complex adaptive systems, evolvability vs robustness, degeneracy

[MartinT]

[Martin Taylor 2010.03.17.10.04 Begorrah]

[From Bruce Gregory (2010.03.17.0930 EDT)]

[From Bill Powers (2010.03.16.1915 MST)]

By the way, I’ve concluded that you just can’t help the constant
sniping,
and I’ve decided that I don’t have to object to it. So feel
free.

That is very generous of you, but I will try to avoid constant
sniping by keeping my thoughts to myself. I can see that most of the
data that I find interesting is of no interest to you; rather, it only
serves to irritate you. I should have realized this when you found
controversial the claim that learning modifies neural networks. Since I
cannot imagine the kind of data that would convince you of the merits
of such an outrageous claim, the most prudent course of action is to
join the majority of those on CSGnet who manage to avoid the
displeasure of the keepers of the gates by remaining silent.

For some reason, the discussion on CSGnet reminds me of the
“debate” about healthcare reform. One such folly is more than enough.

I thought this mailing list was supposed to be about science, not
personalities, and possibly about emotion as a topic of research, not
as the content of a message. Persuading others that you are right isn’t
the objective, is it? At least as I see it, the objective is to put out
the evidence and one’s interpretations of the evidence, and letting
others put out other, possibly contradictory, evidence and
interpretations so that all can investigate where the evidence leads.

We are human, and we probably can’t avoid feeling emotionally involved
with our own ideas, which implies that we should expect others also to
defend theirs. If a person is controlling a perception of self as being
competent to understand the implications of evidence, one contributory
perception that may need to be controlled is that their existing
interpretations of data have been correct. We have to expect people who
have a theory that has been criticised but that has (in their mind)
survived the criticism to have developed a variety of mechanisms to
resist disturbances to that controlled perception, whatever the value
of the theory in the minds of others, and whatever the rational
validity of the criticism. That’s true of the weird US health care
“debate” as much as it is of anything that goes on in CSGnet.

The people you really address when presenting ideas or data in a
discussion group like this are not those with whom you are overtly
conducting a dialogue, but all those onlookers who have no special
commitment to one side or the other of the dialogue. It doesn’t matter
whether Bill P. accepts or rejects a particular claim you make. What
matters is whether to an uncommitted observer his reasons for rejecting
it (if he does) seem more valid than yours do for presenting it. What
matters eventually is the degree to which the theory is able to explain
observations that matter to people.

You may be right in some particular case or you may be wrong, or
somewhere in between. Two of you may not even be thinking of the same
thing when you seem to disagree. But in the end, you have to ask
yourself what you are controlling for. Are you controlling for
perceiving Bill P to be convinced of your correctness, for yourself to
understand the science better by expounding some aspect of it publicly,
for convincing a few lurkers of your ideas, or for something else?
Possibly you are controlling a bit for some of each. I think I am!

If the most important controlled perception for you is that Bill P
accepts what you write, and you perceive that you run into a brick wall
no matter what you present as evidence, then perhaps it makes sense to
use your intellectual resources elsewhere. But if it is more important
that you advance your own understanding by exposing your ideas for
criticism that you can yourself evaluate, then it does not make sense
to remain silent. If it is more important to you that the science be
developed within a larger community, then it makes even less sense to
keep silent and avoid critiquing what others offer for criticism.

In my current one-sided dialogue with Rick, I find it disturbing that
Rick does not ever attempt to address questions that I raise, or
comment on predictions derived from the paper that started this thread.
What is disturbed is my controlling a perception of trying to get
critiques of the science, whereas what I get as feedback is a series of
reiterations of Rick’s religious conviction. But the problem for me is
not that I am failing to convince Rick, but that I am failing to get
ANY response, whether from Rick or from anyone else, that would test
the scientific value of my understanding of that paper. I don’t expect
to convince Rick, but I would like to be able to convince myself (or
disabuse myself of a bad idea). If I am myself the only one to critique
my ideas, I am unlikely to demolish them very easily.

