Self Interest

[Bill_Williams1]

From[Bill Williams 25 April 2004 10:00 PM CST]

[From Boss Man (2004.04.25)]

Probably the best way to learn this skill is to observe several examples,

discern a pattern, and

formulate a rule that allows you to reproduce the pattern.

A lot of weight seems to be resting upon this "formulate a rule." Sounds a
bit like the fabled receipe for elephant soup. Not that there is anything
neccessarily wrong with the receipe.

Is this a PCT account? No, not as far as I

know. Is this account consistent with PCT? Yes, as far as I can tell.

I think I might agree. Formulate a rule.

Bill Williams

[Boss_Man]

[From Boss Man (2004.0425)]

Bill Williams 25 April 2004 10:00 PM CST

A lot of weight seems to be resting upon this "formulate a rule." Sounds a
bit like the fabled receipe for elephant soup. Not that there is anything
neccessarily wrong with the receipe.

I agree. How do we "guess" at what the rule might be? I don't know. But in any case, guess. Does
the rule work for the examples? No? Try another rule. Yes. Congratulations.

[Martin_Taylor2]

[Martin Taylor 2004.04.26.0222]

From[Bill Williams 25 April 2004 6:20 PM CST]

> Martin Taylor 2004.04.25.1530]

>> In [Bill Williams 25 April 2004 12:20 PM CST]

>> I ask, Perhaps it might be well to clarify a point.
Is >>reorganization the same thing as learning?

Martin, in reply argues,

>Absolutely not. However, I find on page 185. of _B:CP_ in a
statement that is listed in the index as a "definition" talks
about learning in terms only of re-organization. In
the glossary "learning" is characterized as " ... a loose term."
(reminds me for some reason of Michelle) , and then there is the
entire chapter 14.

I lost or lent my copy of B:CP a long time ago, so I base my
understanding more on CSGnet discussions, including ones on this
topic with Bill P (again long ago). B:CP was published some twenty
years before those CSGnet discussions, themselves about 10 years ago.

But it doesn't really matter what Bill or I say about it. Anyone can
work out the implications of the theory for themselves. Bill's
discussions of reorganization seem consistent, so far as I can see,
and he approved of my description of it as working by way of the
side-effects on intrinisc variables of actions that control
perceptual variables.

On page 180. Powers appears to define learning in terms of
reorganization. He says, "My objective ... is to develop a theory
of reorganization..." in the following sentences he appears to
minimize the importance of other types of behavioral change that are
or have been called learning. He describes them as being "more
properly categories of behavior."

I guess I don't know what other things called learned he was
referring to, but alterations in the perceptual input functions and
in the reference input functions (other than reconnections) are
learning but are not reorganization in the ordinary sense of the
term. Hebbian learning is more appropriate for them.

Incidentally, at the same time Bill P was publishing B:CP, I
published a paper that argued that Hebbian learning in a perceptual
system with mild lateral inhibition would generate an optimal
orthogonal set of perceptual functions. Such a system in the PCT
hierarchy would minimize side-effect disturbance of related
perceptual controls. But that's by the way, except that my paper was
in the context of a different ( J.G.Taylor, 1963) theory of
perception that argued all perception to be the result of successful
control.

Bruce has asked where new perceptions come from. J.G.Taylor (no
relation) would have argued that the ability to perceive something
can develop only if one's actions affect the state of that
perception. He did quite a few experiments with rather dramatic
results that seemed to support his position. If he was right, it fits
well with PCT learning by mixed Hebbian and reorganization learning.

Now, maybe we are innately disposed to regard a perception of a
disorderly environment as an error.

I'm not clear how you are using several of the words in that
sentence, or what you intend as its implication.

But, don't we go on forever reorganizing our conceptual world
attempting to obtain a better fit between our expectations about the
world and the world we actually experience?

I imagine we do. There's something missing here, however. One of the
levels in the Powers hierarchy involves something like
algorithms--processes that act on perceived categories. Logical
processes (rule-based operations) can produce many different kinds of
results (perceptions) from the same data, and rules for working on
rules also can develop perceptions of methods. That's a kind of
learning that is more Hebbian than reorganizational, though it is
actually neither. It has quasi-instantaneous shifts of perception as
different rules come into play, not a common feature of Hebbian
systems (though it's possible), and in the absence of structural
change. In that, the rule-based operations exhibit some of the same
behaviours as Peter Small discusses in the context of attractor
basins. I think both work in the real mind.

It's late, and I'm going to bed. I hope this isn't too incoherent.

Martin

[Richard_Kennaway]

[From Richard Kennaway (2004.04.26 1100 BST)]

From[Bill Williams 25 April 2004 7:00 PM CST]
You may be right, but then this leaves us with the question of how do we
learn to "differentiate a polynomial?" Imitation is a process that comes to
mind. The role of pain in learning to ride a bicycle seems obvious.

Many things that seem obvious are not at all so, and may be obvious
in completely different ways to different people. To a behaviourist,
for example, it is obvious that the role of pain is to provide
negative reinforcement to drive the learning process. To a control
theorist, it might be obvious that pain is an irrelevant side-effect
of totally losing control of the bicycle, and that when things have
got that far out of control, learning to control the bicycle is no
longer taking place. Another control theorist might come up with a
different interpretation.

FWIW, I learnt to ride a bicycle without ever falling off. So
there's one data point for pain being unnecessary to the task.

···

At 19:00 -0500 25/4/04, Bill Williams wrote:

--
Richard Kennaway, jrk@cmp.uea.ac.uk, Richard Kennaway
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.

[Bill_Williams1]

From[Bill Williams 26 April 2004 9:50 AM CST]

[From Richard Kennaway (2004.04.26 1100 BST)]

The role of pain in learning to ride a bicycle seems obvious.

Many things that seem obvious are not at all so, and may be obvious
in completely different ways to different people. To a behaviourist,
for example, it is obvious that the role of pain is to provide
negative reinforcement to drive the learning process. To a control
theorist, it might be obvious that pain is an irrelevant side-effect
of totally losing control of the bicycle, and that when things have
got that far out of control, learning to control the bicycle is no
longer taking place. Another control theorist might come up with a
different interpretation.

FWIW, I learnt to ride a bicycle without ever falling off. So
there's one data point for pain being unnecessary to the task.

