with ko which is a constant with units "signal units per physical unit" (qo = ko*(r - p)). I understand that qo is an expression that expresses my muscular exercise.

No. That's true only for the most peripheral control systems. All other control systems supply their output signals to the reference inputs of lower-level control systems. Your "perceiving her face" control system is far above that level. But eventually, many, many muscular control systems will be providing outputs that do move muscles. You can't, however, convert those muscular contractions back into the single "qo" from your "face-seeing" control system.

This muscular exercise may result in unintended effects and some actions. The actions have also their effects in the environment. These effects may often be very extensive and other times they are scarcely noticeable. The output effect must therefore be multiplied with a Feedback Gain factor, just as Bill does it in his Live Block model (Feedback effect :=FeedbackGain * OutputQuantity). I understand it as if the FeedbackGain has a unit of "Input Quantity units per physical unit".

You can't really do that in any simple way. Your muscular exertions do have effects on the environment, but all you sense is the impingements of photons on your retinal receptors, the movements of air molecules hitting your eardrum, and the like. There's no translation from newtons of muscular effort to the energy of photons and of air molecules, and it wouldn't be useful even if there were. Where you have to look for this "Feedback Gain" is in the "impulses per second" part of the loop.

Remember the example I am talking about. I started with closed eyes and wished to see the face of my wife. The output quantity or my actions constitute the opening of my eyes, turning my face toward my wife.

Not really. The output quantity provides reference signals for several other control systems, and they in turn for yet lower level ones. Only the combined outputs of several of the very lowest level ones effect the actions you list.

If she for one or another reason will not let me see her face, she starts leaving the room. Then I shout for her to stop and show me her face. I open another door do come in front of her outside the house and so on. My shout and my opening the other door may be the variation in the Feedback Effect. The value of the Feedback Gain is great if my actions lead to great effects and it is smaller if the effects are not that big.
Do I understand the Feedback Effect correct?

I don't think so. You here deal with the effects of many fairly high-level control systems acting in parallel, each with its own disturbances, all providing inputs to the "see her face" system. One might be a system with a reference value for being in a place with no opaque barriers betwee her and you. That system had no error until she left the room, but then it did have error, and generated output that altered the reference values in other control systems.

The face of my wife ( and her shoulders and her hat) constitute the disturbance.

They might provide the input to a perception. To what controlled variable might they constitute a disturbance?

To get the value that the disturbance constitutes of the Input quantity it must be multiplied by a constant with units like "Input units per disturbance unit" (kd*d).

I can't see a constant here. Many processes that may be time-variant functions intervene between the impingement of photons on the retina and the creating of a perception, so at the very least you would have to posit a functional relationship. Since there are usually many different contributions to a perceptual imput function, the function is many-to-one, and not invertible. It's not even clear to me that you could do a dimensional analysis.

Now we have a perceptual signal with units quite like the reference units. The perceptual signal is different from the disturbance units.

I presume you mean that the perceptual signal is "impulses per second" while the external environment represented by that signal involves many different kinds of dimension. That's true, but again, there's no obvious route to imagining a conversion factor.

What I perceive is in reality only the best I can perceive the Real Real World. It is not a one-to-one "copy".

True. Assuming that the PCT model is reasonably correct, for each individual perceptual signal, it's a many-to-one time-varying function -- a variable filter, if you like.

Let me stop with the PCT model of the brain's internal organization and talk about neurology, the nerves and the environment.

How shall we understand the Output quantity (Actions) and the Feedback Effect?
Is it correct to say that the body will move its muscles as long as there is an error?

Not necessarily, even in the simple PCT single loop model.

Is it correct to say that circumstances in the environment may result in different Feedback Effects?

Yes.

Is it correct to say that the Feedback Effect doesn't exist in the world of neurons and the Real Real World around?

Try rephrasing hat question. It makes no sense to me as a question.

Is it correct to say that the Feedback Effect is the Effect that makes the disturbance most like the reference?

I don't understand that either, unless you are asking the simple question "Does a control system control"? In which case the answer is "yes".

Is it correct to say that there is more balance between the perception and the disturbance the less the Feedback Effect is?

What does a "balance between the perception and the disturbance" mean? Any disturbance is a contributor to the controlled perception. How can the two "balance" each other?

I'm afraid this may not be too helpful, since my answers clearly are based on an understanding of PCT that is very different from yours. I may be "writing in Chinese in answer to your question in Swahili".

Martin

[Bj_rn_Simonsen]

[Bjorn Simonsen (2006.11.15,11:30 EUST) --]

From Bill
Powers (2006.11.14.1457 MST)

I think your difficulties stem mainly from trying to use

too simple a model for the complex situation you have

chosen as an example. The Live Block Diagram represents

a situation in which there are continuous functions

relating the output quantity, the disturbance, and the input

quantity over some range of contiguous values. In your

example of turning around to see your wife these

relationships are not present.

Here may be a misunderstanding here. I have never
thought that your Live Block Diagram should explain or describe what you call
my complex situation. I thought upon the Live Block Diagram as a composite
System Diagram.

When I stand with closed eyes and with my back
to my wife, I say to myself “I >>wish to look at her face”. In the PCT
model this wish is represented by a reference >>value, let us say r = 5
signal units (impulses per second).

Unfortunately,this
system could not work. There is no
perceptual signal representing your wife when you open
your eyes. You see some part of the room, perhaps a
wall and a window, but no wife. You require a higher-order
control system to find your wife before you can
establish the kind of simple control relationship shown
in the Live Block Diagram.

Sending my
previous mail I was more in a philosophical mood than in a
Simulation/Construction mood.

Let me repeat
myself. Is it possible, in accordance with PCT, to think that my wish to see
her face is represented with a reference like one or another number e.g. 5 (or
a matrix of references with their values) in a composite Diagram?

Is it further
OK that I set the disturbance like 6 or something else representing a room with
walls and windows and no wife? Is it further OK to think there becomes an error
and output quantity (actions)?

For
example, there might

be a system for doing a scan, with the scan being stopped

when the image of your wife comes into the field of

view. Once that image is visible, its location can be

perceived relative to the center of vision, and a control

system for reducing that distance to zero can operate by

rotating the eyes, head, and body. Of course that

position-control system doesn’t know that the object it

is perceiving is a “wife”. It’s just an object, defined by

contrast with the background. Only your higher-order

systems know anything about this object but its location

in the visual field.

This is OK.
This is how I think a composite System Diagram could function. My point is that
when I turn around I see walls and windows from different visual angles and no
wife.

After a moment
I see her face, I see what I wish to see. Now I stop turning around, I stop my
actions because the error is zero or near zero. Now the perceptual signal in
the composite system is like the reference and the disturbance has its value.
Am I wrong thinking this way? Remember I am in a philosophical mood.

The Live Block Diagram illustrates some basic

properties of a continously-operating analog control

system. To represent the complex task of turning

around to look at your wife, you would need several

levels of control, with multiple control systems at

each level. A single block diagram can’t even come

close to modeling all that must happen. It shows

how a single dimension of control of a single

variable happens. It can’t even show how you would

center your eyes on the wife-object both horizontally

and vertically at the same time.

I must have
expressed myself very unfortunate in my previous mail. I think the way you
think.

You must also remember that this model can’t explain

how the various aspects of the scene come to be

represented as perceptual signals. How does the brain

pick out an object from a scene full of visual intensities?

How is position represented, considering that the retinal

image is affected but the orientation of the eyes, head,

torso, legs, and feet? What coordinate system or

systems does the brain use for calculating distances

and positions?

Yes, I
understand your first sentence and I will remember. I guess there are many,
many different neurons different places in the brain that picks out an object
from a scene full of visual intensities. I also guess I in a PCT way may
explain that many, many loops control their perceptions. When I talk about such
situations and PCT I think upon the System Model as a composite model changing
its values over time. Is there anything that tells me I shall not think so?