If people like you remain silent (as do most readers of CSGnet), it
will be hard to develop the science of PCT within the body of Science
more generally conceived. PCT is probably the most important idea in
biology (including psychology) of the 20th century, but if the dialogue
on the main forum for discussing it remains restricted to convincing
the two people you call “gatekeepers” of the correctness of ideas that
might develop PCT, the odds are that it will fade away, only to be
independently rediscovered in the 21st or 22nd century. I don’t want
that to happen, which is the main reason I continue to contribute to
CSGnet.

So let me implore you not to remain silent. I hope I will disagree with
some of what you say, and agree with others. And vice-versa. Either way
is better than not knowing what you think. (And I address this to all
lurkers as well as to Bruce G).

Martin

[Bill_Powers1]

[From Bill Powers (2010.03.17.0813 MDT)]

Martin Taylor 2010.03.16.21.47–

MT: Both Martins read the paper
at
<
http://www.tbiomed.com/content/7/1/6 >, recognized that the actual
psychology was flawed because of the author’s ignorance of PCT, and
recognized also that the arguments in the paper were directly relevant in
a PCT context. For Rick to reiterate at length his mantra, that if a
paper mentions non-PCT research, it is therefore worthless, is quite
beside the point and damaging to the discussion.

BP: I agree that calling things worthless is not helpful. However, you
have to realize that papers like the one you cite go right over my head,
and don’t suggest anything useful to me as a result. Such papers are
worthless to me, and probably to Rick as well, not because of any
inherent lack of merit (I don’t think any of us would argue for that kind
of objectification) but because they deal with entities that don’t seem
real to either of us, and therefore don’t connect with any of the
problems we’re trying to solve. Unless you can figure out a way to
translate from your level to ours, there isn’t going to be much
communication between these universes of discourse.
The paper speaks of robustness, evolvability, and degeneracy as if they
were causal influences in the environment that anyone can see. Those
words don’t have that meaning for me – that is, as I look at things I
don’t see any robustness, evolvability, or degeneracy in them. I can, of
course, memorize some definitions, but for me they’re just words. I don’t
get that “Aha, that’s what you mean!” effect from them.
Whatever those three words point to is not in my accumulation of
experiences.

This may mean that you possess a level of perception that I don’t have
(and therefore didn’t put into the model). You may be a more evolved kind
of human being than I am. That doesn’t bother me, but it does mean that
there are things you experience and control of which I can know nothing,
so you will never be able to demonstrate to me what you mean when you
refer to those perceptions. You can communicate with other people who
perceive and control at your level, but not with me. I don’t see how it
will ever be possible for you to tell me what you mean by those words,
when I have nothing with which to connect them.

To me, with my perceptual limitations, it seems that a word like
“evolvability” is what Gregory Bateson called a “dormitive
principle.”

http://en.wiktionary.org/wiki/dormitive_principle. If something
evolves, that is explained by saying that it has a property called
“evolvability.” If you like a person, that is because the
person is likeable. If copper is easy to draw through a die, it has
ductility. Things that resist disturbance do so because they contain
robustness. The wiki puts it succinctly:“A type of tautology in
which an item is being explained in terms of the item itself, only put in
different (usually more abstract) words.”

That way of seeing these words may indicate that they are empty of
meaning, or it may mean that the meanings are at a level of perception
that is simply inaccessible to me. I’m obviously not in a position to say
which is the case. I can’t perceive what I can’t perceive. I will never
know what I am missing by being unable to understand those three words,
if there is something to understand.

Best,

Bill P.

···

The main body of my comment, the
one to which Rick claims to be replying, was addressed to his claim that
the author would have been better informed had he been aware that
reorganization used the e-coli mechanism. I asked about what Rick meant
by an “e-coli mechanism” as applied to reorganization, since
there were several ways it could be interpreted, and some difficulties
with some of those ways. I expected Rick’s reply to include some
discussion of the points that I made and the questions I asked, rather
than to contain only an argument that I fail to understand that nothing
can be learned from a paper that isn’t devoted to testing for the
controlled variable.