I like Richard's one data point better than all the "obvious" ones that
I had been assuming. Perhaps with Richard's counter example it
might be possible to re-examine other situations in which pain is
assumed to have an "obvious" role. Following such a re-examination
of enough situations it might then be necessary to re-examine the
"obvious" assumptions that go into contemporary conceptions of the
self, and self-interest.

Bill Williams

[Bill_Williams1]

From[Bill Williams 26 April 2004 10:20 AM CST]

[Martin Taylor 2004.04.26.0222]

>From[Bill Williams 25 April 2004 6:20 PM CST]
>
> > Martin Taylor 2004.04.25.1530]
>
> >> In [Bill Williams 25 April 2004 12:20 PM CST]
>
> >> I ask, Perhaps it might be well to clarify a point.
>Is >>reorganization the same thing as learning?

Martin in reply would prefer to start from first principles rather than
what has been said about this question-- which is OK with me.

But it doesn't really matter what Bill or I say about it. Anyone can
work out the implications of the theory for themselves.

Perhaps there is as much truth to the assertion that "Anyone can
work out the implications" as there is to the assertion that
pedophiles would make good police officers. Besides I wonder
just "How long it would take me to work out the implications?"

But, I don't think the discussion needs to linger over these no doubt
interesting issues, but there are perhaps more pertinent questions.
And, I would aggree with what you say as to .....

> Bill [Powers']

discussions of reorganization seem consistent, so far as I can see,
and he approved of my description of it as working by way of the
side-effects on intrinisc variables of actions that control
perceptual variables.

It is probably a case of read the manuals again. And, you've probably
done this a number of times already, and it seems so elementary, but
this is one of the cases where it would be nice to have a diagram of
how the loops fit together.

>On page 180. Powers appears to define learning in terms of
>reorganization. He says, "My objective ... is to develop a theory
>of reorganization..." in the following sentences he appears to
>minimize the importance of other types of behavioral change that are
>or have been called learning. He describes them as being "more
>properly categories of behavior."

I guess I don't know what other things called learned he was
referring to, but alterations in the perceptual input functions and
in the reference input functions (other than reconnections) are
learning but are not reorganization in the ordinary sense of the
term. Hebbian learning is more appropriate for them.

Here is another instance where I think I follow what you are saying
but resort to the manuals would increase my confidence.

Incidentally, at the same time Bill P was publishing B:CP, I
published a paper that argued that Hebbian learning in a perceptual
system with mild lateral inhibition would generate an optimal
orthogonal set of perceptual functions. Such a system in the PCT
hierarchy would minimize side-effect disturbance of related
perceptual controls. But that's by the way, except that my paper was
in the context of a different ( J.G.Taylor, 1963) theory of
perception that argued all perception to be the result of successful
control.

Is my understand correct that an "optimal orthogonal set of perceptual
input functions" would generate perceptions free from extraneous
confusions resulting from something category mistakes.

J.G. Taylor 1963 -- I haven't encountered this work.

Bruce has asked where new perceptions come from. J.G.Taylor (no
relation) would have argued that the ability to perceive something
can develop only if one's actions affect the state of that
perception. He did quite a few experiments with rather dramatic
results that seemed to support his position. If he was right, it fits
well with PCT learning by mixed Hebbian and reorganization learning.

Sounds as if it is well worth considering.

>Now, maybe we are innately disposed to regard a perception of a
>disorderly environment as an error.

I'm not clear how you are using several of the words in that
sentence, or what you intend as its implication.

I was assuming that a good perceptual system would generate perceptions
that made sense of the perceved environment-- that is presented the world
as a more or less orderly place. I am assuming that the world is more or
less
orderly or organism couldn't exist. So, it seemed to me reasonable to me
to assume that we have an innate disposition to work at understanding what
order there is in an environment. I think I attempt to say approximately
the
same thing later on, and it seeded to make sense to you.

> But, don't we go on forever reorganizing our conceptual world
>attempting to obtain a better fit between our expectations about the
>world and the world we actually experience?

I imagine we do.

This is what I was attempting to communicate above. I assumed that this
must be a biological innate function.

There's something missing here, however. One of the
levels in the Powers hierarchy involves something like
algorithms--processes that act on perceived categories. Logical
processes (rule-based operations) can produce many different kinds of
results (perceptions) from the same data, and rules for working on
rules also can develop perceptions of methods. That's a kind of
learning that is more Hebbian than reorganizational, though it is
actually neither. It has quasi-instantaneous shifts of perception as
different rules come into play,

Would perceptual shifts like necker (sp?) cubes be an instance of this?

not a common feature of Hebbian

systems (though it's possible), and in the absence of structural
change. In that, the rule-based operations exhibit some of the same
behaviours as Peter Small discusses in the context of attractor
basins. I think both work in the real mind.

It's late, and I'm going to bed. I hope this isn't too incoherent.

I don't see what you are saying as "incoherent." It does leave quite a
bit to be filled in-- like re-read the manuals. But I find it helpful
attempting
to improve my understand of how a control theory system could carry out
higher level operations.

But one more question, if I may, when you use the term "rule" would I be
making a mistake to substitute "filter" where you say rule?

Bill Williams

[Boss_Man]

[From Boss Man (2004.04.26)]

Richard Kennaway (2004.04.26 1100 BST)

FWIW, I learnt to ride a bicycle without ever falling off.

Bully!

[Boss_Man]

[From Boss Man (2004.04.26)]

Bill Williams 26 April 2004 9:50 AM CST

I like Richard's one data point better than all the "obvious" ones that
I had been assuming. Perhaps with Richard's counter example it
might be possible to re-examine other situations in which pain is
assumed to have an "obvious" role. Following such a re-examination
of enough situations it might then be necessary to re-examine the
"obvious" assumptions that go into contemporary conceptions of the
self, and self-interest.

I suspect that those who are enamoured of the notion that all actions are motivated by self-
interest have some very creative explanations of the motives of suicide bombers.

[Bill_Williams1]

From[Bill Williams 26 April 2004 12:30 PM CST]

[From Boss Man (2004.04.26)]

I suspect that those who are enamoured of the notion that all actions are

motivated by self-

interest have some very creative explanations of the motives of suicide

bombers.

And, what about explanations for whole cultures going off, so to speak, and
emulating the
"self-interested?" behavior of suicide bombers?"