If you can define a control task so there is only a

single one-dimensional controlled variable (as in a

tracking task) you can apply the control system

equations successfully and make good predictions.

But the examples you are using involve multiple

dimensions of control and multiple levels of control,

so you can’t apply the control equations in such a simple way.

I understand
this very well. Thank you for cared about my mail. May I ask some of the
questions I asked in my previous mail?

If I think
upon the PCT System Diagram as a composite Diagram, is it correct to think upon
the Feedback Effect as the effect my actions have on the scene I am looking.

Here is an
example about what I mean.

A pupil can’t
just leave the classroom. He asks the teacher to leave. This question is a
disturbance for the teacher. The disturbance results in a perceptual signal
(perception), and an error. This error result in an Output variable (action),
the words “Yes please”. The Feedback effect in this instance is that the pupil
is allowed to leave the classroom.

Is it
correct to say that the Feedback Effect doesn’t exist in the world of neurons
and the Real Real World around?

bjorn

[Bj_rn_Simonsen]

[Bjorn Simonsen (2006.11.16,11:35 EUST)]

[From Rick Marken (2006.11.15.1900)]

Bjorn Simonsen (2006.11.15,11:30 EUST)

Here is an example about what I mean.

A pupil can’t just leave the classroom. He
asks the teacher to leave.

This question is a disturbance for the teacher.
The disturbance

results in a perceptual signal (perception),
and an error. This error

result in an Output variable (action), the
words “Yes please”. The

Feedback effect in this instance is that the
pupil is allowed to leave

the classroom.

A pupil can certainly “just” leave the
classroom.

Rick, Why do you object to mine example. It was an
example.

I must have expressed myself very diffuse, because
your answer contains the Feedback Effect in the composite feedback loop
describing the teacher’s control. I hoped for an answer containing the Feedback
Effect in the composite feedback loop describing the pupil’s control. But that
doesn’t matter now. I understand your answer: >The feedback effect in this
example is the teacher hearing

herself answer the student’s question; the teacher’s answer

bring the perception of number of questions left
to be

answered from one back to none.

that the Feedback Effect at the lowest level, after
the actions, is the abstraction saying that the number of questions left to be
answered from one back to none.

After studying Martin’s comments I am inclined to say
that the Feedback Effect box at higher levels is out of the model (?).

Now I’ll try to say the same thing again in
Norwegian…

Kidding;-)

Your last sentence is a disturbing for mine
control “I wish to understand the meaning of your last sentence. The
disturbance (the text) is not equal to my perception of the text. I don’t
perceive myself Kidding, my perception is more like uncertainty if you are
Kidding. The disturbance contributes the perception, as does the output. (The
last sentence is a Feedback Effect after my control “I wish to understand what
Martin writes in his mails).

Back to business.

I often play with Bill’s Live Block. And I
have really problems understanding the effect of changing the Feedback gain fro
0.5 to 1.5

I will talk about this in another thread.

bjorn

[Bj_rn_Simonsen]

[From Bjorn Simonsen (2006.11.16,12:45 EUST)]

Martin Taylor 2006.11.14.17.30

[From Bjorn Simonsen (2006.11.14,19:45 EUST)]

From Rick Marken (2006.11.12.1100)

In many (most?) cases, disturbance
variables are not

perceived at all except via their effects
on controlled variables.

This mail, Rick express uncertainty about what
it is that influence

the disturbance variables so they become
perceived and the

relationship between disturbances and
perception.

I think you misunderstand. What Rick is saying is
that the ONLY thing

represented by the perceptual signal of a control
system is the

summation of the output and the disturbance. The
perceptual signal

does not represent the output, and it does not
represent the

disturbance. There’s no uncertainty about that
simple fact. There is

no “relationship” between perception and
disturbance, other than that

the disturbance contributes to the perception, as
does the output.

Thank you if this is Rick’s meaning.

You seem to have an unconventional view of what a
“disturbance” is.

You can observe all sorts of aspects of the outer
world without

controlling for them. If you aren’t controlling a
perception, all you

have is input from the outer world. There is no “disturbance”.
A

disturbance is some influence on a perceptual
signal from the outer

world that causes its state to deviate from the
value for which you

are controlling that perception. No control, no
disturbance.

You taught me something new here. I have thought upon
all physical variables reaching my sensors through the (in the PCT model)
circle for Input quantity as “disturbance” and “Feedback Effect”. I have never thought upon “Inputs from the
outer world” I hope my view of what a “disturbance” is will be less
unconventional in the future.

When I stand with closed eyes and with my back
to my wife, I say to

myself “I wish to look at her face”.

In the PCT model this wish is represented by a
reference value, let

us say r = 5 signal units (impulses per
second). This value results

in an output quantity. The output quantity is
a physical unit

Still “impulses per second”, though.

Yes, OK, but I have a tendency to think upon a
composite System Diagram.

and
in the PCT model this output quantity get its correct physical

units by multiplying with ko which is a
constant with units "signal

units per physical unit" (qo = ko*(r -
p)). I understand that qo is

an expression that expresses my muscular
exercise.

No. That’s true only for the most peripheral
control systems.

Isn’t it also true when we in a philosophical mood
talk about a composite system?

This muscular exercise may result in
unintended effects and some

actions. The actions have also their effects
in the environment.

These effects may often be very extensive and
other times they are

scarcely noticeable. The output effect must
therefore be multiplied

with a Feedback Gain factor, just as Bill does
it in his Live Block

model (Feedback effect :=FeedbackGain *
OutputQuantity). I

understand it as if the FeedbackGain has a
unit of "Input Quantity

units per physical unit".

You can’t really do that in any simple way. Your
muscular exertions

do have effects on the environment, but all you
sense is the

impingements of photons on your retinal receptors,
the movements of

air molecules hitting your eardrum, and the like.
There’s no

translation from newtons of muscular effort to the
energy of photons

and of air molecules, and it wouldn’t be useful
even if there were.

Where you have to look for this “Feedback
Gain” is in the "impulses

per second" part of the loop.

I haven’t thought upon it this in this way. Neither I
have heard about any translations from newtons of muscular effort to energy of
photons and energy of air molecules. But there is a feedback signal from the
tendons and there I thought we could use Feedback effect :=FeedbackGain *
OutputQuantity.

Remember the example I am talking about. I
started with closed eyes

and wished to see the face of my wife. The
output quantity or my

actions constitute the opening of my eyes,
turning my face toward my

wife.

Not really. The output quantity provides reference
signals for

several other control systems, and they in turn
for yet lower level

ones. Only the combined outputs of several of the
very lowest level

ones effect the actions you list.

Doesn’t my composite system thinking explain my
expression?

The face of my wife ( and her shoulders and
her hat) constitute the

disturbance.

They might provide the input to a perception. To
what controlled

variable might they constitute a disturbance?

My wish to see my wife’s face.

To get the value that the disturbance
constitutes of the Input

quantity it must be multiplied by a constant
with units like "Input

units per disturbance unit" (kd*d).

I can’t see a constant here. Many processes that
may be time-variant

functions intervene between the impingement of
photons on the retina

and the creating of a perception, so at the very
least you would have

to posit a functional relationship. Since there
are usually many

different contributions to a perceptual imput
function, the function

is many-to-one, and not invertible. It’s not even
clear to me that

you could do a dimensional analysis.

I agree that the disturbance variables may influence
Input quantity as time variant functions do. Therefore it was wrong when I
wrote “To get the value that the disturbance constitutes of the Input quantity
it MUST be multiplied by a CONSTANT …”

I’m afraid this may not be too helpful, since

my answers clearly are based on an understanding

of PCT that is very different from yours. I may be

“writing in Chinese in answer to your question in Swahili”.

