More on GAP-ACT

[Fred_Nickols1]

[From Fred Nickols (2006.11.06.1036 ET)] --

Thanks to those who read and commented on the GAP-ACT piece I sent out
earlier.

I've written a second, brief piece to accompany that first one. It is
attached and it is more or less an answer to the "So what?" question that
might be raised about the first piece. In this second piece, the issue is
one of how management might use a grasp of the GAP-ACT model to influence
human performance.

Any comments, suggestions, ideas or criticisms are welcome.

Regards,

Fred Nickols
"Assistance at a Distance"
nickols@att.net
www.nickols.us

Influencing Human Performance.doc (157 KB)

[Richard_H_Pfau]

Fred,

As a management trainer interested in PCT, I’d like to commend you on the layout of the GAP-ACT model as well as the clear explanations and examples given in the two papers recently sent. The model’s name is also seemingly useful in helping others grasp the concepts involved.

A very nice job and a nice way to introduce ideas from PCT into work settings.

The approach seems quite compatible with and complementary to Locke and Latham’s findings and preachings concerning relationships between clear specific goals and improved performance in work situations. Providing improved feedback and influencing confounding and complicating conditions/obstacles are also often recognized as important aspects of performance improvement (e.g., Mager and Pipe, “Analyzing Performance Problems”) – as I’m sure you well know.

In short, the GAP-ACT model seems like a natural for introduction into the mainstream of management and supervisory training, given the way it both uses many concepts that are now a standard part of such training and links those concepts into an understandable structure. I hope you’ll popularize the model in some management publications so that it hopefully does become a part of workplace thought processes.

With Regards,

Richard Pfau

HRD Consultant

[Bj_rn_Simonsen]

[From Bjorn Simonsen
(2006.11.07,12:00 EUST)]

From
Fred Nickols (2006.11.06.1036 ET)

In this view, the performer is
attempting to control some

targeted variable (T) so as to keep it aligned with a goal state (G).

I appreciate your GAP-ACT model,
Fred.

I worked with a comment after your [From Fred Nickols (11.03.2000.1124 ET)], but I put it
aside.

The job is to convince people working
with organization development to see that people control their perceptions.

This is general knowledge among us
in CSG. Nevertheless we often express that people control their actions (you
are the last one (look at your two lines above)). Implicitly we emphasize that
people control their perceptions, but for one or another reason we express
ourselves as if they control their actions. Maybe Rick explained why in his:

From Rick Marken (2000.09.26.1850)]\

So I think there is a sense in
which the success of the control

of perception model can be
considered evidence that what we

actually control, when we control
our perceptions, is the real

world. Once, based on PCT, you
accept the fact that the apparent

real world you control is
actually control of perception, you can

move on, using PCT, to accept the
assumption that the perceptions

you control are the real
world…Grasshopper.

I think that such argumentation guides us in
a wrong direction. I appreciate the way Bill express himself.

[From Bill Powers (2004.06.28.0133 MDT)]

Remember that what we say about experience
does not

determine the experience, but only what we
think about the experience, how

we explain the experience, what we do about
the experience, and so on.

I think Rick and Bill agree about Bill’s presentation,
but it is difficult for Rick to stop saying that the real world is just what
two or a million people say they experience when a stone is falling against the
earth. The world isn’t so, but a million people agree that they perceive the
real world as they do.

I think PCT gives the world an important message. And that
is: “Thrust in what you perceive. We have not enough knowledge to say that the
connection between neurons and states of mind (perceptions) are a true
representation of the disturbances from the real world. What we perceive is the
best be have. It is good enough and we need our perceptions, but we shall
always remember that they are our perceptions. “

Those who live up to this message will always respect
that other people may perceive things different. Sometimes they will change
their own goals to perceive different| and sometimes they will live their lives
disagreeing with themselves or with other people. They will never try to force
other people to change their actions, neither will they fool them to do so. Sometimes
they will try to find an agreement at a higher level that makes a disappearing
act of the earlier goals they disagreed.

I think you describe this very well in your paragraph “Influencing Goals”.

If I should have written your piece I would not have used
the headings “Influencing

Perceptions”, ” Influencing Actions” and “Influencing
Confounding and Complicating

Conditions”. I
would have stopped with the heading “
Influencing Goals” and
explained

that if people change their goals, they also change
their actions, perceptions and Confounding and complicating conditions. But you
know this better than me. You explained that very well in your Rubber Band Demonstration.
And you are the specialist on Organization Development.

Again. I think you have done a good job. Pardon me for
having some different perceptions.

bjorn

···

[Fred_Nickols1]

[From Fred Nickols (2006.11.07.0628 EST)] --

from Richard Pfau (no date time stamp)

[mailto:CSGNET@LISTSERV.UIUC.EDU] On Behalf Of Richard H. Pfau
Sent: Monday, November 06, 2006 8:02 PM

Fred,

As a management trainer interested in PCT, I'd like to commend you on the
layout of the GAP-ACT model as well as the clear explanations and examples
given in the two papers recently sent. The model's name is also seemingly
useful in helping others grasp the concepts involved.