Apart from my comments and questions about e-coli reorganization (on
which Rick claimed expertise), I further noted that the arguments in the
Whiteacre paper made some quite specific predictions about what should be
expected from studies of reorganization, when such studies become
feasible (e.g. that reorganization should occur in overlapping modular
segments of the perceptual control structure). Not only that, but I
pointed out that the paper provides hints about how the e-coli mechanism
actually could apply in reorganization, despite that on the surface
e-coli seems inappropriate for what I called types 1, 2, and 3
reorganization. The paper also provides reasons to anticipate that most
elementary control units should be expected to show a small tolerance
dead-band for small errors. This last is quite easy to check out in
tracking studies, though it is more likely to be important at higher
levels. I’m surprised (not really) that Rick did not comment on any of
these implications of Whiteacre’s arguments for PCT.

Let me ask Rick (though I suppose I ask only for rhetorical effect, as I
think I can guess the answer): did you re-read the paper to check out
these implications for PCT either before or after I mentioned them? Did
you understand how these implications derive from the arguments in the
paper? Do you understand how the concept of “evolvability”
applies in the context of reorganization, and why some possible control
structure organizations are more rigid and less evolvable (i.e. amenable
to successful adaptation to novel conditions) than others?

In other words, Rick, do you understand why I, a person reasonably
conversant with PCT and with some possible mechanisms of reorganization,
and one wishing to further the development of PCT as a theory, would find
this paper to be potentially useful?

Martin

[rsmarken]

[From Rick Marken (2010.03.17.0815)]

Bill Powers (2010.03.16.1610 MDT) to Rick Marken (2010.03.16.1410) --

Defensiveness begets defensiveness, and to the other side, looks like an
attack. I applaud your effort to de-escalate. It� would help even more if
you laid out more specifically what it is about conventional psychology that
needs to be discarded, and what difference it makes.

I've been trying to do that for the last 30 years;-) I have a couple
papers that specifically address it; the 1997 _Psych Methods_ paper
and the 2009 _Review of General Psychology_ paper. And a couple of
other papers touch on the issue (liek the 2005 JEP:HPP paper). I think
I've been getting better at describing the issue. But I am still
working on trying to improve my arguments.

I have objections to conventional methods that would hold even if PCT didn't
exist. The use of population statistics to explain or predict individual
behavior is, to my mind, a mistake that needs to be rejected ...

Sure, I have those same objections. And there are many conventional
psychologists who have them as well. In my field
(perception/cognition) as well as in operant conditioning research it
is not uncommon for researchers to use individual data as the basis of
prediction (I did this in my PhD thesis, for example). The objection
to the use of population data as a basis for understanding individual
behavior is a good one, and I often mention it when I write about
problems with conventional research (saying that research on control
is done on an individual basis). But, as you note, that objection is
not informed by an understanding of control. The objections I
emphasize are the ones that are informed by an understanding of
control (which are the objections you describe in many of your
papers).

Simple logical errors should be edited out, such as petitio principii.

Of course. That's "begging the question" and it certainly should be eliminated.

The error of post hoc ergo propter hoc should be caught and eradicated before
publication.

This is the flaw of "confounding" in experimental research and, of
course, every effort should be made to eliminate it though
experimental control. And, of course, this would be the case in
research based on an understanding of the nature of control as well.

My insistence on high correlations has nothing to do with PCT: I never argue
that low correlations are useless simply because PCT correlations are high.

In my latest paper (which no one seems to want to publish;-) I make
the point that by using the correct model of behavior in experiments
(a control rather than an input-output model) it should be possible to
get very high correlations between model and behavior even if control
is poor, as evidenced by low correlations between IV (disturbance) and
output (DV). So getting "high correlations" is part of my current
argument for using a control model as the basis for research in
psychology. The correlations of behavioral data with the input-output
model (as measured by statistics such as R squared) are quite low:
averaging about .6 (that's an r, not an R squared value).