I think the category "suicide bomber" is an very nice anomaly in the
context of the theory of
self-interest. Thanks for bringing it to my attention. However, in your
case I suspect that I
know of a self-interested motive that prompted you to do so. But, whether
self-interested or
not, thanks for the comment.

Bill Williams

[Mary_Powers]

from Mary Powers 2004.04.26

[From Boss Man (2004.04.25)]

Bill Williams 25 April 2004 7:00 PM CST
>

>You may be right, but then this leaves us with the question of how do we
>learn to "differentiate a polynomial?" Imitation is a process that comes to
>mind. The role of pain in learning to ride a bicycle seems obvious. But,
>I've found I tend to imitate other people without any conscious effort on my
>part. Is there a PCT account of imitation? Or, do you have some
>explanation for how we learn to do things such as "differentiate a
>polynomial?"

Probably the best way to learn this skill is to observe several examples,
discern a pattern, and
formulate a rule that allows you to reproduce the pattern. Is this a PCT
account? No, not as far as I
know. Is this account consistent with PCT? Yes, as far as I can tell.

I also think it is consistent, though incomplete. It may be that you can
formulate a rule on your own, or have the rule told to you. A rule, PCP
suggests, is a principle level concept, and can set reference signals at
the program level (if-then). That in turn sets reference signals for
sequences, and so on down to actually writing on a piece of paper or
pedalling the bike. However, I don't think that riding a bike requires
starting at such a high level, at least not consciously, unless you are
interested in the physics of it. Following the pattern is probably enough,
or maybe even simply reproducing a sequence. But learning to do math does
involve learning the rules.

As for whether learning is reorganization, I'm inclined to think that it is
not, unless what you are learning conflicts with something you already know
or believe. Then you must either reorganize or reject what is to be learned.

Mary P.

···

At 06:08 PM 4/25/2004, you wrote:

[Kenneth_Kitzke_Value]

In a message dated 4/24/2004 10:53:23 AM Eastern Daylight Time, powers_w@FRONTIER.NET writes:

[From Bill Powers (2004.04.24.0812 MST)]

I experience persistent error in my golf shots. The ball does not go
where I want—almost never. Now, tell me whether and when and how I
reorganize?

Sounds like you have the same problem as Tiger Woods. I hope all that money
doesn’t trouble you too much.
You are really trying to be a comedian in your old age, aren’t you Bill?

Reorganization as I define it is basically a random process, which is all
that remains when there is no systematic way (already learned) of solving a
problem. You can’t aim it consciously or on purpose – it just shuffles the
deck, and it’s up to you to keep or discard the result.
My point in my reply above to Bossman was merely that his assertion that persistent error triggers “reorganization” was misleading if not outright erroneous.

Since you are indeed the definer of the concept called “reorganization,” I hope you would correct such misleading statements and help Bossman learn something about HPCT.

It is my understanding that having persistent error in our hierarchy is indicative that we are alive. Persistent error disappears completely only when we die. Without any perception of error, we would not behave or act to try to reduce the error. It would be like being in the state of coma.

In the tracking experiments, error is persistent. No reorganization is triggered or required to deal with the error. We just behave/act to reduce the error. Is this the correct understanding?

And, I asked Bossman if when I play golf, and experience persistent error with my shots and score, how and when does reorganization come into play. I don’t think it does. Reorganization is needed for other reasons, some of which you address below. So, perhaps we can learn something from your conception of it?

If you’re already at your peak of performance, reorganizing can only make
it worse. What happens is that the performance gets better in one way and
worse in another as random changes go on, so the overall performance just
sort of wanders around near the best performance, but never remains constant.
I wonder how you know this to be true? Have you tested this? Is this true in all circumstances of behavior to some performance level?

I would think that by your definition if a golfer wants to improve his handicap (having reached a peak in personal performance level), reorganization is the only way to lower their handicap?

For example, he could buy new clubs with anti-slice characteristics, he could take lessons from a pro, he could increase his strength or agility or endurance. I would not call these changes in behavior reorganization, would you?

If they are examples of reorganization, would you claim they are random because that person has never tried them to solve their problem of a too high handicap before?

If they are not, what would be an example of reorganization be for a golfer who, unless he can win a tournament this year, will jump off a bridge? I am trying to hypothesize an intrinsic variable. Life is no longer worth living if I can’t win a golf tournament. It may be somewhat farfetched, but is not inconceivable.

If reorganization is driven by intrinsic error states as I propose, then
it’s possible to get into a condition where you’re so upset by failure (and
elated by success) that the intrinsic error never goes away, even when
there is a period of success. So you go right on reorganizing when it would
be better to let the errors die down and for reorganization to slow way
down. In my models of reorganization, you can tell when the gain is too
high: the final solution is approached but then reorganization starts
moving the system farther from the goal as often as closer to it. The
quickest results occur when the rate of random change (or the amount of
change per reshuffle) gets smaller as the error gets smaller. Of course if
it gets smaller too fast, reorganization stops when there is still
significant error left. There is an optimum gain for correcting the
intrinsic error as quickly as possible but without continuing when the
error is as small as it’s going to get.
So, please clarify this. Does reorganization come into play only when an intrinsic variable is in persistent error? Can you tell me what might be an intrinsic variable that a golfer would experience in a state of persistent error? How would the gain or rate of change apply to a golfer at any performance level? Are gain and rate of change separate and distinct attributes in reorganization by a golpher? Examples might help.

Is that last paragraph practically adoptable or determinable or is it just a theoretical construct?

The connection of reorganization with emotion works, I think, for both
positive and negative emotions. The main difference between them is that
when the error is gone, the controlled variable is at a high level for
positive and at a low level for negative emotions. In other words, we seek
the perceptions associated with positive emotions, and avoid those that go
with negative emotions.

I like this. But, if there is no error why would there be a negative emotion?

But errors are errors, and reorganization can occur as a result of trying
and failing either to achieve good perceptions or avoid having bad ones. So
reorganization can occur as a result of being in any strong state of
feeling/thinking, pleasant or unpleasant. This is, perhaps, the lesson that
the Zen masters have been teaching athletes and sports enthusiasts for
thousands of years. If you want too much either to avoid failure or to
achieve success, you will stir yourself up and keep the reorganization
going even when it becomes a liability instead of the essential means of
changing. The very intensity of your desire to achieve perfection and your
despair at falling short keep you from getting to the ultimate state of
calm and skill. To reach the peak of skill, you basically have to stop
feeling either self-critical or self-encouraging. In other words, you have
to stop reorganizing.
This is the change in your conception of reorganization that made me more comfortable with your perceptions of the role of “reorganization” by some “reorganization system” in human behavior. For, I have long maintained that the highest level human and self goals can be brought about through imagination and be done intentionally without any “random” reorganization taking place and without persistent error.