Everything you write Martin is helpful. I
think it happens when I read your mails that I understand how you understand
PCT. I am sorry that I express myself
in Swahili, but I understood your Chinese very well this time.

bjorn

[Bill_Powers1]

[From Bill Powers (2006.11.16.0735 MST)]

Bjorn Simonsen
(2006.11.15,11:30 EUST) – –

I think that a lot of confusion in this discussion arises from
inexact use of English. For example, both you and Rick seem to agree that
if “a student asks the teacher to leave the room,” it is the
student who will leave the room if the answer is yes. But the normal
meaning of “A asks B to do X”, is that B, not A, is being
asked to do X. There are several ways to make sure the right subject is
being designated: “The student asks the teacher for permission to
leave the room,” or less gracefully, “The student asks the
teacher if the teacher will please leave the room.” Even for native
English speakers, the use of pronouns can get very sloppy, with no
indication of which preceding noun the pronoun is intended to indicate.
The fact that the writer knows is no help to the reader.

[One way this happens in posts is that someone will cite a long paragraph
in which the writer makes a series of different statements, and then add
the comment, “That’s right!” To which statement does
“that” refer? ]

You say, "Now the perceptual signal in the composite system is like
the reference and the disturbance has its value. Am I wrong thinking this
way? "

What way is “this”? Your sentence says that the disturbance has
the same value as the reference signal, or perhaps the perceptual signal.
That is wrong. But perhaps that is not what you thought you were saying
in English.

The pronoun “its” in the phrase “its value” refers to
some previous noun, and we have the choices (from right to left) of
disturbance, reference (signal), and perceptual signal. Used that
way, the usual meaning would not be that the disturbance has the value of
the disturbance, but that it has the value of some other preceding noun,
either reference or perception. If you meant to say that the disturbance
has the value of the disturbance, you would say “the disturbance has
its own value” or just “a value.” That, in fact, would be
correct. The disturbance has a value that is independent of the values of
both the perceptual signal and the reference signal. Of course if you
said that last sentence as I just did, there would be no ambiguity caused
by pronouns.

“Like” is vague in that it doesn’t specify the manner in which
A is like B. You say the perceptual signal is like the reference, but
does that mean the perceptual signal specifies the intended state of a
controlled variable as the reference signal does? Does it mean the
perceptual signal, like the reference signal, is a train of neural
impulses? What the sentence should say is that the perceptual signal has
(almost) the same value as the reference signal when the steady state is
reached. “Like” is the wrong word for indicating numerical
similarity.

Here’s another of your comments (to Martin Taylor):

You taught me something new
here. I have thought upon all physical variables reaching my sensors
through the (in the PCT model) circle for Input quantity as “disturbance”
and “Feedback Effect”. I have never thought upon “Inputs from the
outer world” I hope my view of what a “disturbance” is will be less
unconventional in the future.

The circle for input quantity indicates a physical variable, outside
the control system, that is affected by both the disturbance and
the feedback effect, additively. Everything below the horizontal line
separating “system” from “environment” is in the
outer world relative to the controlling system. For example, consider the
lateral position of a car on the road, which we treat as a variable that
exists in the outer world. That position is affected by forces from the
front wheels of the car and by forces from side winds, these forces also
being in the outer world. The input quantity is the car’s position; the
feedback effect is an adjustable lateral force generated by pointing the
front wheels of the car to one side or the other as the car travels, and
the disturbance is the force on the car caused by a wind coming from the
right or left side. The perceptual signal inside the driver depends ONLY
on the position of the car relative to the center of its lane on the
road; there is no way for the driver to perceive either of the forces
acting on the car, or to distinguish the effects of one force from the
effects of the other. I repeat that last statement in other
words:

Qi = d + Ff(qo). The magnitude of the input quantity is equal to
the sum of the magnitudes of the disturbing variable and the feedback
function of the output quantity. But the perceptual signal depends on qi
alone, so there is no way for it to represent either d or Ff(qo)
separately. If d is 3 and Ff(qo) is 4, then qi is 7. But qi is also 7 if
d = 75 and Ff(qo) = -68. The value of the input quantity (or the
perceptual signal) does not indicate either the value of d or the value
of Ff(qo). The input quantity and perceptual signal will be the same for
any combination of d and Ff(qo) that adds up to the same magnitude.

···

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

I think that many of the problems of understanding come from
grammatical differences between languages. In the grammar of some
languages, “John hit Bill” means that Bill was the one who
lifted his fist and directed it onto John’s nose, whereas in English it
means the opposite. “Disturbance affects perception” might be
read, using Swedish grammar (I don’t know if this is true), as saying
that the perception is the causal agent and that it has an effect on the
disturbance. When different languages used different word-order
conventions, there is a high risk of confusion in a technical
discussion.

Also, there are problems with little words. the British say
“different to” whereas in American English we say
“different from” Translating these small words from one
language to another can result in changes of meaning going both
ways.

The point here is that speakers of foreign languages communicating in
English need to acquire detailed knowledge of similarities and
differences between the languages, and that those who hear or read those
communications should know enough about the other language to detect
possible miscommunications, or at least see where ambiguity of
translation is possible. I suspect that Dag Forssell can tell us quite a
bit about that, in relation to Swedish at least. And of course Bruce
Nevin is our language guru (not to leave out Martin Taylor who seems to
have some expertise in this regard – or anyone else I’ve overlooked). By
the way, I’ve seen British mysteries in which “overlook” is
used as Americans would use “oversee.” Overlook, American
style, includes the meaning of “fail to see” or
“miss”, and “oversee” means “supervise, maintain
observation or management” of something or someone. Heck, in
English, “loosen” means the same thing as “unloosen”,
and “flammable” means the same as
“inflammable”.

The most important feedback loop in communication is the effect of the
speaker’s words on the speaker’s own perceptions. Even just in one’s
native tongue, reading what one has just written can reveal alternate
meanings that are far from those that are intended. Guess which meanings
the reader will pick. Throw in language differences, and it’s a wonder
that such things as international journals can exist at all.

Best,

Bill P.

[Martin_Taylor2]

[From Bjorn Simonsen (2006.11.16,12:45 EUST)]
Martin Taylor 2006.11.14.17.30

>>[From Bjorn Simonsen (2006.11.14,19:45 EUST)]

>>When I stand with closed eyes and with my back to my wife, I say to
>>myself "I wish to look at her face".
>>In the PCT model this wish is represented by a reference value, let
>>us say r = 5 signal units (impulses per second). This value results
>>in an output quantity. The output quantity is a physical unit

>Still "impulses per second", though.
Yes, OK, but I have a tendency to think upon a composite System Diagram.

As you should. But the "Gain" nevertheless must be in commensurate units if you are going to use it in calculations.

In my message to which you respond, I emphasized the many-to-one aspect of the transformation from the sensory input from the outer word to the signle-values perceptual signal. The transform the other way is even more problematic. The input function is at least computable in principle (though probably not in practice). The output signal (impulses per second from the control system's output function) has many different effects on the outer world -- you mentioned some. It's a one-to-many transformation, and is therefore _in principle_ underdetermined. The effects on the outer world depend on the states of all the control systems through which the output signal works its magic.

If you think of a composite system diagram, you certainly can draw a braided stream of influences from the ouput signal to the perceptual input function, but you can't presume a functional relationship. What you can do is look at the global influence on the input quantity of all those "Feedback Effects" through the environment. They counter the myriads of different disturbance influences.

>> and in the PCT model this output quantity get its correct physical
>>units by multiplying with ko which is a constant with units "signal
>>units per physical unit" (qo = ko*(r - p)). I understand that qo is
>>an expression that expresses my muscular exercise.

>No. That's true only for the most peripheral control systems.

Isn't it also true when we in a philosophical mood talk about a composite system?

I don't think so. I grant you that we sense only physical variables, but the variations in the perceptual variables don't seem to correspond in any meaningful way to physical units measured in the language of grams, metres, and seconds.

>>The face of my wife ( and her shoulders and her hat) constitute the
>>disturbance.

>They might provide the input to a perception. To what controlled
>variable might they constitute a disturbance?

My wish to see my wife's face.