Your argument about the
behavioral illusion is an argument about PCT versus other theories. It's not
a matter of how reliable the observations of cause and effect are; it's a
matter of a difference of opinion about what is cause and what is effect.
Any arguments about the behavioral illusion can be settled by ordinary
scientific methods.

Of course. I'm not arguing for new scientific methods. But I would
argue that you can't settle arguments about cause and effect (as in
the behavioral illusion) unless you are aware of the possibility that
the system under study might be a control system and that, therefore,
there might be a controlled variable between the IV and DV in the
experiment.

Best

Rick

···

--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[rsmarken]

[From Rick Marken (2010.03.17.0840)]

Bill Powers (2010.03.17.0813 MDT) to Martin Taylor (2010.03.16.21.47)--

BP: I agree that calling things worthless is not helpful. However, you have
to realize that papers like the one you cite go right over my head, and
don't suggest anything useful to me as a result. Such papers are worthless
to me, and probably to Rick as well, not because of any inherent lack of
merit (I don't think any of us would argue for that kind of objectification)
but because they deal with entities that don't seem real to either of us,
and therefore don't connect with any of the problems we're trying to solve.
Unless you can figure out a way to translate from your level to ours, there
isn't going to be much communication between these universes of discourse.

Thanks. This expresses my feelings exactly.

To me, with my perceptual limitations, it seems that a word like
"evolvability" is what Gregory Bateson called a "dormitive principle."

I have the same limitations, apparently. Indeed, I said the same thing
in an earlier post: "evolvability" sounds like a "dormitive
principle". By the way, I Wiki'd "dormitive principle" and found that
the term was apparently first coined by Moli�re in one of his plays as
a mock explanation of why opium puts you to sleep. So it's Moli�re
rather than Bateson who we can thank for the term. I like thinking of
it that way because I have always esteemed Moli�re's work far above
Bateson's

Best

Rick

···

--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[rsmarken]

[From Rick Marken (2010.03.17.0900)]

Bill Powers (2010.03.16.1915 MST) to Bruce Gregory

By the way, I've concluded that you just can't help the constant sniping,
and I've decided that I don't have to object to it. So feel free.

Sometimes that sniping really pays off. BG's snip about us only
studying tracking experiments was the basis for my "Studying Control"
post, upon which no one has commented. But I think the ideas in that
post will be the basis for my talk in Manchester. It helped me realize
something that I haven't really understood -- or at least haven't
articulated to myself -- in the 30+ years that I've been working in
this field (study of living control systems). What I realized is that
many people don't grasp the _general_ implications of the basic
demonstrations of control, such as the computer tracking experiments
and the rubber band demo. I think my talk in Manchester will be an
extension of what I started discussing in that "Studying Control"
post, and it's all thanks to BGs sniping about "just tracking tasks".
So any comments, suggestions (or snipes) on that post would be most
welcome.

Best

Rick

···

--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[MartinT]

[Martin Taylor 2010.03.17.11.25]

[From Bill Powers (2010.03.17.0813 MDT)]

Martin Taylor 2010.03.16.21.47–

MT: Both Martins read
the paper
at
<
http://www.tbiomed.com/content/7/1/6 >, recognized that the
actual
psychology was flawed because of the author’s ignorance of PCT, and
recognized also that the arguments in the paper were directly relevant
in
a PCT context. For Rick to reiterate at length his mantra, that if a
paper mentions non-PCT research, it is therefore worthless, is quite
beside the point and damaging to the discussion.

BP: I agree that calling things worthless is not helpful. However, you
have to realize that papers like the one you cite go right over my
head,
and don’t suggest anything useful to me as a result. Such papers are
worthless to me, and probably to Rick as well, not because of any
inherent lack of merit (I don’t think any of us would argue for that
kind
of objectification) but because they deal with entities that don’t seem
real to either of us, and therefore don’t connect with any of the
problems we’re trying to solve. Unless you can figure out a way to
translate from your level to ours, there isn’t going to be much
communication between these universes of discourse.