Tiger Woods’ problem is that when he makes a mistake he suffers agonies of
self-reproach (or else is hypersensitive to the click of a camera shutter).
So he is reorganizing (theoretically, you understand) much too rapidly for
the actual size of the performance error. As a result, instead of making
small changes and waiting to see the result, he makes too many changes that
are too large, and his game deteriorates. The harder he tries, the faster
it falls apart. The mantra I would have him repeat before and during each
game would be “I have 100 million dollars. Does it really matter if I win
or lose?” The answer is, of course, “Well sure, but not enough to get upset
over.”

PCT Zen.

Best,

Bill P.
This is fascinating. I doubt that you understand why Tiger Woods game deteriorates. Yet, you propose a solution? I suspect there are many reasons and angles to Tiger’s play that you, nor many of his competitors, have ever even considered. I also suspect that if small errors were not treated with high gain, Tiger would be a caddy, or a duffer like me, instead of a super star golfer.

BTW, have you heard the story about the golf match between Tiger Woods and Stevie Wonder? If not, ask me at the conference. It was written by a comedian at the peak of his performance! I know it will make you laugh out loud.

[Martin_Taylor2]

[Martin Taylor 2004.04.26 1605]

[From Boss Man (2004.04.26)]

I suspect that those who are enamoured of the notion that all
actions are motivated by self-
interest have some very creative explanations of the motives of
suicide bombers.

What's to be creative about it? I suspect that the conventional
answer that they expect an excellent reward in heaven might have
soemthing to do with self-interest.

What's more interesting is the question of the propagation of
martyrdom memes. How do such beliefs come to be so strong as to make
it worthwhile to risk losing the known elements of life (which
perhaps don't offer much reward for many of the suicide bombers).

A quite separate kind of answer might be that no actions have served
to reduce much the errors in many perceptions, in their social
milieu. They therefore may be reorganizing, but without changing
reference values at the higher levels. They find that they can
achieve at least a bit of control over their perception of the "good
life" of their enemies, by their suicide act. That's perhaps better
than the degree of control they might otherwise have.

None of the above is defensible. They are only speculative possibilities.

Martin

[Bill_Powers1]

[From Bill Powers (2004.04.26.1350 MST)]

[From Boss Man (2004.04.26)]

Bill Williams 26 April 2004 9:50 AM CST
>
>I like Richard's one data point better than all the "obvious" ones that
>I had been assuming. Perhaps with Richard's counter example it
>might be possible to re-examine other situations in which pain is
>assumed to have an "obvious" role. Following such a re-examination
>of enough situations it might then be necessary to re-examine the
>"obvious" assumptions that go into contemporary conceptions of the
>self, and self-interest.

I suspect that those who are enamoured of the notion that all actions are
motivated by self-
interest have some very creative explanations of the motives of suicide
bombers.

This is a very old argument, but it confuses achieving goals with achieving
advantages for one's self. Some goals have to do with advantages for
yourself, others with advantages for others, and the rest are neutral with
respect to who benefits. It's nice to achieve any goal, no matter what it's
about, but that doesn translate into a law of nature saying that your
reference signals must pertain only to your own state of being.

In PCT terms acheiving a goal is nothing more than making a perception
match a reference signal in one's head. So all goal-achievement involves
"self-satisfaction" and "getting what you want" and all those loaded terms
that are called "hedonistic" or "selfish." The debate gets pretty confusing
when the goal you selfishly want to achieve is for someone else to be
helped at your expense. The mere fact that it's your goal and achieving it
will lead to reduction of an error in you doesn't necessarily mean that
achievement of the goal will leave you better off in any other way but
making a perception match a reference signal. It might, but it doesn't have to.

Best,

Bill P.

···

At 11:13 AM 4/26/2004, you wrote:

[Boss_Man]

[From Boss Man (2004.04.26)]

Martin Taylor 2004.04.26 1605

What's to be creative about it? I suspect that the conventional
answer that they expect an excellent reward in heaven might have
soemthing to do with self-interest.

As a Canadian, you are perhaps unfamiliar with the American philosopher Satchel Paige who
noted, "Everybody wants to go to heaven, but nobody wants to die."

[Boss_Man]

[From Boss Man (2004.04.26)]

Bill Powers (2004.04.26.1350 MST)

In PCT terms acheiving a goal is nothing more than making a perception
match a reference signal in one's head. So all goal-achievement involves
"self-satisfaction" and "getting what you want" and all those loaded terms
that are called "hedonistic" or "selfish." The debate gets pretty confusing
when the goal you selfishly want to achieve is for someone else to be
helped at your expense. The mere fact that it's your goal and achieving it
will lead to reduction of an error in you doesn't necessarily mean that
achievement of the goal will leave you better off in any other way but
making a perception match a reference signal. It might, but it doesn't have to.

Nicely put. Dare one hope it is clear enough for Peter Small to appreciate?

[Bill_Williams1]

From[Bill Williams 26 April 2004 3:40 PM CST]

[From Bill Powers (2004.04.26.1350 MST)]

.

This is a very old argument, but it confuses achieving goals with

achieving

advantages for one's self.

I think this is a fundamental distinction that despite the age of the
argument
often is over looked. In the case of an agent who maximizes allowing an
agent to have goals that have to do with other agents results in a lot of
difficulty. It makes it hard to "prove" that a market solution is always
best.
A control theory conception of economic agency, it appears to me, does
not experience this problem. So, the problems that arise in an analysis
based upon maximization and interdependency do not occur in a control
theory analysis-- or at least it appears so to me. The control theory
analysis
with interferences between agents goals does not support the conclusion,
as best I can tell, that markets always work best. Sometimes they may,
sometimes they may not.

When you assume that an economic agent maximizes and the agents goals
are free from interdependencies then as it has been widely claimed it can be
proven that market solutions are best. There is no need to look and any
evidence. If you assume that agents are control theory type creatures and
that they have interdependencies between goals, then you need to examine
the situation to determine how best to organize things.