A very nice job and a nice way to introduce ideas from PCT into work
settings.

Thanks for the kind words.

The approach seems quite compatible with and complementary to Locke and
Latham's findings and preachings concerning relationships between clear
specific goals and improved performance in work situations. Providing
improved feedback and influencing confounding and complicating
conditions/obstacles are also often recognized as important aspects of
performance improvement (e.g., Mager and Pipe, "Analyzing Performance
Problems") -- as I'm sure you well know.

Yes, I'm quite familiar with the Mager & Pipe book, going all the way back
to the first edition in 1970. As it happens, Mager & Pipe drew their
inspiration (if not much of their actual material) from a paper they cited
in that first edition. It's titled "Maintenance Systems: The Neglected Half
of Behavior Change" and it was written by Karen Brethower. Her paper was
published in a book titled "Managing the Instructional Programming Effort"
(1967) which was edited by Geary Rummler. In Karen's paper, is a deficiency
analysis guide devised by Rummler, which seems to be the basis for the Mager
& Pipe book, including the gun to the head question. (Mager & Pipe used the
gun to the head question and Geary asked simply if the performer could do
the task if his life depended on it). FWIW, you can find a copy of Karen's
paper on my web site. (I recently received permission from her and from
UMich to reprint it.) It's at the following link:

http://home.att.net/~nickols/Brethower.pdf

As I'm sure you know, human performance technology enjoys the fruits of many
empirical findings but it's short on theory. PCT seems to me to offer a
much more solid theoretical basis for all those empirical findings.

In short, the GAP-ACT model seems like a natural for introduction into the
mainstream of management and supervisory training, given the way it both
uses many concepts that are now a standard part of such training and links
those concepts into an understandable structure. I hope you'll popularize
the model in some management publications so that it hopefully does become
a part of workplace thought processes.

That is part of my aim. The underlying aim is to focus attention on PCT.

If you're interested, the GAP-ACT model is part of a paper I'm working on
titled "Roadmaps to Results" which deals with three domains of performance:
Financial, Operational and Human. That paper is about how models (diagrams
and schematics) of the situations in which we intervene guide our
interventions. The GAP-ACT model is used in the human performance domain.
You can find the Roadmaps paper at the following link:

http://home.att.net/~essays/roadmaps.pdf

It contains what I think is a neat example of the GAP-ACT model used to
analyze factors related to the way applicants fill out registration forms.
It also points to Bill's books.

Thanks again for the kind words and comments.

Regards,

Fred Nickols
nickols@att.net

[Fred_Nickols1]

[From Fred Nickols (2006.11.07.0649 EST)] --

From Bjorn Simonsen (2006.11.07,12:00 EUST)]
From Fred Nickols (2006.11.06.1036 ET)
>In this view, the performer is attempting to control some
>targeted variable (T) so as to keep it aligned with a goal state (G).

I appreciate your GAP-ACT model, Fred.
I worked with a comment after your [From Fred Nickols (11.03.2000.1124
ET)], but I put it aside.
The job is to convince people working with organization development to see
that people control their perceptions.

I hope they come to see that but that's not the job I've chosen.

This is general knowledge among us in CSG. Nevertheless we often express
that people control their actions (you are the last one (look at your two
lines above)).

I don't see how what I wrote in those lines can be construed as people
trying to control their actions. I said quite explicitly that they are
attempting to control some targeted variable, what would ordinarily be
called the controlled variable on this list and they are trying to keep it
aligned with a goal state, which would be called the reference condition on
this list. How does that translate to controlling actions?

Implicitly we emphasize that people control their
perceptions, but for one or another reason we express ourselves as if they
control their actions. Maybe Rick explained why in his:

>>>From Rick Marken (2000.09.26.1850)]\
>>>So I think there is a sense in which the success of the control
>>>of perception model can be considered evidence that what we
>>>actually control, when we control our perceptions, is the real
>>>world. Once, based on PCT, you accept the fact that the apparent
>>>real world you control is actually control of perception, you can
>>>move on, using PCT, to accept the assumption that the perceptions
>>>you control are the real world...Grasshopper.

I think that such argumentation guides us in a wrong direction. I
appreciate the way Bill express himself.

>>[From Bill Powers (2004.06.28.0133 MDT)]

>>Remember that what we say about experience does not
>>determine the experience, but only what we think about the experience,
how
>>we explain the experience, what we do about the experience, and so on.

I think Rick and Bill agree about Bill's presentation, but it is difficult
for Rick to stop saying that the real world is just what two or a million
people say they experience when a stone is falling against the earth. The
world isn't so, but a million people agree that they perceive the real
world as they do.

And that's your perception of reality, right?

I think PCT gives the world an important message. And that is: "Thrust in
what you perceive. We have not enough knowledge to say that the connection
between neurons and states of mind (perceptions) are a true representation
of the disturbances from the real world. What we perceive is the best be
have. It is good enough and we need our perceptions, but we shall always
remember that they are our perceptions. "

I know many people who already believe this. They also believe that there
is enough congruence between their individual perceptions and what other
sources say is reality that they can pretty much get away with acting as
though their perceptions are an accurate reflection of reality. And so they
proceed as though perception equals reality - which, if course, it doesn't
but it's pretty close most of the time.