What you see is your perception. Some (unknowable) external influence that alters what you see is a disturbance to your perception. The face of your wife and her shoulders and her hat are perceptions, which may or may not be what you want to see. A plausible disturbance might be whatever induced her to turn away. It isn't something you perceive (at least not within the control system controlling for seeing her face).

A few years ago, we had an important discussion on "points of view". When you talk about identifiable disturbances, you are taking "The Analyst's Viewpoint". The analyst can see all of the signal values all the way around the loop, as well as the values of the disturbance and reference signals. The "External Observer" could see the physical influences of the controller on the outer world, and the physical phenomena corresponding to the external input to the controller. But from inside the control system, the only way external signals are manifest is in the value of the perceptual signal and of the error signal. As Bill P. pointed out, what's "inside" is bounded on one end by the perceptual input function, and on the outher by the output function. (If you like, we could add a Reference Input Function, too, but that's irrelevant to this discussion).

It is really important, in talking about PCT, to keep straight which viewpoint you are using at any point in the discussion, and not mix viewpoints within the discussion. The analyst cal see the disturbance; the control system can't.

Martin

[Bj_rn_Simonsen]

[From Bjorn Simonsen (2006.11.17, 11:45 EUST)]

Martin Taylor – received 2006.11.16, 23:01 EUST.

Yes, OK, but I have a tendency to think upon a
composite System Diagram.

As you should. But the “Gain”
nevertheless must be

in commensurate units if you are going to use it in calculations.

Yes, commensurate units makes it difficult (to me) to
make complex simulations.

In my message to which you respond, I emphasized
the many-to-one

aspect of the transformation from the sensory
input from the outer

word to the signle-values perceptual signal.

Are you saying that if you go into a not illuminated
box, sit down on a chair and look at a illuminated picture of a banana, you
activate perceptual signals in some negative feedback loops. Will you activate
other valued perceptual signals if you leave the box and re-enter it 99 times?

What about the experiment when people looked at a grey
banana and said it was yellow?

The transform the other

way is even more problematic. The input function
is at least

computable in principle (though probably not in
practice). The output

signal (impulses per second from the control
system’s output

function) has many different effects on the outer
world – you

mentioned some. It’s a one-to-many transformation,
and is therefore

in principle underdetermined. The effects on the
outer world depend

on the states of all the control systems through
which the output

signal works its magic.

Do I misunderstand what you say if I think one certain
output variable (action) may have many different effects? And these one-to-many
transformations depend on environment factors not on the one output variables?

If you think of a composite system diagram, you

certainly can draw a braided stream of influences

from the output signal to the perceptual input

function, but you can’t presume a functional relationship.

Let us keep conflicts irrelevant. Can’t we presume
that the Feedback Effect (almost ever) counteract changes in the disturbance variable?

and in the PCT model this output quantity get its correct
physical

units by multiplying with ko which is a constant with units
"signal

units per physical unit" (qo = ko*(r - p)). I understand
that qo is

an expression that expresses my muscular exercise.

No. That’s true only for the most peripheral control systems.

Isn’t it also true when we in a philosophical
mood talk about a

composite system?

I don’t think so. I grant you that we sense only
physical variables,

but the variations in the perceptual variables
don’t seem to

correspond in any meaningful way to physical units
measured in the

language of grams, metres, and seconds.

Let us forget the banana above and let people look at
squares with different colours. Do you say that the perceptual variables don’t
seem to correspond to:

Red colour if the physical units that hit our retina
is between 625 and 740 nm,

Orange colour if the physical units that hit our
retina is between 590 and 625 nm

Yellow colour if the physical units that hit our
retina is between 565 and 590 nm

Green colour if the physical units that hit our retina
is between 500 and 565 nm

Turquoise colour if the physical units that hit our
retina is between 485 and 500 nm

Blue colour if the physical units that hit
our retina is between 440 and 485 nm

Violet colour if the physical units that hit
our retina is between 380 and 440 nm

It is really important, in talking about PCT, to
keep

straight which viewpoint you are using at any point

in the discussion, and not mix viewpoints within
the

discussion. The analyst cal see the disturbance;
the

control system can’t.

And often I take the analyst view and talk
about both the controlling person’s perceptions and the Extern Observer’s
perceptions.

I will return to your one-to-many
transformations and your many-to-one transformation in another thread.

bjorn

[Martin_Taylor2]

[Martin Taylor 2006.22.17.09.48]

[From Bjorn Simonsen (2006.11.17, 11:45 EUST)]
Martin Taylor - received 2006.11.16, 23:01 EUST.

>>Yes, OK, but I have a tendency to think upon a composite System Diagram.

>As you should. But the "Gain" nevertheless must be

in commensurate units if you are going to use it in calculations.

Yes, commensurate units makes it difficult (to me) to make complex simulations.

I think you mean "lack of" commensurate units.

>In my message to which you respond, I emphasized the many-to-one
>aspect of the transformation from the sensory input from the outer
>word to the signle-values perceptual signal.
Are you saying that if you go into a not illuminated box, sit down on a chair and look at a illuminated picture of a banana, you activate perceptual signals in some negative feedback loops.

You activate all sorts of perceptual signals. Whether any of them are in negative feedback loops depends on whether the person is controlling for particular values of any of them.

Will you activate other valued perceptual signals if you leave the box and re-enter it 99 times?

Possibly, possibly not. I would imagine you would "activate" a perceptual signal that we characterize as "feeling bored" the 100th time, but not the first. Also, in many situations, repeated exposure to similar physical input patterns results in a neutralized perception of the pattern. Colours tend toward grey, angles tend toward vertical or horizontal, and so forth.

What about the experiment when people looked at a grey banana and said it was yellow?

What about it? Context often influences the relation of a perceptual signal to the physical inputs.

Here is a personal anecdote that I think relates to the "banana" effect. When I was typing out my first draft of my thesis many years ago on a manual typewriter, I used yellow paper. I also had stacks of papers of other colours, blue, red, white and green. Apart from stopping to eat and sleep, typing the thesis was just about the only thing I did for four or five days. A day or two later, I wanted to type a letter. I took a sheet of paper off a stack and put it in the typewriter. When I started to type, I got annoyed at myself because I thought I had taken a page off the blue stack, when I wanted white. But when I took the sheet out of the typewriter, I saw that it "really" was white. I put it back into the typewriter and it turned blue again. This effect lasted for a day or two. Context influences colour, and can do so quite dramatically.

There's a follow-up to this story. I tried to demonstrate the effect in simpler situations, but didn't manage to get the conditions right. Several years later, Celeste McCullough did, and her name is now always attached to "the McCullough Effect" (look it up on Wikipedia; I imagine it's there).

>The output
>signal (impulses per second from the control system's output
>function) has many different effects on the outer world -- you
>mentioned some. It's a one-to-many transformation, and is therefore
>_in principle_ underdetermined. The effects on the outer world depend
>on the states of all the control systems through which the output
>signal works its magic.

Do I misunderstand what you say if I think one certain output variable (action) may have many different effects?

Absolutely. It's a cornerstone of PCT, isn't it? Many routes to the same result.

And these one-to-many transformations depend on environment factors not on the one output variables?

If you mean the environment of the single control loop, yes. If you mean the environment of the person, the answer there is "yes, too. But the real point is that for any given value of the output signal on different occasions, the lower-level loops that implement the influence of the output signal on its own perceptual signal are likely to be in different states. By "its own perceptual signal" I mean the one in the same control unit.

>If you think of a composite system diagram, you

certainly can draw a braided stream of influences
from the output signal to the perceptual input
function, but you can't presume a functional relationship.

Let us keep conflicts irrelevant. Can't we presume that the Feedback Effect (almost ever) counteract changes in the disturbance variable?