Bill Williams

[Boss_Man]

[From Boss Man (2004.04.26)]

···

On Mon, 26 Apr 2004 16:09:27 EDT, Kenneth Kitzke Value Creation Systems <KJKitzke@AOL.COM> wrote:

Since you are indeed the definer of the concept called "reorganization," I
hope you would correct such misleading statements and help Bossman learn
something about HPCT.

I appreciate Mr. Kitzke's concern for the state of my education. I note that he asks questions, but
only, apparently, when he believes he already knows the answers. Or perhaps he is simply
uninterested in the answers.

[Martin_Taylor2]

[Martin Taylor 2004.04.26.1749]

[From Boss Man (2004.04.26)]

Martin Taylor 2004.04.26 1605

What's to be creative about it? I suspect that the conventional
answer that they expect an excellent reward in heaven might have
soemthing to do with self-interest.

As a Canadian, you are perhaps unfamiliar with the American
philosopher Satchel Paige who
noted, "Everybody wants to go to heaven, but nobody wants to die."

Satchel Paige may have been a great pitcher, but even he didn't have
what Bill W likes to call Lady Princess Diana's spectacles, which
allow you to see inside other peoples heads, to know their reference
levels for their controlled perceptions. We've seen people who have
even gone as far as the Supreme Court (of Canada) to try to assert
their right to die. I'm not sure I wouldn't, if I were trying to live
in the conditions we read about for so many of the world's people.

I've often wondered whether the Church Fathers made suicide a sin
that barred you from heaven simply as a device to prevent Christians
from besieging the pearly gates, and thereby losing the population
race with other religions.

Martin

[Bill_Powers1]

[From Bill Powers (2004.04.26.1739 MST)

My point in my reply above to
Bossman was merely that his assertion that persistent error triggers
“reorganization” was misleading if not outright
erroneous.

Since you are indeed the definer of the concept called
“reorganization,” I hope you would correct such misleading
statements and help Bossman learn something about
HPCT.

I think (on admittedly sparse evidence) that he already knows
considerable about HPCT. I do prefer to think of reorganization as a
process that occurs at varying rates depending on the magnitude of
intrinsic error, so I might take exception to the “triggering”
idea. Maybe the problem you’re concerned with here is that of how much
intrinsic error is needed to produce significant rates of reorganization.
Normally, as you say, control errors are always present, but they are
usually quite small relative to the total range of the associated
perceptual signal. In a tracking task, where we can vary the magnitude
and speed of disturbances and so adjust the amount of error the
controller is experiencing, a mean tracking error of 10% of the total
range of target movement is enough to make most people give up – it
feels like failure.

The trickiest question you implicitly raise is the difference between
just plain error in the hierarchy, and intrinsic error, which is what I
have proposed as the driver of reorganization. Hierarchical error (i.e.,
tracking error in a learned control system) can have an indirect
connection to intrinsic error: the person may go into conflict between
continuing and quitting when hierarchical error gets large enough, or the
excessive efforts required to counteract disturbances that are large and
fast could lead to tension and fatigue. I’ve seen people get pretty
emotional when they’re not used to computers and are afraid they’re
failing some sort of test. So that kind of connection to intrinsic error
isn’t hard to understand. Ray Pavloski actually measured heart rate
during difficult tracking tasks (in which a television picture of the
participant’s hand was shown in relation to the target – but rotated by
amounts from a little to a lot). Heart rate went up with difficulty, so
that may have been a symptom of intrinsic error.

But it’s been suggested a number of times (Martin Taylor proposed this
quite a while ago) that error signals in the hierarchy could themselves
function as intrinsic error signals, in addition to their normal role in
control systems. If error signals get too high – higher than the normal
range – that can indicate (to some monitoring system) a problem without
the nature of the problem having to be understood. I assume that the
reorganizing process can’t be intelligent or make use of higher mental
functions, since it has to work from birth or perhaps before that, and it
has to work in animals, too. This is why I don’t want error signals to be
recognized in terms of what they mean – only in the sense that any large
and protracted error signifies loss of control, no matter what is being
controlled.

It is my understanding that having
persistent error in our hierarchy is indicative that we are alive.
Persistent error disappears completely only when we die. Without
any perception of error, we would not behave or act to try to reduce the
error. It would be like being in the state of
coma.

Right, That’s what I think, too.

In the tracking experiments, error
is persistent. No reorganization is triggered or required to deal
with the error. We just behave/act to reduce the error. Is
this the correct understanding?

I don’t know if it’s correct, but that agrees with my
understanding.

And, I asked Bossman if when I play
golf, and experience persistent error with my shots and score, how and
when does reorganization come into play. I don’t think it
does. Reorganization is needed for other reasons, some of which you
address below. So, perhaps we can learn something from your
conception of it?

I think it really does. Mary was commenting today about the effects of
practice, where you try something over and over (i.e., attempt to match
perceptions to the same reference picture over and over), and gradually
get better at it, you hope. That just seems like ordinary learning, but
when you delve a little deeper, I think you’ll find reorganization
lurking just beneath the surface. A record of achievement during practice
doesn’t yield just a simple monotonic curve rising to an ultimate highest
level. Instead the curve goes up and down fairly randomly, especially
when the curve starts to level off, getting better sometimes and worse
sometimes, with an overall upward trend. I think Dick Robertson did some
literature searching on that subject for his Phantom Plateau paper. It’s
that randomness that makes me think of reorganization. Why should a
systematic process ever go the wrong way? But we know that the e. Coli
method of biased random reorganization often goes the wrong way, though
it’s biased in the right direction.

If
you’re already at your peak of performance, reorganizing can only makeit
worse. What happens is that the performance gets better in one way and
worse in another as random changes go on, so the overall performance just
sort of wanders around near the best performance, but never remains
constant.

I wonder how you know this to be
true? Have you tested this? Is this true in all circumstances
of behavior to some performance level?

How I know what to be true? If reorganization is a process of random
change (trial and error), then it follows that when you’re doing as well
as possible, any change of organization will make performance worse.
That’s just logic: if the premised is true, the conclusion is true… In
support of this I then offered the observation from ordinary experience
that when one is performing at the peak, the measure of performance
wanders around some average but ceases to improve. The champion runner
does the 100-meter dash is 9.4, 9.8, 9.6. 10.2, 9.3 seconds, in
successive races. I think this hold true in just about any sport I know
of. Do the best bowlers roll nothing but perfect games? Do the tennis
champions serve nothing but aces? Does the winning bridge player never
make a bad bid? Does Tiger Woods always hit the fairway?