Those who live up to this message will always respect that other people
may perceive things different. Sometimes they will change their own goals
to perceive different| and sometimes they will live their lives
disagreeing with themselves or with other people. They will never try to
force other people to change their actions, neither will they fool them to
do so. Sometimes they will try to find an agreement at a higher level that
makes a disappearing act of the earlier goals they disagreed.

I agree.

I think you describe this very well in your paragraph "Influencing Goals".
If I should have written your piece I would not have used the headings
"Influencing
Perceptions", " Influencing Actions" and "Influencing Confounding and
Complicating
Conditions". I would have stopped with the heading " Influencing Goals"
and explained
that if people change their goals, they also change their actions,
perceptions and Confounding and complicating conditions. But you know this
better than me. You explained that very well in your Rubber Band
Demonstration. And you are the specialist on Organization Development.

Well, lurking in there is a might big issue; namely, the very profound
differences between control in a setting where the controlled variable (T)
is something that is affected directly and immediately by the person's
actions (like the rubber band demo) and a setting where the controlled
variable is some far removed in space and time from the direct and immediate
effects of one's actions (e.g., the error rate in a process). In the first
case, our perceptions do a reliable job of informing us rather immediately
of any changes in the controlled variable. In the second case, delay
creates some interesting effects.

Again. I think you have done a good job. Pardon me for having some
different perceptions.

That's fine, Bjorn. I appreciate you taking the time to read and respond.

I have the gnawing feeling that I'm botching the job in that you feel I'm
not accurately presenting PCT via the GAP-ACT model (or explanations of it).
If that's the case, we need to continue this discussion because getting it
right is indeed one of my goals. So, if you have alternate wording, please
feel free to suggest it.

Regards,

Fred Nickols
www.nickols.us
nickols@att.net

[Bill_Powers1]

[From Bill Powers (2006.11.07.0730 MST)]
Fred Nickols (2006.11.07.0649 EST)] --

Bjorn:

>It is good enough and we need our perceptions, but we shall always
> remember that they are our perceptions.

Fred:

I know many people who already believe this. They also believe that there
is enough congruence between their individual perceptions and what other
sources say is reality that they can pretty much get away with acting as
though their perceptions are an accurate reflection of reality. And so they
proceed as though perception equals reality - which, if course, it doesn't
but it's pretty close most of the time.

Can you give me an example of "pretty close," Fred? More specifically, I'm interested in what you compare the perception to, to determine that the perception is pretty close to it.

Best,

Bill P.

[Bill_Powers1]

[From Bill Powers (2006.11.07.0750 MST)]

Richard Pfau (2006.11.07) –

The approach seems quite compatible
with and complementary to Locke and Latham’s findings and preachings
concerning relationships between clear specific goals and improved
performance in work situations.

You might try writing to Locke and Latham about this. Both my wife Mary
and I tried for at least four years to communicate with Locke
(as well as Bandura), but he was violently against my work and that of
anyone else who used it. When Mary asked Locke f he had read my book,
Behavior: the control of perception, he said he didn’t need to, since the
title itself was “incorrect.”

I thought I’d warn you that if you have some idea of bringing the two
views together you might run into some unexpected opposition.

Best,

Bill P.

[bnhpct]

[From Bruce Nevin (2006.11.07.11.21 EST)] --

Fred Nickols (2006.11.07.0948 EST)] --

  As for the dictate at the core of this discussion - namely,
  that behavior is the control of perception - I'm not sure I
  buy it. PCT, as you've explained it, Bill, deals with a
  closed-loop view of human functioning. In loops, any choice
  of cause and and effect is pretty darn arbitrary. I could
  just as easily say that behavior is the control of discrepancies

  beween perceptions and reference conditions. Or perhaps that
  discrepancies between reference conditions and perceptions
  control behavior. Or, heaven forbid, I might even say that
  disturbances are controlling perceptions.

Causality may be traced around the loop, but it is not true that each
effect has one exclusive cause.

In particular, the state of the input perception is influenced not only
by the behavioral outputs of the system but also by disturbances which,
outside an artificial laboratory situation, are indeterminately many.
(There may be disturbances to values at any point in the loop, including
control of one perception interfering with control of another.)

Despite disturbances, the signature effect of control is to reduce the
difference between the reference perception and the input perception and
to maintain that difference near zero. This effect is not caused by any
one thing in the loop. It has as its cause the functioning through time
of the entire control loop. This effect specifically is not caused by
disturbances; to the contrary, the control loop nullifies any causative
effect that disturbances might have on the difference between the input
perception and the reference perception.

But I don't think it's necessary to say "behavior is the control of
differences between input perceptions and reference perceptions" (or
"between what you want and what you get", or "control of error", etc.).
You alluded to the perhaps necessary conceit that our perceptions *are*
reality. Accordingly as we conceive that our behavior controls reality,
it is appropriate to acknowledge that our behavior controls our
perceptions of reality, which we trust are real.