Here, it's hard to talk about "the disturbance variable" as though it is a unitary concept. The problem is that the influences that disturb the perception happen in the outer world, which means that what is usually drawn as a summation point where disturbance meets output is actually a whole lot of places within the braided stream. Some stream components may be undisturbed, some disturbance elements may not be compensated at all. All we know is that overall, the effect of the output more or less balances the opposite effect of the disturbances, and that none of this is identifiable (even by the analyst) inside the control element itself (the part between the perceptual input function and the output function by way of the comparator).

> I grant you that we sense only physical variables,
>but the variations in the perceptual variables don't seem to
>correspond in any meaningful way to physical units measured in the
>language of grams, metres, and seconds.

Let us forget the banana above and let people look at squares with different colours. Do you say that the perceptual variables don't seem to correspond to:
Red colour if the physical units that hit our retina is between 625 and 740 nm,
Orange colour if the physical units that hit our retina is between 590 and 625 nm
Yellow colour if the physical units that hit our retina is between 565 and 590 nm
Green colour if the physical units that hit our retina is between 500 and 565 nm
Turquoise colour if the physical units that hit our retina is between 485 and 500 nm
Blue colour if the physical units that hit our retina is between 440 and 485 nm
Violet colour if the physical units that hit our retina is between 380 and 440 nm

You are talking about the most peripheral (lowest level) perceptions in the conventional HPCT hierarchy, but even there these statements aren't always true, even in a person with normal colour vision. The colour that is perceived (consciously, at least) depends on the lighting, on the colours of surrounding parts of the visual field, and on what colours you have been looking at most recently -- as well as on effects like the "banana effect". Those variations can be pretty dramatic.

But that wasn't the point of my comment, which was that controlled perceptions of the kinds you were asking about (seeing your wife's face, student wanting to leave the room, for example), don't have any definable relationship to the physical variables. There is a relationship, of course, but it is one constructed by the various perceptual processes in the brain. If the relation of theperception of colour is so malleable when compared to the physical variables, how much more so would we expect higher-level percepts to be?

>It is really important, in talking about PCT, to keep

straight which viewpoint you are using at any point

>in the discussion, and not mix viewpoints within the
>discussion. The analyst cal see the disturbance; the
>control system can't.

And often I take the analyst view and talk about both the controlling person's perceptions and the Extern Observer's perceptions.

So do we all. The important thing is to be clear which viewpoint is used to make an analytic point. Early in this thread, you were talking "analyst" while treating what was accessble to the control unit, when you treated the disturbance as something available to it independent of the perception. The value of the disturbance is available to the analyst separately from the value of the perception, but that distinction is not available to the control unit.

In my very early days on CSGnet, Allan Randall and I had a long argument with Rick on this very matter, which I now see as a failure of communication that none of us identified at the time. Allan and I pointed out that informationally, the disturbance signal could largely be recovered from signal values within the control unit. Rick argued that the control unit has no "zero, nada" information about the disturbance. We arrived at an impasse, neither being convinced by the other. Now I understand why that was so: we were both right, but we were talking from different viewpoints.

Rick, correctly, was arguing that there is no way that the perceptual signal can discriminate the various components that generate its value (i.e., in the simple control loop diagram, the output and the disturbance). That's taking the viewpoint of the control system. Allan and I correctly argued that the time course of the disturbance can be recovered from the signal values within the control loop (specifically the perceptual signal and the output signal), and that this recovery was more accurate, the better the control. What none of us realized was that Rick was arguing "control system viewpoint", whereas Allan and I were arguing "analyst's viewpoint". Failure to recognize this distinction caused months of difficult discussion.

Viewpoint matters.

Martin

[Richard_Marken2]

[From Rick Marken (2006.11.17.0830)]

Bjorn Simonsen (2006.11.16,11:35 EUST)

Rick, Why do you object to mine example. It was an example.

I didn't object to the example. I just objected to your description of it.

After studying Martin�s comments I am inclined to say that the Feedback Effect box at higher levels is out of the model (?).

The feedback is the effect of system output on input. Therefore, it is always outside of the control model inasmuch as it is the environmental connection between output and input

>Now I'll try to say the same thing again in Norwegian....

>Kidding;-)

�Your last sentence is a disturbing for mine control �I wish to understand the meaning of your last sentence. The disturbance (the text) is not equal to my perception of the text. I don�t perceive myself Kidding, my perception is more like uncertainty if you are Kidding.

I meant _I_ was kidding about writing my reply in Norwegian. No way I could do that!

The disturbance contributes the perception, as does the output.

I think it's better to think of the disturbance as contributing to the state of the physical variables (qi) on which a perception depends. I think your way of thinking about it leads to the notion that perception _depends on_ a disturbance. But this is not the case. Think of it in terms of a tracking task. You would still perceive the cursor even if there were no disturbance pushing it around. Same with output. Output contributes to the variables on which a perception depends; but even if there is no output (if you don't move the handle in a tracking task) there is still a perception of the state of the cursor.

Best

Rick

···

--
Richard S. Marken Consulting
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

[Acree_Michael]

[Mike Acree (2006.11.17.10.30 PST)]

Martin Taylor 2006.22.17.09.48--

There's a follow-up to this story. I tried to demonstrate the effect in

simpler

situations, but didn't manage to get the conditions right.
Several years later, Celeste McCullough did, and her name is now always

attached to "the >McCullough Effect" (look it up on Wikipedia; I imagine
it's there).

Celeste McCullough was a professor of mine at Oberlin in the mid '60s,
one of my top favorites. Some years later another alumnus told me that
she had hit a child with her car, and disappeared after that. I've
never been able to find out what happened to her.

Mike

[Bj_rn_Simonsen]

[From Bjorn Simonsen (2006.11.17, 22:30 EUST)]

Martin Taylor 2006.22.17.09.48

Martin Taylor - received 2006.11.16, 23:01 EUST.

I think you mean “lack of” commensurate
units.

Yes, I think there are problems with what we have, but
I think there are more problems with what we ( I ) don’t have.

The output

signal (impulses per second from the
control system’s output

function) has many different effects on
the outer world – you

mentioned some. It’s a one-to-many
transformation, and is therefore

in principle underdetermined. The
effects on the outer world depend

on the states of all the control systems
through which the output

signal works its magic.

Do I misunderstand what you say if I think one
certain output

variable (action) may have many different
effects?

Absolutely. It’s a cornerstone of PCT, isn’t it? Many
routes to the

same result.

And these one-to-many transformations depend
on environment factors

not on the one output variables?

If you mean the environment of the single control
loop, yes. If you

mean the environment of the person, the answer
there is "yes, too.

But the real point is that for any given value of
the output signal

on different occasions, the lower-level loops that
implement the

influence of the output signal on its own
perceptual signal are

likely to be in different states. By “its own
perceptual signal” I

mean the one in the same control unit.

Martin, may I ask if you will read the >>>,

and the > above, once or twice.

Please comment my comments below.

The output

signal (impulses per second from the
control system’s output

function) has many different effects on
the outer world – you

mentioned some. It’s a one-to-many
transformation, and is therefore

in principle underdetermined. The
effects on the outer world depend

on the states of all the control systems
through which the output

signal works its magic.

Here you say there is a one-to-many transformation. That
is OK.

In your last sentence you say that the different (my
word) effects on the outer world depend on the states of all control systems
through which the output signal works its magic.

When we talk about the effects on the outer world and
a one-to-many transformation, I think the one output quantity has different
effects on the outer world because of the outer world, not because of all the
control systems through which the output quantity works its magic. Do me we
disagree?

Do I misunderstand what you say if I think one
certain output

variable (action) may have many different
effects?

Absolutely. It’s a cornerstone of PCT, isn’t it? Many
routes to the

same result.

Here I ask if I misunderstand you if one-to-many. And
you say: Yes, absolutely. And then you confirm that there is one-to-many.

I think Bill must step inn with language comments.

And these one-to-many transformations depend
on environment factors

not on the one output variables?

If you mean the environment of the single control
loop, yes. If you

mean the environment of the person, the answer
there is "yes, too.