I would think that by your
definition if a golfer wants to improve his handicap (having reached a
peak in personal performance level), reorganization is the only way to
lower their handicap?

For example, he could buy new clubs with anti-slice characteristics, he
could take lessons from a pro, he could increase his strength or agility
or endurance. I would not call these changes in behavior
reorganization, would you?

Pretty much, yes. Buying a club with anti-slice characteristics is a stab
in the dark; your slice may be caused by something quite different from
the club, such as opening the club face on the downswing. As you know,
golf is a game with about 50 interacting parameters to adjust, and people
get better mostly by trial and error. Look at the pros who keep trying
new putters by the dozen. They’re reorganizing, not systematically
finding the root of the problem and fixing it. I’m not saying that there
is never any systematic method behind learning – only that when
systematic methods have done all they can, which often isn’t a lot, all
that is left is reorganization.

And beyond that, once you have reorganized to the point where inherent
uncertainties at various places in your control systems cause irreducible
variations in performance, reorganization can cause things to change but
there will be no further systematic improvements.

If they are examples of
reorganization, would you claim they are random because that person has
never tried them to solve their problem of a too high handicap
before?

No, I’d say they were definitely random if measures of performance showed
a fairly large random variation. There is some hope of measuring such
things.

If they are not, what would be an
example of reorganization be for a golfer who, unless he can win a
tournament this year, will jump off a bridge? I am trying to
hypothesize an intrinsic variable. Life is no longer worth living
if I can’t win a golf tournament. It may be somewhat farfetched,
but is not inconceivable.

It’s not the jumping off the bridge that shows intrinsic error, it’s the
state of the biochemical organism being far from its “design
center.” Phil Michelson, who finally won his first major tournament
(the Masters) after something like 20 years, was in the kind of bind you
may be thinking of. He really, really wanted to win a major, meaning that
he had a GREAT BIG error signal about winning. And until this year, he
was getting more and more erratic, taking changes when he knew he
shouldn’t, trying different swings and different clubs, and generally
showing what looked to me like symptoms of somebody reorganizing himself
out of top shape. This year he seemed to settle down, playing more
conservatively and more relaxed, and Bingo. The green jacket.

So, please clarify this. Does
reorganization come into play only when an intrinsic variable is in
persistent error? Can you tell me what might be an intrinsic
variable that a golfer would experience in a state of persistent
error? How would the gain or rate of change apply to a golfer at
any performance level? Are gain and rate of change separate and
distinct attributes in reorganization by a golpher? Examples might
help.

I’m just giving you my best guesses, which maybe some day could become
the basis for some experiments. Do the experiments, then we’ll really
know.

But while we’re guessing, intrinsic errors aren’t that hard to imagine.
If you find yourself in a screaming rage, stomping on your driver,
throwing your bag (or caddy) in the lagoon, your heart pounding and your
lungs heaving and your face purple, I think it would be reasonable to
guess that your intrinsic variables are not all matched nicely to their
intrinsic reference levels. I would think it safe to assume that
reorganization is going on. How you get yourself into a state like this
is an individual matter, but clearly we can do this to ourselves. If you
keep flagellating yourself for not doing better, feel shame about your
performance and outrage at what others are doing to do, you’re probably
going into reorganization. Even if you just want something nice for
yourself, so much that you’re aching and longing and yearning and
pleading for it, you’re probably putting yourself into a state conducive
to reorganization. I really don’t think you have to get any more subtle
about it than this.

But, if there is no error why would
there be a negative emotion?

Remember that in my theory of emotion, a true emotion arises when, for
any reason, the action needed to correct the error doesn’t happen or is
ineffective. So you have a state of high error, plus preparation to
action, and are unable to correct the error. So you have a negative
emotion while you want to avoid something and can’t, or while you are
yearning for something positive but haven’t yet achieved it (and here, I
go along with suggestions that the rate of change of the error probably
has something to do with the emotion). Remember the cynical phrase from
the days before women’s lib: “Sure, he’s crazy about you. Give him
what he wants and then you’ll see just how much he loves you.”
Meaning, of course, that when the error’s gone, the emotion’s gone too –
though I’m not quite that cynical about love.

This is the change in your
conception of reorganization that made me more comfortable with your
perceptions of the role of “reorganization” by some
“reorganization system” in human behavior. For, I have
long maintained that the highest level human and self goals can be
brought about through imagination and be done intentionally without any
“random” reorganization taking place and without persistent
error.

However, you were once in a state where that was not happening, and then
you were in a state where it was happening. How did that change come
about? I know you have a theory about it, but reorganization is
also able to explain that kind of change. I don’t, offhand, have any test
that would settle the issue, but I hope you agree that it is an
issue.

This is fascinating. I doubt
that you understand why Tiger Woods game deteriorates. Yet, you
propose a solution?

Well, I’ve observed his game for six or seven years now, pretty
regularly. I’ve seem him indulge in some displays of temper and disgust
at his own bad shots or bad breaks, and I’ve seen his long accurate shots
gradually develop random dispersion, to the point where right now he’s
lucky to hit half the fairways with his drives. I don’t think there’s any
doubt that the organization of his golf-playing systems has been
changing, becoming less precise, more erratic. Watching and hearing, it
seems to me he is struggling not to let his slump get him down, and
partly from my own experience and partly from theory, my best guess would
be that he’s reorganizing – that he feels worse about his game than he
wants anyone to know, and that things about his game are changing that he
would rather not change. Of course I don’t know why this is happening. As
you say,

I suspect there are many reasons
and angles to Tiger’s play that you, nor many of his competitors, have
ever even considered.

You also say,

I also suspect that if small errors
were not treated with high gain, Tiger would be a caddy, or a duffer like
me, instead of a super star golfer.

Yes, as long as his control systems are organized optimally, as they were
for five years. High gain means small error – in the hierarchy. But
even in the hierarchy, there is such a thing as too high a gain. All
hierarchical control involves irreducible time delays, and that puts a
limit on the combination of speed and accuracy of control. If you start
overcorrecting, control goes down the tube. Cranking the gain up as high
as it will go is a guarantee of loss of control.