Does that help make it seem less a "dictate" and more a finding?

  /BN

[Martin_Taylor2]

[Martin Taylor 2006.11.07.11.52]

[From Bruce Nevin (2006.11.07.11.21 EST)] --

Fred Nickols (2006.11.07.0948 EST)] --

As for the dictate at the core of this discussion - namely,
that behavior is the control of perception - I'm not sure I
buy it. PCT, as you've explained it, Bill, deals with a
closed-loop view of human functioning. In loops, any choice
of cause and and effect is pretty darn arbitrary. I could
just as easily say that behavior is the control of discrepancies
beween perceptions and reference conditions. Or perhaps that
discrepancies between reference conditions and perceptions
control behavior. Or, heaven forbid, I might even say that
disturbances are controlling perceptions.

Causality may be traced around the loop, but it is not true that each
effect has one exclusive cause.

In particular, the state of the input perception is influenced not only
by the behavioral outputs of the system but also by disturbances which,
outside an artificial laboratory situation, are indeterminately many.
(There may be disturbances to values at any point in the loop, including
control of one perception interfering with control of another.)

Despite disturbances, the signature effect of control is to reduce the
difference between the reference perception and the input perception and
to maintain that difference near zero. This effect is not caused by any
one thing in the loop. It has as its cause the functioning through time
of the entire control loop. This effect specifically is not caused by
disturbances; to the contrary, the control loop nullifies any causative
effect that disturbances might have on the difference between the input
perception and the reference perception.

To expand on what Bruce said: In the canonical loop there are exactly two inputs from the outer world, the reference input and the disturbance input. There are two places in the loop where their effects can be directly measured (assuming one could probe anywhere that is diagrammed), the comparator output (the "eror") for the reference input, and the place where d and o (in the usual diagram, C and A in Fred's) meet (the "perceptual input"). What controls what can be assessed by asking which of the two external signals has more influence on the values at these two "probe" points.

In a well-functioning control system, if you change the reference input, the value out of the comparator (the "error") approaches zero, and the value out of the d-o summation (the "perceptual input") approaches the reference signal. If you change the disturbance signal value, the value out of the comaprator approaches zero, and the value out of the d-o convergence -- which we call the perceptual input -- still approaches the reference signal (not the disturbance signal value). So the disturbance has almost no effect at the critical points within the loop, whereas the reference signal has a considerable effect at those points.

The control loop _implements_ control of the perceptual input in response to variations in the reference signal value. The implementation, as described in the canonical loop, involves several cause-effect connections, none of which is arbitrary, and none of which individually is "control". Control is an emergent property of the loop connection, and requires an imbalance in the cause-effect gain between the two paths perception->comparator and error->output. What controls what depends on which half of the loop has the higher gain. In itself, this says that control is an emergent property of the loop, inherent in none of its separate connections.

I suppose that technically "Behaviour" (a.k.a "output" or "action") is not "control of perception". It is merely the externally visible part of the implementation of control of perception.

Martin

[Fred_Nickols1]

[From Fred Nickols (2006.11.07.1225 EST)] --

[Martin Taylor 2006.11.07.11.52]

To expand on what Bruce said: In the canonical loop there are exactly
two inputs from the outer world, the reference input and the
disturbance input.

Whoa, there! Maybe I'm dumber than I look. I thought the reference input
was part of the person, an internal factor, not something from the outer
world. Hmm. I think we need to talk more about this.

There are two places in the loop where their
effects can be directly measured (assuming one could probe anywhere
that is diagrammed), the comparator output (the "eror") for the
reference input, and the place where d and o (in the usual diagram, C
and A in Fred's) meet (the "perceptual input"). What controls what
can be assessed by asking which of the two external signals has more
influence on the values at these two "probe" points.

In a well-functioning control system, if you change the reference
input, the value out of the comparator (the "error") approaches zero,
and the value out of the d-o summation (the "perceptual input")
approaches the reference signal. If you change the disturbance signal
value, the value out of the comaprator approaches zero, and the value
out of the d-o convergence -- which we call the perceptual input --
still approaches the reference signal (not the disturbance signal
value). So the disturbance has almost no effect at the critical
points within the loop, whereas the reference signal has a
considerable effect at those points.

In a well-functioning control system I would assume that the perceptual
input is being kept aligned with the reference signal and thus there is zero
error. That's a static look, I think. In a dynamic situation, all this
stuff is varying (but let's have the reference signal stay constant). So, a
disturbance affects the controlled variable which changes the perceptual
input, an error signal results, action is taken and the controlled variable
aligns with the reference signal and there is no again error. Do I have
that right? If I do, I don't understand how changing the reference signal
would produce zero error out of the comparator; I would think it would
produce an error until such time as output (action/behavior) affects the
controlled variable and perceptual input once again aligns with the
reference signal.

There's obviously something I'm not getting.