But the real point is that for any given value of
the output signal

on different occasions, the lower-level loops that
implement the

influence of the output signal on its own
perceptual signal are

likely to be in different states. By “its own
perceptual signal” I

mean the one in the same control unit.

Here you say that these one-to-many transformations
depend on environment factors, the way you think upon environment factors.

But in your second sentence above I understand that
you say that it is the lower-level loops that implement the influence of the
output signal. And these lower level loops are not in the environment, are
they?

I
grant you that we sense only physical variables,

but
the variations in the perceptual variables don’t seem to

correspond
in any meaningful way to physical units measured in the

language of grams, metres, and seconds.

I still think that the physical units, (a number)nm,
e.g. 520 nm, representing the light from a green paper very, very often lead to
the perception of a green paper in a meaningful way.

I still think it is heavy to lift a 200 kg weight and
I think it is meaningful.

I still have problems when I try to run 100 meters on
5.0 seconds. This is meaningful.

bjorn

···

[Martin_Taylor2]

[From Bjorn Simonsen (2006.11.17, 22:30 EUST)]
Martin Taylor 2006.22.17.09.48

Martin Taylor - received 2006.11.16, 23:01 EUST.

>> >The output
>>>signal (impulses per second from the control system's output
>>>function) has many different effects on the outer world -- you
>>>mentioned some. It's a one-to-many transformation, and is therefore
>>>_in principle_ underdetermined. The effects on the outer world depend
>>>on the states of all the control systems through which the output
>>>signal works its magic.

>>Do I misunderstand what you say if I think one certain output
>>variable (action) may have many different effects?

>Absolutely. It's a cornerstone of PCT, isn't it? Many routes to the
>same result.

I see an ambiguity in my response. I meant that you do not misunderstand. "Absolutely" was intended to refer to the part after "I think".

>> And these one-to-many transformations depend on environment factors
>>not on the one output variables?

>If you mean the environment of the single control loop, yes. If you
>mean the environment of the person, the answer there is "yes, too.
>But the real point is that for any given value of the output signal
>on different occasions, the lower-level loops that implement the
>influence of the output signal on its own perceptual signal are
>likely to be in different states. By "its own perceptual signal" I
>mean the one in the same control unit.

Martin, may I ask if you will read the >>>, >> and the > above, once or twice.
Please comment my comments below.
>> >The output
>>>signal (impulses per second from the control system's output
>>>function) has many different effects on the outer world -- you
>>>mentioned some. It's a one-to-many transformation, and is therefore
>>>_in principle_ underdetermined. The effects on the outer world depend
>>>on the states of all the control systems through which the output
>>>signal works its magic.

Here you say there is a one-to-many transformation. That is OK.
In your last sentence you say that the different (my word) effects on the outer world depend on the states of all control systems through which the output signal works its magic.
When we talk about the effects on the outer world and a one-to-many transformation, I think the one output quantity has different effects on the outer world because of the outer world, not because of all the control systems through which the output quantity works its magic. Do me we disagree?

We don't disagree, if you change "not because" into "and because". If you leave it as "not because", then we disagree.

>>Do I misunderstand what you say if I think one certain output
>>variable (action) may have many different effects?

>Absolutely. It's a cornerstone of PCT, isn't it? Many routes to the
>same result.

Here I ask if I misunderstand you if one-to-many. And you say: Yes, absolutely. And then you confirm that there is one-to-many.
I think Bill must step inn with language comments.

See above. As Bill said, the unintended side of an ambiguity is, by Murphy's Law, the interpretation chosen!

>> And these one-to-many transformations depend on environment factors
>>not on the one output variables?

>If you mean the environment of the single control loop, yes. If you
>mean the environment of the person, the answer there is "yes, too.
>But the real point is that for any given value of the output signal
>on different occasions, the lower-level loops that implement the
>influence of the output signal on its own perceptual signal are
>likely to be in different states. By "its own perceptual signal" I
>mean the one in the same control unit.

Here you say that these one-to-many transformations depend on environment factors, the way you think upon environment factors.

But in your second sentence above I understand that you say that it is the lower-level loops that implement the influence of the output signal. And these lower level loops are not in the environment, are they?

They most definitely ARE in the environment of the control system we are considering. They are the recipiens of its output signal, and they contribute inputs to its perceptual input function. The are part of its feedback path -- the "braided stream" we have been talking about. And their reference signals are derived from other higher-level loop output signals besides the single one we are dealing with. So not only are they subject to variable disturbances from the world outside the person, but also they have varying levels of their reference signals, reference signals to which "this" control loop contributes. Even if they did't, their error values could well be different because of the history of their efforts to control their own perceptual variables.

The consequence is that the actions that control a particular high-level perception may change, even if the environment external to the person doesn't (a possibility that is extremely unlikely).

> > I grant you that we sense only physical variables,
> >but the variations in the perceptual variables don't seem to
> >correspond in any meaningful way to physical units measured in the
> >language of grams, metres, and seconds.

I still think that the physical units, (a number)nm, e.g. 520 nm, representing the light from a green paper very, very often lead to the perception of a green paper in a meaningful way.

Oh, sure. It's usual. Control would be hard to establish if it weren't. Most of what we perceive probably has "usual" relationships of some kind to something in the "real reality". Usually when we let something go, it falls to the floor; but some (unusual) things float to the ceiling. And provided the illuminating light doesn't vary too much from a moderately flat spectrum, other contextual changes usually won't take the perceived colour of a green paper into red or blue. But isn't that a very peripheral point to the discussion, as well as being dealing with a very peripheral perception, which, in the hierarchy, is not mediated by lower-level control loops?

I still think it is heavy to lift a 200 kg weight and I think it is meaningful.
I still have problems when I try to run 100 meters on 5.0 seconds. This is meaningful.

All you are saying here is that there are limits on our muscular output. Sure, and there are limits on the "impulses per second" output of any neuron. But it's a mystery to me how this or the colour comments relate to the question which was how to describe in pure PCT "trying to see my wife's face" or "wanting to leave the room with the teacher's permission".

Martin

[Bill_Powers1]

[From Bill Powers (2006.11.16.1545 MST)]

I think something came out backward in this:

Martin Taylor 2006.22.17.09.48--

Bjorn:

>>Do I misunderstand what you say if I think one certain output
>>variable (action) may have many different effects?

Martin:

>Absolutely. It's a cornerstone of PCT, isn't it? Many routes to the
>same result.

I see an ambiguity in my response. I meant that you do not misunderstand. "Absolutely" was intended to refer to the part after "I think".

Meaning A:
One output can have many effects: a controlled effect, and a number of side-effects relative to the control system in question. That is a one-to-many relation.

Meaning B:
Many different outputs can be used to control the same controlled variable, so if something interferes with one method of control, there is usually some other action that can be learned which will restore control by a different path.

Meaning C:
A control system can involve many lower-order outputs, which affect the individual variables which are combined to produce one higher-order controlled perception. The output fans out to affect multiple aspects of the environment (one-to-many) and then the perceptual signal combines many aspect of the environment into a smaller number of controlled varibles (many-to-one)\.

All these statements are true, but it's good to specify which meaning is intended.

Best,

Bill P.

[Martin_Taylor2]

[Martin Taylor 2006.11.17.19.42]

Isn't language wonderful!

[From Bill Powers (2006.11.16.1545 MST)]

I think something came out backward in this:

Martin Taylor 2006.22.17.09.48--

Bjorn:

>>Do I misunderstand what you say if I think one certain output
>>variable (action) may have many different effects?

Martin:

>Absolutely. It's a cornerstone of PCT, isn't it? Many routes to the
>same result.

I see an ambiguity in my response. I meant that you do not misunderstand. "Absolutely" was intended to refer to the part after "I think".

Meaning A:
One output can have many effects: a controlled effect, and a number of side-effects relative to the control system in question. That is a one-to-many relation.