Maybe I should try to paraphrase the arguments of the paper in PCT
terminology, because I do think it offers an important advance in
understanding reorganization. I won’t do that in this message, but I
will try to answer some of your comments.

The paper speaks of robustness, evolvability, and degeneracy as if they
were causal influences in the environment that anyone can see…[Much
later] To me, with my perceptual limitations, it seems that a word like
“evolvability” is what Gregory Bateson called a “dormitive
principle.”

I didn’t read it that way at all. They are all measures of different
properties of a system. It doesn’t make much sense to say an object is
hot because it has the property “temperature”. Nor is is sensible to
say (as you did) that one can draw a copper wire because it has
“ductility”. One can measure the temperature of the object, and having
measured it, one can say that it is hot. One can measure the ductility
of copper, and having measured it, one can say that if one tries to
draw it into a wire, that will be possible. measures are not dormitive
principles.

Robustness is a measure of the level of disturbance that can change the
state of the more or less robust entity. A control system is robust
because it counters disturbance with action of its own, but is robust
only so far as its perceptual speed and output power are capable of
countering the disturbance. An armoured entity like a tortoise is
robust only insofar as the energy of the disturbance fails to crack its
shell. A steel building skeleton is more robust if it can stand when
many random girders are removed than if it collapses when one or two
are removed (though it may also have tortoise-style robustness based on
the strength of its girders). The robustness of a scientific theory is
of this structural kind, and the structural kind of robustness is what
the paper concentrates on.

Evolvability is also a measure, for which a surrogate is defined in a
footnote to the paper. It isn’t a principle, dormitive or otherwise,
any more than is “temperature”, for which several surrogate measures
are used (e.g. the length of a column of mercury in a glass tube). Just
as “temperature” is an abstraction that we currently think of as
correlating with the average momentum of the molecules of a substance,
so “evolvability” is an abstraction that Whiteacre thinks of as
correlating with the the freedom of entities to alter in response to
changing environmental circumstances. In the context of reorganization,
the “entity” would be a control unit that might have its input or
output connections changed so that its feedback loop through a changed
environment gives the control unit a different behavioural repertoire.

Degeneracy is a third measure. A system of N simultaneous equations in
U unknowns is degenerate if N<U (among other cases), and is more
degenerate the greater the excess of equations. Likewise, a control
structure is degenerate if N control units control P perceptions and
N<P.

We have seen degeneracy in a control structure only as a source of
conflict. Whiteacre, considering a different domain, sees degeneracy as
the source of both robustness and evolvability. That is what transfers
to the domain of control structures. If the control structure is
strictly orthogonal (meaning zero degeneracy), then if one of the
control units starts to control a new perception that is not a function
of the pre-existing ones, some previously controlled perception will
cease to be controlled (or to exist). In the sense of the steel
building skeleton, the control system will not be robust. Likewise, if
the control structure is degenerate, some control units can alter their
inputs or outputs to take advantage of novel environmental feedback
paths without destroying the existing ability to control, and if the
new way of doing things is effective, other control units can follow.
It offers a way e-coli optimization can work in the context of radical
reconnection events. The system gains evolvability.

Those
words don’t have that meaning for me – that is, as I look at things I
don’t see any robustness, evolvability, or degeneracy in them.

No, but with appropriate instruments you might measure them. You don’t
see the gamma radiation around you, but you can measure it with
appropriate instrumentation. I assume that the term “gamma radiation”
does have some meaning for you. If not, try “radio waves” or
“atmospheric pressure”.

I can, of
course, memorize some definitions, but for me they’re just words. I
don’t
get that “Aha, that’s what you mean!” effect from them.
Whatever those three words point to is not in my accumulation of
experiences.
…
I don’t see how it
will ever be possible for you to tell me what you mean by those words,
when I have nothing with which to connect them.

Do the above paragraphs alter this belief?