The control loop _implements_ control of the perceptual input in
response to variations in the reference signal value. The
implementation, as described in the canonical loop, involves several
cause-effect connections, none of which is arbitrary, and none of
which individually is "control". Control is an emergent property of
the loop connection, and requires an imbalance in the cause-effect
gain between the two paths perception->comparator and error->output.
What controls what depends on which half of the loop has the higher
gain. In itself, this says that control is an emergent property of
the loop, inherent in none of its separate connections.

I suppose that technically "Behaviour" (a.k.a "output" or "action")
is not "control of perception". It is merely the externally visible
part of the implementation of control of perception.

My head hurts. I'm gonna have to chew on all this. But the two paragraphs
above prompt a couple of comments. That control is a property of the loop
seems and sounds right. Control of what is of course the issue right now.
And, right now, I'm interested in the meaning of "what controls what depends
on which half of the loop has the higher gain." That sounds like signal
strength determines what is controlling what. If so, then it would seem to
further be the case that whatever controls (i.e., determines) signal
strength is controlling whatever is being controlled.

I hate loops!

Regards,

Fred Nickols
www.nickols.us
nickols@att.net

[bnhpct]

[From Bruce Nevin (2006.11.07.13.30 EST)]

Gain is not signal strength. The higher the loop gain, the closer to
zero the error signal is maintained.

  /BN

[Martin_Taylor2]

[Martin Taylor 2006.11.07.13.34]

[From Fred Nickols (2006.11.07.1225 EST)] --

[Martin Taylor 2006.11.07.11.52]

To expand on what Bruce said: In the canonical loop there are exactly
two inputs from the outer world, the reference input and the
disturbance input.

Whoa, there! Maybe I'm dumber than I look. I thought the reference input
was part of the person, an internal factor, not something from the outer
world. Hmm. I think we need to talk more about this.

Sure, the reference signal is inside the person, but it comes from outside the control loop. If PCT is to be believed (in any of its possible forms), there are myriads of control loops within any one person -- probably millions, perhaps billions. Each of them must have a reference signal coming from outside itself (unless, of course, it is a homeostatic mechanism that has an intrinsic reference signal value prmanently set, as Bill P. construes the top-level loops to be).

There are two places in the loop where their
effects can be directly measured (assuming one could probe anywhere
that is diagrammed), the comparator output (the "eror") for the
reference input, and the place where d and o (in the usual diagram, C
and A in Fred's) meet (the "perceptual input"). What controls what
can be assessed by asking which of the two external signals has more
influence on the values at these two "probe" points.

In a well-functioning control system, if you change the reference
input, the value out of the comparator (the "error") approaches zero,
and the value out of the d-o summation (the "perceptual input")
approaches the reference signal. If you change the disturbance signal
value, the value out of the comaprator approaches zero, and the value
out of the d-o convergence -- which we call the perceptual input --
still approaches the reference signal (not the disturbance signal
value). So the disturbance has almost no effect at the critical
points within the loop, whereas the reference signal has a
considerable effect at those points.

In a well-functioning control system I would assume that the perceptual
input is being kept aligned with the reference signal and thus there is zero
error. That's a static look, I think. In a dynamic situation, all this
stuff is varying (but let's have the reference signal stay constant).

That's why I used the active form "approaches zero".

  So, a
disturbance affects the controlled variable which changes the perceptual
input, an error signal results, action is taken and the controlled variable
aligns with the reference signal and there is no again error. Do I have
that right?

Yep. Except that I would use the word "approaches" rather than "aligns with", since zero error is never actually achieved.

If I do, I don't understand how changing the reference signal
would produce zero error out of the comparator; I would think it would
produce an error until such time as output (action/behavior) affects the
controlled variable and perceptual input once again aligns with the
reference signal.

Exactly (except again substitute "approach" for "align").

There's obviously something I'm not getting.

Doesn't seem so, except possibly for the conflation of a single contol loop with an entire person.

> The control loop _implements_ control of the perceptual input in
> response to variations in the reference signal value....
>

I suppose that technically "Behaviour" (a.k.a "output" or "action")
is not "control of perception". It is merely the externally visible
part of the implementation of control of perception.

My head hurts. I'm gonna have to chew on all this. But the two paragraphs
above prompt a couple of comments. That control is a property of the loop
seems and sounds right. Control of what is of course the issue right now.
And, right now, I'm interested in the meaning of "what controls what depends
on which half of the loop has the higher gain." That sounds like signal
strength determines what is controlling what.

Not at all. Gain is what counts. if by "signal strength" you mean the value of a signal at a given moment, then a high gain and a low gain system can each output large values or small values. The difference is in how much the output value changes when the input value changes by a unit amount.

If so, then it would seem to
further be the case that whatever controls (i.e., determines) signal
strength is controlling whatever is being controlled.

Not at all. What is controlled (i.e. stabilized against the effects of external disturbances) is determined by where in the loop the different gains exist.

I hate loops!

Cheerio!

Martin

[Fred_Nickols1]

[From Fred Nickols (2006.11.07.1352)] --

Bruce Nevin (2006.11.07.13.30 EST)]

Gain is not signal strength. The higher the loop gain, the closer to
zero the error signal is maintained.