The reorganizing system is, hypogthetically of course, much slower than
the hierarchy, since it deals in tides of biochemistry in the whole body,
not signals zipping around in a brain. If reorganization goes too fast,
there will not be time to recognize that a change in behavior has reduced
intrinsic error, and to reduce the rate of reorganization appropriately.
(I say this as if these processes involved conscious decisions; poetic
license).

I think of the whole reorganization story as a proposal being set up for
testing, not as an established truth. I’ve done quite a lot of homework
in trying to fit this theory to experiences I’ve had or heard about, but
that’s just how you get your hypotheses in shape to be put to
experimental test. Now if we just had some nifty experiments that got
right to the point …

Best,

Bill P.

[Bill_Powers1]

[Sorry for duplication – there was an error in the To field that may or
may not have bombed the transmision]

[From Bill Powers (2004.04.26.1739 MST)

My point in my reply above to
Bossman was merely that his assertion that persistent error triggers
“reorganization” was misleading if not outright
erroneous.

Since you are indeed the definer of the concept called
“reorganization,” I hope you would correct such misleading
statements and help Bossman learn something about
HPCT.

I think (on admittedly sparse evidence) that he already knows
considerable about HPCT. I do prefer to think of reorganization as a
process that occurs at varying rates depending on the magnitude of
intrinsic error, so I might take exception to the “triggering”
idea. Maybe the problem you’re concerned with here is that of how much
intrinsic error is needed to produce significant rates of reorganization.
Normally, as you say, control errors are always present, but they are
usually quite small relative to the total range of the associated
perceptual signal. In a tracking task, where we can vary the magnitude
and speed of disturbances and so adjust the amount of error the
controller is experiencing, a mean tracking error of 10% of the total
range of target movement is enough to make most people give up – it
feels like failure.

The trickiest question you implicitly raise is the difference between
just plain error in the hierarchy, and intrinsic error, which is what I
have proposed as the driver of reorganization. Hierarchical error (i.e.,
tracking error in a learned control system) can have an indirect
connection to intrinsic error: the person may go into conflict between
continuing and quitting when hierarchical error gets large enough, or the
excessive efforts required to counteract disturbances that are large and
fast could lead to tension and fatigue. I’ve seen people get pretty
emotional when they’re not used to computers and are afraid they’re
failing some sort of test. So that kind of connection to intrinsic error
isn’t hard to understand. Ray Pavloski actually measured heart rate
during difficult tracking tasks (in which a television picture of the
participant’s hand was shown in relation to the target – but rotated by
amounts from a little to a lot). Heart rate went up with difficulty, so
that may have been a symptom of intrinsic error.

But it’s been suggested a number of times (Martin Taylor proposed this
quite a while ago) that error signals in the hierarchy could themselves
function as intrinsic error signals, in addition to their normal role in
control systems. If error signals get too high – higher than the normal
range – that can indicate (to some monitoring system) a problem without
the nature of the problem having to be understood. I assume that the
reorganizing process can’t be intelligent or make use of higher mental
functions, since it has to work from birth or perhaps before that, and it
has to work in animals, too. This is why I don’t want error signals to be
recognized in terms of what they mean – only in the sense that any large
and protracted error signifies loss of control, no matter what is being
controlled.

It is my understanding that having
persistent error in our hierarchy is indicative that we are alive.
Persistent error disappears completely only when we die. Without
any perception of error, we would not behave or act to try to reduce the
error. It would be like being in the state of
coma.

Right, That’s what I think, too.

In the tracking experiments, error
is persistent. No reorganization is triggered or required to deal
with the error. We just behave/act to reduce the error. Is
this the correct understanding?

I don’t know if it’s correct, but that agrees with my
understanding.

And, I asked Bossman if when I play
golf, and experience persistent error with my shots and score, how and
when does reorganization come into play. I don’t think it
does. Reorganization is needed for other reasons, some of which you
address below. So, perhaps we can learn something from your
conception of it?

I think it really does. Mary was commenting today about the effects of
practice, where you try something over and over (i.e., attempt to match
perceptions to the same reference picture over and over), and gradually
get better at it, you hope. That just seems like ordinary learning, but
when you delve a little deeper, I think you’ll find reorganization
lurking just beneath the surface. A record of achievement during practice
doesn’t yield just a simple monotonic curve rising to an ultimate highest
level. Instead the curve goes up and down fairly randomly, especially
when the curve starts to level off, getting better sometimes and worse
sometimes, with an overall upward trend. I think Dick Robertson did some
literature searching on that subject for his Phantom Plateau paper. It’s
that randomness that makes me think of reorganization. Why should a
systematic process ever go the wrong way? But we know that the e. Coli
method of biased random reorganization often goes the wrong way, though
it’s biased in the right direction.

If
you’re already at your peak of performance, reorganizing can only makeit
worse. What happens is that the performance gets better in one way and
worse in another as random changes go on, so the overall performance just
sort of wanders around near the best performance, but never remains
constant.

I wonder how you know this to be
true? Have you tested this? Is this true in all circumstances
of behavior to some performance level?

How I know what to be true? If reorganization is a process of random
change (trial and error), then it follows that when you’re doing as well
as possible, any change of organization will make performance worse.
That’s just logic: if the premised is true, the conclusion is true… In
support of this I then offered the observation from ordinary experience
that when one is performing at the peak, the measure of performance
wanders around some average but ceases to improve. The champion runner
does the 100-meter dash is 9.4, 9.8, 9.6. 10.2, 9.3 seconds, in
successive races. I think this hold true in just about any sport I know
of. Do the best bowlers roll nothing but perfect games? Do the tennis
champions serve nothing but aces? Does the winning bridge player never
make a bad bid? Does Tiger Woods always hit the fairway?

I would think that by your
definition if a golfer wants to improve his handicap (having reached a
peak in personal performance level), reorganization is the only way to
lower their handicap?

For example, he could buy new clubs with anti-slice characteristics, he
could take lessons from a pro, he could increase his strength or agility
or endurance. I would not call these changes in behavior
reorganization, would you?