Martin

[Dick_Robertson]

[From Dick Robertson,2010.03.17.1225CDT]>

[From: Martin Taylor 2010.03.17.10.04 Begorrah]

Martin,

I often find your comments like a breath of fresh air, even when I’m not convinced that you covered all bases. I appreciate your openness and good will.

As for this statement:

[From Bruce Gregory (2010.03.17.0930 EDT)]

That is very generous of you… I should have realized this when you found controversial the claim that learning modifies neural networks… etc.

I can’t imagine Bill Powers finding it controversial that “learning modifies neural networks” since he hypothesized, at least as early as 1973 that learning IS the modification of neural networks, some score, or more, years before neurologists claimed to have discovered that.

Of course, if by saying “learning modifies neural networks” you are referring to some mysterious function, called “learning” that is an independent agent of some kind that works upon the brain, then yes, he would doubtless dispute that and/or call for evidence for the existence of such an independent agent.

But I didn’t think you meant it that way. I regard this kind of argument as mainly an example of too loosely defined meanings.

Best,

Dick R

[Bruce_Gregory6]

[From Bruce Gregory (2010.03.17.1616 EDT)]

[From Bill Powers (2010.03.17.0813 MDT)]

That way of seeing these words may indicate that they are empty of
meaning, or it may mean that the meanings are at a level of perception
that is simply inaccessible to me. I’m obviously not in a position to say
which is the case. I can’t perceive what I can’t perceive. I will never
know what I am missing by being unable to understand those three words,
if there is something to understand.

I think you state the fundamental problem very clearly.

Bruce

[_Boris_Hartman]

Hi everybody !

BG: I am beginning to understand the source of your frustration. Almost
nothing in the study of living systems can live up to your standards.

BP: Yes, that's right. I have trouble living up to them myself, but at least
I try to.

BH : I bet you do. You demonstrated it several times before. And I can agree
with you, that you are trying not so much anymore to control other people
with you opinion. But nobody can "jump out of his skin". It's your style of
conversation. That's maybe one of the reasons why PCT "revolution" can't begin.

BP : By the way, I've concluded that you just can't help the constant
sniping, and I've decided that I don't have to object to it. So feel free.

BG : That is very generous of you, but I will try to avoid constant sniping
by keeping my thoughts to myself. I can see that most of the data that I
find interesting is of no interest to you; rather, it only serves to
irritate you. I should have realized this when you found controversial the
claim that learning modifies neural networks. Since I cannot imagine the
kind of data that would convince you of the merits of such an outrageous
claim, the most prudent course of action is to join the majority of those on
CSGnet who manage to avoid the displeasure of the keepers of the gates by
remaining silent.

BG : For some reason, the discussion on CSGnet reminds me of the "debate"
about healthcare reform. One such folly is more than enough.

HB : Can we call these an attempt of control on both sides ? Can we say that
both of you feel "frustrated" ? Both of you didn't achieve your goals in
trying to persuade each other who's right. Don't we call this conflict ? As
far as I understand PCT, that's what PCT is about. To understand that
control of other people is causing damage to people relationship and to the
humanity. It leads (escalate) into quarrel and conflicts and maybe to wars
if we took for example religious conflicts about which god is right.

MT : I thought this mailing list was supposed to be about science, not
personalities, and possibly about emotion as a topic of research, not as the
content of a message. Persuading others that you are right isn't the
objective, is it? At least as I see it, the objective is to put out the
evidence and one's interpretations of the evidence, and letting others put
out other, possibly contradictory, evidence and interpretations so that all
can investigate where the evidence leads.

BH : Bravo, Martin.

BH : I'd like to ask, if anybody know, were there any points in discussion
where Bill and Gregory agreed upon ? I wasn't quite "on-line".

Best,

Boris

[_Boris_Hartman]

Hi Rick !

RM : The PCT revelation of the existence of the "behavioral illusion" is a
huge deal to me.

BH : Are you sure about this ? Is it possible that PCT is an upgrade of some
other theory which revealed the existance of the "behavioral illusion" ?