Okay; so that sounds sort of like a tolerance level; the higher the gain,
the closer or tighter the tolerance. I ask because when I turned up the
gain in the amplifiers controlling the servomechanisms in those old
electromechanical fire control computers I used to baby sit, the torque or
tightness of the servo increased. Too much gain and it got the jitters,
which we would offset with a tweaking of the other adjustment, known to us
as stability. What we were looking for was very close (i.e., tight)
tracking with no oscillation or jitter. So I think I get what you're
saying.

On the other hand, in the fire control radar, when we turned up the gain, we
were increasing the signal strength.

Thanks for taking the time to deal with a dabbler...

Regards,

Fred Nickols
www.nickols.us
nickols@att.net

[Fred_Nickols1]

[From Fred Nickols (2006.11.07/1358 EST)] --

[Martin Taylor 2006.11.07.13.34]

>[From Fred Nickols (2006.11.07.1225 EST)] --
>
> [Martin Taylor 2006.11.07.11.52]
>>
>> To expand on what Bruce said: In the canonical loop there are exactly
>> two inputs from the outer world, the reference input and the
>> disturbance input.
>
>Whoa, there! Maybe I'm dumber than I look. I thought the reference
input
>was part of the person, an internal factor, not something from the outer
>world. Hmm. I think we need to talk more about this.

Sure, the reference signal is inside the person, but it comes from
outside the control loop. If PCT is to be believed (in any of its
possible forms), there are myriads of control loops within any one
person -- probably millions, perhaps billions. Each of them must have
a reference signal coming from outside itself (unless, of course, it
is a homeostatic mechanism that has an intrinsic reference signal
value prmanently set, as Bill P. construes the top-level loops to be).

Hmm. The control loop is both inside and outside the person but does not
include the reference signal. Hmm. What does the control loop include?

>
>> There are two places in the loop where their
>> effects can be directly measured (assuming one could probe anywhere
>> that is diagrammed), the comparator output (the "eror") for the
>> reference input, and the place where d and o (in the usual diagram, C
>> and A in Fred's) meet (the "perceptual input"). What controls what
>> can be assessed by asking which of the two external signals has more
>> influence on the values at these two "probe" points.
>>
>> In a well-functioning control system, if you change the reference
>> input, the value out of the comparator (the "error") approaches zero,
>> and the value out of the d-o summation (the "perceptual input")
>> approaches the reference signal. If you change the disturbance signal
>> value, the value out of the comaprator approaches zero, and the value
>> out of the d-o convergence -- which we call the perceptual input --
>> still approaches the reference signal (not the disturbance signal
>> value). So the disturbance has almost no effect at the critical
>> points within the loop, whereas the reference signal has a
>> considerable effect at those points.
>
>In a well-functioning control system I would assume that the perceptual
>input is being kept aligned with the reference signal and thus there is
zero
>error. That's a static look, I think. In a dynamic situation, all this
>stuff is varying (but let's have the reference signal stay constant).

That's why I used the active form "approaches zero".

> So, a
>disturbance affects the controlled variable which changes the perceptual
>input, an error signal results, action is taken and the controlled
variable
>aligns with the reference signal and there is no again error. Do I have
>that right?

Yep. Except that I would use the word "approaches" rather than
"aligns with", since zero error is never actually achieved.

Well, that's good news - and bad. It's good news because I got it right.
It's bad news because a long, long time ago, when I first joined up, I said
something about being taught that a servomechanism relied on the presence of
an error signal in order to function; that without any error signal, it
would just sit there. Moreover, zero error was never realized and so the
servomechanisms never really got to zero error; instead, they oscillated
about the point of alignment or synchronization even though this oscillation
was detectable without instruments and didn't make any functional
differences. If I remember correctly, Bill corrected my thinking. Now I'm
really confused.

> If I do, I don't understand how changing the reference signal
>would produce zero error out of the comparator; I would think it would
>produce an error until such time as output (action/behavior) affects the
>controlled variable and perceptual input once again aligns with the
>reference signal.

Exactly (except again substitute "approach" for "align").

No problem; I think. Approach is good; I guess align is okay too if
stipulated that it doesn't signify zero error.

>There's obviously something I'm not getting.

Doesn't seem so, except possibly for the conflation of a single
contol loop with an entire person.

I can see how you might get that from what I've been saying but, FWIW, I
don't think of a person as a single control loop. As you indicated above,
there are probably millions if not billions (I don't know).

> > The control loop _implements_ control of the perceptual input in
> > response to variations in the reference signal value....
> >
>> I suppose that technically "Behaviour" (a.k.a "output" or "action")
>> is not "control of perception". It is merely the externally visible
>> part of the implementation of control of perception.
>
>My head hurts. I'm gonna have to chew on all this. But the two
paragraphs
>above prompt a couple of comments. That control is a property of the
loop
>seems and sounds right. Control of what is of course the issue right
now.
>And, right now, I'm interested in the meaning of "what controls what
depends
>on which half of the loop has the higher gain." That sounds like signal
>strength determines what is controlling what.