Meaning B:
Many different outputs can be used to control the same controlled variable, so if something interferes with one method of control, there is usually some other action that can be learned which will restore control by a different path.

Meaning C:
A control system can involve many lower-order outputs, which affect the individual variables which are combined to produce one higher-order controlled perception. The output fans out to affect multiple aspects of the environment (one-to-many) and then the perceptual signal combines many aspect of the environment into a smaller number of controlled varibles (many-to-one)\.

All these statements are true, but it's good to specify which meaning is intended.

I didn't even see the possibility of A, and I don't really discriminate between B and C, except as a matter of focus. I saw an ambiguity in whether I "Absolutely" thought Bjorn misunderstood or whether he didn't. You see two extra possibilities in his non-misunderstanding, which leaves open the question of whether he yet misunderstood!

Ye gads! Will the mysteries of language never cease?

Martin

[Martin_Taylor2]

Re: Elements in the PCT model
Martin Taylor 2006.11.21.15.37]

[From Bill Powers (2006.11.21.0715
MST)]
Martin Taylor (2006.11.21)

You’re arguing about whether the
physics model and the PCT model are being correctly applied. All
arguments of this kind about “is” and “is not” are
really arguments about who is following the rules of the model more
correctly. No resolution can be reached through purely verbal
arguments.

and you reply:

I’m afraid I have to dispute that.
Mathematics is just another kind of verbal argument, after all.

Yes, of course, and numbers are just another kind of written
symbol.

You say that with a touch of sarcasm, or so it seems. But it is
more true than you give credit for.

However, in verbal arguments terms can
change their definitions from one use to the next, and premises can be
constructed on the fly, as needed,

The essence of philosophical argument

and all sorts of other logical
perversions are permitted. A mathematical argument requires agreement
on definitions to start it,

Aye, there’s the rub!

Mathematics proceeds from axioms, and uses classical logic
(ordinarily), in which a statement is true or false (whether provably
so or not). In teh axioms, definitions are not required. Elements just
are themselves. More complex stuff is defined by means of processes
applied to the axioms. Nowhere do the definitions link tothe
observable world (other than the observable world of symbols).

It so happens that an awful lot of mathematics works very well
when applied to the (perceived) real world. One is tempted to think
that the mathematics “really” describes the world, and maybe
it does, in the way a skeleton describes the body. The mathematics we
use is based on axioms in which the elements do have implicit
metaphors in the real world. Numbers reflect the cattle the farmer
sells; the miracle is that it still works with negative and imaginary
numbers and quaternions.

We seldom apply mathematics in a way that goes back to the
axioms, nor do we use formal logic of anyone’s preferred description.
We start with much more complex entities and try to define them in
terms of perceptions. Take “energy” for example, since it is

I sketched in a case (the ball falling at
midnight) in which one could demonstrate that information is
transmitted by means of absorbing energy into the transmitter rather
than sending it from transmitter to receiver.

“Information is transmitted by absorbing energy into the
transmitter”. Setting asside the question of defning
“energy”, which otherwise we have to assume is defined
identically by the communicating partners, it is not at all clear what
energy is intended here. Is it the photon energy imparted by
reflection off the ball, which is actuall zero if the photon doesn’t
change wavelength? Is it the graviational ptential energy lost by the
ball as it falls? is it the kinetic energy the ball gains as it
accelerates? is it the heat energy of friction as the ball slides down
the pole? Or, the only plausible one, since it coincides with the
moment specified by the event, is it the heat energy generated by the
conversion of kinetic energy when the ball stops at the bottom?

I don’t really think you mean the only plausible one, but if you
are going to generate a mathematical description, you have to make at
least that definition precise. The problem with using the transfer
of kinetic enery into heat at the bottom of the frop is that the human
perceiver senses neither. The only sensible connection between the
ball and the observer is the photon stream, and what energy transfer
there is comes from the bsorption of the photon and its conversion
into chemical energy, That photon did come from the ball to the eye,
even though the ball changed only its momentum vector, not its energy.
So far as I can see, there isn’t any obvious transfer of energy
between the ball and the perceiver, in either direction.

At this point, I would have no basis for a mathematical
description of the relation between energy transfer and information
transfer. But I presume you do have a mental sketch of how such a
description would go, and what definitions would be required.

and adherence to those definitions
throughout. It requires that the means of reasoning be known to all,
and laid out in advance, and adhered to.

Violation of ths is the way that mathematics has advanced over
the centuries, but let that pass.

For example, the many-to-one versus
one-to-many issue: You and Bjorn are talking about different parts of
the models in different contexts,

I hadn’t detected that. I’ll have to go back and see where your
impression comes from. I’ve tried to address specifically what Bjorn
has been saying. If we’ve been talking past each other (as, evidently,
Rick and I were doing 14 years ago), it would be nice to know
where.

If you speak of the effects of
first-order output quantities in the world-model, it is clear that one
output quantity has many effects that fan out into the physical world,
only some of which are involved in the local control system. If you
speak of alternative paths to the same result in the environment, a
many-to-one effect, only some of them will be in effect at a given
time and the others will come into play only if the first one fails,
and then only after some reorganization to hook up new output effects
to the old error signal. If, on the other hand, you’re speaking of
output signals at higher levels, one set of connections to lower
systems has effects on only the reference inputs to lower orders that
are actually connected. And finally, the many-to-one connections in
the upgoing paths are subject to similar comments, with many inputs,
some controlled and some not controlled, contributing. There seems
little room for ambiguity here, so it’s hard to see what prevents
agreement except imprecise language and shifting
meanings.

All of which I’ve been trying to get across, and I think Bjorn
gets it. It seems to be the implications that don’t get across.

Moreover, in a situation in which
the premises and relationships cover a wide range of possibilities, it
is often vry difficult to create an intelligible mathematical
description from which generic conclusions can be derived. I’d be
interested, for example, in how you would presnet in an intelligble
form, a mathematical demonstration of this:

As I pointed out some years ago, at least
10 and perhaps 15, the energy budget of acting and sensing has nothing
to do with the transmission of information. Information can easily
move in a direction opposite to the direction of energy transfer.
Information has nothing to do with energy (or entropy).

I assume that “this” refers to “Information can easily
move in a direction opposite to the direction of energy transfer”
rather than “I pointed out…” or “10 and perhaps 15
years ago” or “information has nothing to do with energy or
entropy.” The first statement contains the kernel of my argument
and can be demonstrated.

Questioned, above. The one I was most interested in was the last,
since except at the low-energy limit, which doesn’t concern most of
our perceptions, information does seem to have little or nothing to do
with energy (how to mathematize that statement is another matter). But
if physicists, the most mathematically inclined of scientists, do
think that information and entropy are deeply interlinked, I think it
would be even harder to find an appropriate mathematical description
of “Information has nothing to do with entropy”.

I’m saying only that the direction
of energy transfer is not, in general, important. It may even be
irrelevant, Shannon notwithstanding.

I don’t remember Shannon saying anything on that topic. Could you
point me to where he did? Shannon is someone I admire greatly, but I
really only know the BTJ articles that became the book.

There is simply no point in trying to
settle any matter of this kind without setting up a formal argument,
agreeing on what the data are, and following agreed-upon rules to
derive conclusions. Then the only questions are whether you have
considered all relevant data, measured accurately, and reasoned
without blunders. If you have handled the preliminaries properly,
there is no room left for disagreement.

That really is the Bertrand Russell view, isn’t it? How many hundred
pages did it take him to demonstrate that you could say 1+1=2,
starting with the NAND operator (or something like that). And then
Goedel went and blew his program to pieces!

I said, “a matter of this kind,”
meaning a matter involving the PCT model and the physics model. All
you have to do is say what part of which model at what level you are
talking about, and I believe all the disagreements will disappear. I
can’t see what that has to do with Principia or
Godel.