Best,

Boris

[rsmarken]

[From Rick Marken (2010.03.17.1810)]

RM : The PCT revelation of the existence of the "behavioral illusion" is a
huge deal to me.

BH : Are you sure about this ?

Yes, to the extent that I know myself at all;-)

Is it possible that PCT is an upgrade of some
other theory which revealed the existance of the "behavioral illusion" ?

Sure, but I don't know what that theory is. I certainly never heard
anything about such a theory before or after I encountered PCT and I'm
sure Bill was not aware of the existence of such a theory when he
discovered the behavioral illusion. Is there a theory you know about
that recognizes the existence of such an illusion? I think it would be
great if such a theory exists.

Best

Rick

···

2010/3/17 Boris Hartman <boris.hartman@masicom.net>:
--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[_Boris_Hartman]

Hi Rick !

RM : Sure, but I don't know what that theory is. I certainly never heard
anything about such a theory before or after I encountered PCT and I'm
sure Bill was not aware of the existence of such a theory when he
discovered the behavioral illusion. Is there a theory you know about
that recognizes the existence of such an illusion? I think it would be
great if such a theory exists.

BH : Well I just lined up some facts. I must admit that I was a little
confused when I read Ashby's book of "Adaptive behavior".

BP (earlier) : That's where my reorganization theory came from: Ashby's
homeostat.

BH : I got a feeling that many heuristics for PCT came from Ashby's book
specially "diagram of immediate effects", which I suppose makes the whole
mainframe for Ashby's "Adaptive behavior" and PCT. But I don�t know how much
of Ashby's knowledge was presented in his first book "Design for a brain",
which Bill read. I didn't got through yet. My assumption is that Ashby's
feed-back heuristics with control of stability (ultrastability) of
essential variables is the same in both books. But what I read (it's not
necessary that it's right) I got a feeling that Bill was strongly influenced
by Ashby.

BP (1998) : When the intrinsic error signal are restored to zero, all the
associated critical variables are once again near the built-in reference
states�I took these idea, incidentally from the cyberneticist W. Ross Ashby
because I thought he was right though I couldn't prove it. All that remains
is to turn the set if intrinsic error signals into learning. And here we run
into wall, not because we can't think of ways to do this, but because there
are many ways and we have no data that will tell us which of these ways the
human system actually usesďż˝ (page 50-51).

BP (2005) : My model is a direct extension of Ashby's concept of
"ultrastability" the propertyďż˝(page 184).

BH : So I concluded that mainframe for PCT came from Ashby, although we
musn't deny Bill's huge perceptual control upgrade and wonderfull
contribution to science he made with HPCT. But still I think we can't deny
some important heuristics influence from Ashby.
I also got a feeling that some other heuristics which is present in PCT was
included in Ashby's knowledge. So I found some interesting thoughts about
"behavioral illusion".

Ross A. (1960) : As the book is on internet (Arthur Dijkstra), we can all
read it. I interpreted Ashby's thinking (speccialy on pages 2-12) as
"mainframe of behavioral illussion". I think the whole book is representing
the "behavioral illusion". And as there are some similarities in Ashby'y
book that I also met in Bill's, I concluded that Bill upgraded Ashby's
theory of "adapted behavior". For example :

BP (2005) : What an organism senses affects what it does, and what it does
affects what it senses (page 41).

Ross A. (1960) : The organism affects the environment, and the environment
affects the organism. Such a system is said to have "feed-back" (page 37).
Example with riders and birds.

Ross A. : It's really important to point out to the existence of feed-back
in the relation between the free-living organism and it's environment
because most physiological experiments are deliberatelly arranged to avoid
feed-backďż˝(page 38). I could say this is an aspect of "behavioral illusion".

There are some other similar heuristics as for example Ashby mentioned
"Great machine" and Bill started the book with it. The diagram of immediate
effects and diagram in Bill's book 2005, page 191 include "practically the
same" heuristics with double feed-back and so on. Did you read Ashby's book ?