Not at all. Gain is what counts. if by "signal strength" you mean the
value of a signal at a given moment, then a high gain and a low gain
system can each output large values or small values. The difference
is in how much the output value changes when the input value changes
by a unit amount.

> If so, then it would seem to
>further be the case that whatever controls (i.e., determines) signal
>strength is controlling whatever is being controlled.

Not at all. What is controlled (i.e. stabilized against the effects
of external disturbances) is determined by where in the loop the
different gains exist.

I still need to chew some more on this gain and its location thing...

Thanks for the tutoring...

Regards,

Fred Nickols
www.nickols.us
nickols@att.net

[Bill_Powers1]

[From Bill Powers (2006.11.07.1230 MST)]

Fred Nickols (2006.11.07.0948 EST)] --

>I'm interested in what you compare the perception to,
> to determine that the perception is pretty close to it.

I think we compare our perceptions against (1) previous perceptions, (2) conceptual constructs, (3) reported perceptions/descriptions of others, and (4) alternate perceptions (e.g., tactile and visual).

These are all perceptions inside the person doing the comparing, aren't they? What I'm asking is how you compare one of your perceptions with something in the outside world that isn't a perception of your own.

A friend of mine and I play pool whenever we have an opportunity to be with one another. He and I agree as to which of the 16 balls on the table is the cue ball.

So if you perceive yourself pointing to what you see as a white ball, you hear your friend saying words like "Yes, that's the cue ball." So you're comparing one of your own perceptions with other instances of your own perceptions. What I'm asking is how either you or your friend, or both of you at once, compare your individual perceptions of a white ball with whatever it is in the outside world that is causing those perceptions to appear.

As for the dictate at the core of this discussion - namely, that behavior is the control of perception - I'm not sure I buy it. PCT, as you've explained it, Bill, deals with a closed-loop view of human functioning. In loops, any choice of cause and and effect is pretty darn arbitrary. I could just as easily say that behavior is the control of discrepancies beween perceptions and reference conditions. Or perhaps that discrepancies between reference conditions and perceptions control behavior. Or, heaven forbid, I might even say that disturbances are controlling perceptions.

Since all you can ever know about is a perception, it follows that all you can ever control is a perception. But before we can deal with this point, we have to agree on the statement that all we can know about is perceptions. If you know about something in the world that is not a perception, now is the time to point it out and explain why it is not a perception.

Best,

Bill P.

[Martin_Taylor2]

Re: More on GAP-ACT
[Martin Taylor 2006.11.07.15.00]

[From Fred Nickols (2006.11.07/1358 EST)]

to

[Martin Taylor
2006.11.07.13.34]

Hmm. The control loop is both inside and outside the person but
does not
include the reference signal. Hmm.
What does the control loop include?

In your diagram,

the control loop is PAT, and the external inputs are G (reference
signal) and C (disturbance signal).

Fred to Bruce Nevin:

[From Fred Nickols (2006.11.07.1352)] –

Bruce Nevin (2006.11.07.13.30
EST)]

Gain is not signal strength. The
higher the loop gain, the closer to
zero the error signal is
maintained.

Okay; so that sounds sort of like a
tolerance level; the higher the gain,
the closer or tighter the tolerance.
I ask because when I turned up the
gain in the amplifiers controlling the
servomechanisms in those old
electromechanical fire control computers
I used to baby sit, the torque or
tightness of the servo increased.
Too much gain and it got the jitters,
which we would offset with a tweaking of
the other adjustment, known to us
as stability. What we were looking
for was very close (i.e., tight)
tracking with no oscillation or jitter.
So I think I get what you’re
saying.

On the other hand, in the fire control
radar, when we turned up the gain, we
were increasing the signal
strength.

That’s mixing two things, link input-output gain and “loop
gain”. If you have a simple input-output connection through an
amplifier:

 Input----->|

Amplifier>---->Output,

and you keep the input constant while increasing the amplifier
gain, then certainly the output signal strength increases as the
amplifier gain increases.

If you have a control loop, the different individual input-output
links can have different gains, but because they are connected in a
loop, that doesn’t necessarily mean higher gain means greater signal
strength in any particular part of the loop. As you noted above,
higher gain (up to a point) in your amplifiers meant a lower signal
strength for the error signal.

<(Aside)> However, when you increased the gain too much,
you got to a situation where the gain and the phase shift around the
loop worked together to give you unwanted oscillation. That’s really
outside the scope of the present discussion, although it’s very
important when dealing with gain and tolerance in real control
systems. <(/Aside)>

Bruce talked about “loop gain”, which is simply the
product of all the link gains multiplied around the loop. What he said
is true only if changes in the loop gain are achieved by changing the
gain of the output function (leading to A in your diagram). Quite the
reverse happens if the loop gain is increased by increasing the
perceptual input gain (leading to P in your diagram) rather than the
output gain.