It has to do with using axiom-based definitions, or verbally
precise definitions grounded in a fuzzy world, and expecting to be
able to prove everything in the real world by applying formal
reasoning to them. It often works, and works amazingly well. But it’s
no basis for a discussion or even an argument, unless between partners
who are equally skilled with the processing tools and who understand
the same placement of the precise definitions within the fuzziness of
the real world.

Ordinary language is more ambiguous, but it’s often easier to
understand, and to understand correctly, than is the precision of
mathematics. It’s the usual tradeoff between brittle precision and
robust but imprecise strength.

Martin

···

at issue. You said:

[Bill_Powers1]

[From Bill Powers (2006.11.21.1646 MST)]

Martin Taylor 2006.11.21.15.37 –

and all sorts of other
logical perversions are permitted. A mathematical argument requires
agreement on definitions to start it,

Aye, there’s the rub!

Mathematics proceeds from axioms, and uses classical logic (ordinarily),
in which a statement is true or false (whether provably so or not). In
teh axioms, definitions are not required. Elements just are themselves.
More complex stuff is defined by means of processes applied to the
axioms. Nowhere do the definitions link tothe observable world (other
than the observable world of symbols).

You’re taking “mathematics” too literally. I think it’s
possible to offer a verbal argument using constraints like those we apply
in formal mathematics. I have tried to do that, and of course have met
with resistance because people can see that if we do it this way,
agreement is easy – but the conclusions they’re trying to reach are no
longer reachable. Remember the fuss about the word “control”?
This word has a number of mutually contradictory usages, but there is
great reluctance to settle on just one. As a consequence, pointless
disagreements continue and we continue trying to construct Towers of
Babel with predictable results.

We seldom apply mathematics in a
way that goes back to the axioms, nor do we use formal logic of anyone’s
preferred description. We start with much more complex entities and try
to define them in terms of perceptions. Take “energy” for
example, since it is at issue. You said:

I sketched in a case (the ball
falling at midnight) in which one could demonstrate that information is
transmitted by means of absorbing energy into the transmitter rather than
sending it from transmitter to receiver.

“Information is transmitted by absorbing energy into the
transmitter”. Setting asside the question of defning
“energy”, which otherwise we have to assume is defined
identically by the communicating partners, it is not at all clear what
energy is intended here.

I said which kind: the potential energy of the ball. If you want more
details, I’m imagining that the person transmitting the message (passage
of time before midnight) is varying the finger pressure of a hand to let
the rope supporting the ball slide through it at a controlled rate. All
the energy flow is from the rope into the hand, heating the skin at the
output of the transmitter. No energy flows from the transmitter into the
ball; it goes the other way.

The only sensible connection
between the ball and the observer is the photon stream, and what energy
transfer there is comes from the bsorption of the photon and its
conversion into chemical energy, That photon did come from the ball to
the eye, even though the ball changed only its momentum vector, not its
energy. So far as I can see, there isn’t any obvious transfer of energy
between the ball and the perceiver, in either
direction.

All that is irrelevant because at the output of the transmitter, there is
no outflow of energy into the ball. Whatever happens from then on doesn’t
matter, because the transmitter is not sending energy outward.

At this point, I would have no
basis for a mathematical description of the relation between energy
transfer and information transfer. But I presume you do have a mental
sketch of how such a description would go, and what definitions would be
required.

The entropy equations, as I understand them, describe the fractional
transfer of energy at a given energy level: for heat, dh = dQ/T. The
expression dQ is a transfer of energy. This form of equation has also
been used to describe information flow, and much has been made of the
relation between information flow and entropy changes, because the
mathematical forms look similar. But if information can be transmitted by
either a positive or a negative change in entropy. the apparently
similarity is only an accident and is not physically significant. In a
similar way, a mass on a spring and a negative feedback control system of
a particular type can be described by identical second-order differential
equations, yet there is no connection between their mechanisms of
operation.

and adherence to those
definitions throughout. It requires that the means of reasoning be known
to all, and laid out in advance, and adhered to.

Violation of ths is the way that mathematics has advanced over the
centuries, but let that pass.

Why? Do you recomment that we change the meaning of “+” in the
middle of transforming an equation? That we redefine the physical
variable represented by X in the middle of solving for X?

direction of energy transfer.
Information has nothing to do with energy (or
entropy).

I assume that “this” refers to “Information can easily
move in a direction opposite to the direction of energy transfer”
rather than “I pointed out…” or “10 and perhaps 15 years
ago” or “information has nothing to do with energy or
entropy.” The first statement contains the kernel of my argument and
can be demonstrated.

Questioned, above. The one I was most interested in was the last, since
except at the low-energy limit, which doesn’t concern most of our
perceptions, information does seem to have little or nothing to do with
energy (how to mathematize that statement is another matter). But if
physicists, the most mathematically inclined of scientists, do think that
information and entropy are deeply interlinked, I think it would be even
harder to find an appropriate mathematical description of
“Information has nothing to do with entropy”.

I sketched in one way above.

I said, “a matter of this
kind,” meaning a matter involving the PCT model and the physics
model. All you have to do is say what part of which model at what level
you are talking about, and I believe all the disagreements will
disappear. I can’t see what that has to do with Principia or
Godel.

It has to do with using axiom-based definitions, or verbally precise
definitions grounded in a fuzzy world, and expecting to be able to prove
everything in the real world by applying formal reasoning to
them.

No, not everything. Just the things we have spend the last 20 years or so
defining and discussing. Has all that effort been wasted?

Ordinary language is more
ambiguous, but it’s often easier to understand, and to understand
correctly, than is the precision of mathematics. It’s the usual tradeoff
between brittle precision and robust but imprecise
strength.

I think that’s a wish rather than a fact. When there is only a fuzzy way
of communicating, determine what “correct” means is also a
fuzzy process, and the fuzzier it gets the easier it is to perceive
agreement when there is actually disagreement. It’s easier to understand
ordinary language because less is demanded by way of demonstrating
correct understanding. I agree that mathematics can be pretty opaque, but
that’s largely because of the need to remember lots of conventions and
shorthand notations. It’s not necessary to be so terse, however,
especially when using ordinary language under constraints like those we
use in mathematics. Like keeping definitions the same for the duration of
an argument

Best,

Bill P.

[Bill_Powers1]

[
From Bill Powers (2006.11.24.1045 MST)]

Martin Taylor 2006.11.23.20.10]

Bjorn Simonsen (20066.11.23, 11:10 EUST) --

You guys are still talking about different things. Bjorn says that light is a disturbance. You say it is not. So you are talking completely past each other.

Bjorn:

The loop can be presented as a composed loop, as a loop expressing what happens when we wish to perceive something or what happens when we are imputed a disturbance e.g. in form of a light.

Martin:

I made a point that the loop I was considering was NOT subject to ANY disturbance. The error in the starting condition came from the fact that the screen was showing a colout (say red, for example) different from the reference colour (blue, you suggest).

The more you talk past each other, the longer the posts get. If you settle on what is meant by the word "disturbance," and explicitly agree on the definition and stick to it, a great many of those words will become unnecessary.

Best,

Bill P.

[Fred_Nickols1]

[From Fred Nickols (2006.11.24.1431 EST)] --

From Bill Powers (2006.11.24.1045 MST)]

Martin Taylor 2006.11.23.20.10]

Bjorn Simonsen (20066.11.23, 11:10 EUST) --

You guys are still talking about different things. Bjorn says that
light is a disturbance. You say it is not. So you are talking
completely past each other.

Bill's comment prompts a question. Setting aside the particulars of the discussion in question, I would like to deal with the situation in which one person says light is a disturbance and another says it isn't.

It seems to me that that statement is somewhat akin to one person saying that food is a reinforcer and another saying it isn't. Don't both statements depend on context?

If I'm trying to look at something and a bright light is causing me to squint or shade my eyes, it seems to me that this bright light is a disturbance to my goal of vewing something. On the other hand, if the light doesn't interfere with my ability to do that, then it's not a disturbance.

Do I have the correct?

Regards,

Fred Nickols
nickols@att.net