Pretty much, yes. Buying a club with anti-slice characteristics is a stab
in the dark; your slice may be caused by something quite different from
the club, such as opening the club face on the downswing. As you know,
golf is a game with about 50 interacting parameters to adjust, and people
get better mostly by trial and error. Look at the pros who keep trying
new putters by the dozen. They’re reorganizing, not systematically
finding the root of the problem and fixing it. I’m not saying that there
is never any systematic method behind learning – only that when
systematic methods have done all they can, which often isn’t a lot, all
that is left is reorganization.

And beyond that, once you have reorganized to the point where inherent
uncertainties at various places in your control systems cause irreducible
variations in performance, reorganization can cause things to change but
there will be no further systematic improvements.

If they are examples of
reorganization, would you claim they are random because that person has
never tried them to solve their problem of a too high handicap
before?

No, I’d say they were definitely random if measures of performance showed
a fairly large random variation. There is some hope of measuring such
things.

If they are not, what would be an
example of reorganization be for a golfer who, unless he can win a
tournament this year, will jump off a bridge? I am trying to
hypothesize an intrinsic variable. Life is no longer worth living
if I can’t win a golf tournament. It may be somewhat farfetched,
but is not inconceivable.

It’s not the jumping off the bridge that shows intrinsic error, it’s the
state of the biochemical organism being far from its “design
center.” Phil Michelson, who finally won his first major tournament
(the Masters) after something like 20 years, was in the kind of bind you
may be thinking of. He really, really wanted to win a major, meaning that
he had a GREAT BIG error signal about winning. And until this year, he
was getting more and more erratic, taking changes when he knew he
shouldn’t, trying different swings and different clubs, and generally
showing what looked to me like symptoms of somebody reorganizing himself
out of top shape. This year he seemed to settle down, playing more
conservatively and more relaxed, and Bingo. The green jacket.

So, please clarify this. Does
reorganization come into play only when an intrinsic variable is in
persistent error? Can you tell me what might be an intrinsic
variable that a golfer would experience in a state of persistent
error? How would the gain or rate of change apply to a golfer at
any performance level? Are gain and rate of change separate and
distinct attributes in reorganization by a golpher? Examples might
help.

I’m just giving you my best guesses, which maybe some day could become
the basis for some experiments. Do the experiments, then we’ll really
know.

But while we’re guessing, intrinsic errors aren’t that hard to imagine.
If you find yourself in a screaming rage, stomping on your driver,
throwing your bag (or caddy) in the lagoon, your heart pounding and your
lungs heaving and your face purple, I think it would be reasonable to
guess that your intrinsic variables are not all matched nicely to their
intrinsic reference levels. I would think it safe to assume that
reorganization is going on. How you get yourself into a state like this
is an individual matter, but clearly we can do this to ourselves. If you
keep flagellating yourself for not doing better, feel shame about your
performance and outrage at what others are doing to do, you’re probably
going into reorganization. Even if you just want something nice for
yourself, so much that you’re aching and longing and yearning and
pleading for it, you’re probably putting yourself into a state conducive
to reorganization. I really don’t think you have to get any more subtle
about it than this.

But, if there is no error why would
there be a negative emotion?

Remember that in my theory of emotion, a true emotion arises when, for
any reason, the action needed to correct the error doesn’t happen or is
ineffective. So you have a state of high error, plus preparation to
action, and are unable to correct the error. So you have a negative
emotion while you want to avoid something and can’t, or while you are
yearning for something positive but haven’t yet achieved it (and here, I
go along with suggestions that the rate of change of the error probably
has something to do with the emotion). Remember the cynical phrase from
the days before women’s lib: “Sure, he’s crazy about you. Give him
what he wants and then you’ll see just how much he loves you.”
Meaning, of course, that when the error’s gone, the emotion’s gone too –
though I’m not quite that cynical about love.

This is the change in your
conception of reorganization that made me more comfortable with your
perceptions of the role of “reorganization” by some
“reorganization system” in human behavior. For, I have
long maintained that the highest level human and self goals can be
brought about through imagination and be done intentionally without any
“random” reorganization taking place and without persistent
error.

However, you were once in a state where that was not happening, and then
you were in a state where it was happening. How did that change come
about? I know you have a theory about it, but reorganization is
also able to explain that kind of change. I don’t, offhand, have any test
that would settle the issue, but I hope you agree that it is an
issue.

This is fascinating. I doubt
that you understand why Tiger Woods game deteriorates. Yet, you
propose a solution?

Well, I’ve observed his game for six or seven years now, pretty
regularly. I’ve seem him indulge in some displays of temper and disgust
at his own bad shots or bad breaks, and I’ve seen his long accurate shots
gradually develop random dispersion, to the point where right now he’s
lucky to hit half the fairways with his drives. I don’t think there’s any
doubt that the organization of his golf-playing systems has been
changing, becoming less precise, more erratic. Watching and hearing, it
seems to me he is struggling not to let his slump get him down, and
partly from my own experience and partly from theory, my best guess would
be that he’s reorganizing – that he feels worse about his game than he
wants anyone to know, and that things about his game are changing that he
would rather not change. Of course I don’t know why this is happening. As
you say,

I suspect there are many reasons
and angles to Tiger’s play that you, nor many of his competitors, have
ever even considered.

You also say,

I also suspect that if small errors
were not treated with high gain, Tiger would be a caddy, or a duffer like
me, instead of a super star golfer.

Yes, as long as his control systems are organized optimally, as they were
for five years. High gain means small error – in the hierarchy. But
even in the hierarchy, there is such a thing as too high a gain. All
hierarchical control involves irreducible time delays, and that puts a
limit on the combination of speed and accuracy of control. If you start
overcorrecting, control goes down the tube. Cranking the gain up as high
as it will go is a guarantee of loss of control.

The reorganizing system is, hypogthetically of course, much slower than
the hierarchy, since it deals in tides of biochemistry in the whole body,
not signals zipping around in a brain. If reorganization goes too fast,
there will not be time to recognize that a change in behavior has reduced
intrinsic error, and to reduce the rate of reorganization appropriately.
(I say this as if these processes involved conscious decisions; poetic
license).

I think of the whole reorganization story as a proposal being set up for
testing, not as an established truth. I’ve done quite a lot of homework
in trying to fit this theory to experiences I’ve had or heard about, but
that’s just how you get your hypotheses in shape to be put to
experimental test. Now if we just had some nifty experiments that got
right to the point …

Best,

Bill P.