Martin

[Fred_Nickols1]

[From Fred Nickols (2006.11.07.1830 EST)] --

Bill Powers (2006.11.07.1230 MST)]

Fred Nickols (2006.11.07.0948 EST)] --

> >I'm interested in what you compare the perception to,
> > to determine that the perception is pretty close to it.
>
>I think we compare our perceptions against (1) previous perceptions,
>(2) conceptual constructs, (3) reported perceptions/descriptions of
>others, and (4) alternate perceptions (e.g., tactile and visual).

These are all perceptions inside the person doing the comparing,
aren't they? What I'm asking is how you compare one of your
perceptions with something in the outside world that isn't a
perception of your own.

Aw, c'mon, Bill; you know better than that. I know as well as you do that
"all we know of our world we know by way our perceptions" (to quote myself).

You asked me what I compare my perceptions against "to determine that the
perception is pretty close to it." Is it my fault that all I have is other
perceptions?

<snip>

>As for the dictate at the core of this discussion - namely, that
>behavior is the control of perception - I'm not sure I buy it. PCT,
>as you've explained it, Bill, deals with a closed-loop view of human
>functioning. In loops, any choice of cause and and effect is pretty
>darn arbitrary. I could just as easily say that behavior is the
>control of discrepancies beween perceptions and reference
>conditions. Or perhaps that discrepancies between reference
>conditions and perceptions control behavior. Or, heaven forbid, I
>might even say that disturbances are controlling perceptions.

Since all you can ever know about is a perception, it follows that
all you can ever control is a perception.

I believe your paragraph immediately above is what the logicians call a "non
sequitur."

Given that all I can ever know about is a perception, I don't believe it
necessarily follows that all I can ever control is a perception. Might it
be the case that I can control things I don't "know about"? (Naturally, my
assertion is subject to what you mean by "know about.")

Might it be the case that I am controlling "controlled quantities" - to use
your term - of which I am unaware? If so, does it not then follow that
"control" might quickly sort into "conscious" control and "unconscious" (or
unaware) control? If not, does it follow that "control" is a conscious
process? (I think not because at the lower levels of your hierarchy,
awareness in the sense of a conscious, sentient being does not seem to be a
factor).

But before we can deal with
this point, we have to agree on the statement that all we can know
about is perceptions. If you know about something in the world that
is not a perception, now is the time to point it out and explain why
it is not a perception.

I think we agree on that; see my self-referenced quote above.

Regards,

Fred Nickols
www.nickols.us
nickols@att.net

[Fred_Nickols1]

[From Fred Nickols (2006.11.08.0908 EST)] --

Bill Powers (2006.11.07.0750 MST)]

> Richard Pfau (2006.11.07) --

> The approach seems quite compatible with and complementary to Locke
> and Latham's findings and preachings concerning relationships between
> clear specific goals and improved performance in work situations.

You might try writing to Locke and Latham about this. Both my wife Mary
and I tried for at least four years to communicate with Locke (as well
as Bandura), but he was violently against my work and that of anyone else
who used it. When Mary asked Locke f he had read my book, Behavior: the
control of perception, he said he didn't need to, since the title itself
was "incorrect."

I thought I'd warn you that if you have some idea of bringing the two
views together you might run into some unexpected opposition.

One of the annoying aspects of hanging out on this list is that I am
regularly reminded of all the stuff I don't know about. Locke & Latham is a
case in point. So, off I went to the web to see what I could find - and I
found a lot. One piece I found was a 2002 paper Locke & Latham published in
the American Psychologist titled "Building a Practically Useful Theory of
Goal Setting and Task Motivation" subtitled "A 35-Year Odyssey."

After reading it, I was reminded of Tom Peters' zinging of psychology when
he called it "A field known for its non-findings." In any event, I read the
paper. Here is all they had to say of Powers and control theory (two
paragraphs in the Feedback section on page 708):

  "Control theory (Carver & Scheier, 1981) also emphasizes the
importance of goal setting and feedback for motivation. The assumptions
that underlie control theory, however, are questionable (Locke, 1991a, 1994;
Locke & Latham, 1990). In essence, the theory is based on a machine model
derived from cybernetic engineering (Powers, 1978). The source of
motivation is asserted to be a negative feedback loops (such as that
characterizing a thermostat) that eliminates goal-performance discrepancies.
The natural state of the organism is, by implication, one of motionlessness
or rest.

  "Control theory is in effect a mechanistic version of Hull's drive
reduction theory, which was abandoned decades ago. However, machines do not
possess internal motivational states and do not have goals of their own.
Their "goals" are those of the machine's builders. Furthermore, discrepancy
reduction is a consequence rather than a cause of goal-directed behavior.
As Bandura (1989) stated, goal setting is first and foremost a
discrepancy-creating process. Motivation requires feed-forward control in
addition to feedback. After people attain the goal they have been pursuing,
they generally set a higher goal for themselves. This adoption of higher
goals creates rather than reduces motivation discrepancies to be mastered.
"Self motivation thus involves a dual cyclic process of disequilibratory
discrepancy production followed by equilibratory reduction" (Bandura, 1989,
p.38)."

In short, Locke & Latham seem rather dismissive of control theory.

Regards,

Fred Nickols
www.nickols.us
nickols@att.net