I can't deny Bill's huge contribution to diagram of feed-back control
(perceptual control) and the huge contribution he made to hierarchical
behavioral perceptual control structure. That's also what Carver and Scheier
(1997) don't deny, although they were caught in trap when they used Bill's
diagram and excluded him as a reference.

I assume that here are really tiny borders between knowledge of Wiener,
Ashby and Powers.

So I start wondering how much is in PCT really Bill's knowledge and how much
is integrated knowledge or contribution of others.
My final conclusion is that we must be very careful with such an excluding
statement as PCT revealed thatďż˝, PCT is the only theory thatďż˝, and so on.
PCT is by my opinion still in developing phase and it's still on the way to
it's perfection.

This is the second reason why I think PCT "revolution" can't start.

But of course. Please accept my apology and Bill too, if I missed something
or if I made a wrong interpretation. You know my poor English language :))
and "not understanding the PCT"...:))

Best,

Boris

P.S. But something is still puzzling me. The term reorganization. Could we
say it's similar to meaning such as "re-structuring" or changing
(modification) of organization in the moleculs in the body while energy is
acting on them or biochemical reactions are going on. Or the term means
tendency to reach the same organization as it was in the state of stability
? Probably there are some others meanings beside these.

What does it mean organization in PCT sense ? The same as in Maturana's
autopoiesis ?

Best,

Boris

[Bill_Powers1]

[From Bill Powers (2010.03.18.0845 MDT)]

BP (earlier) : That's where my reorganization theory came from: Ashby's
homeostat.

BH : I got a feeling that many heuristics for PCT came from Ashby's book
specially "diagram of immediate effects", which I suppose makes the whole
mainframe for Ashby's "Adaptive behavior" and PCT.

BP: No, that diagram is just a standard block diagram as used in electronics, drawn by a person who was unfamiliar with electronics. I was fairly adept at the design of electronic systems by the time I finished college in 1950. I learned that from textbooks.

But I don�t know how much
of Ashby's knowledge was presented in his first book "Design for a brain",
which Bill read. I didn't got through yet. My assumption is that Ashby's
feed-back heuristics with control of stability (ultrastability) of
essential variables is the same in both books. But what I read (it's not
necessary that it's right) I got a feeling that Bill was strongly influenced
by Ashby.

Yes, his "homeostat" was the basis of my concept of reorganization, as I have repeatedly said. He did understand some of the basics of negative feedback control, though I had already encountered control system engineering by that time, as a college student. Unfortunately, Ashby figured out a different way to achieve control, essentially the old notion of "compensatory responses" which was as close as conventional psychology ever came to explaining control behavior (not very close). He actually concluded that this sort of open-loop control was simpler, faster, and more accurate than negative feedback control, all three of which conclusions were wrong.

BP (1998) : When the intrinsic error signal are restored to zero, all the
associated critical variables are once again near the built-in reference
states�I took these idea, incidentally from the cyberneticist W. Ross Ashby
because I thought he was right though I couldn't prove it. All that remains
is to turn the set if intrinsic error signals into learning. And here we run
into wall, not because we can't think of ways to do this, but because there
are many ways and we have no data that will tell us which of these ways the
human system actually uses� (page 50-51).

BP (2005) : My model is a direct extension of Ashby's concept of
"ultrastability" the property�(page 184).

BH : So I concluded that mainframe for PCT came from Ashby, although we
musn't deny Bill's huge perceptual control upgrade and wonderfull
contribution to science he made with HPCT. But still I think we can't deny
some important heuristics influence from Ashby.

Of course. I have taken ideas from many sources, including textbooks on control-system engineering (which had nothing in them about reorganization), and of course my Navy training before I went to college, in which I learned how negative feedback systems of several kinds work, and how to repair them.
However, in your quote of BP (2005), the term "my model" refers not to control theory, but to reorganization theory. That would have been clear if you had quoted some of the context of that statement. I knew something about negative feedback control before I read Ashby, and learned a lot more about it later, from studying engineering textbooks and working with analog computers.

BH: I also got a feeling that some other heuristics which is present in PCT was included in Ashby's knowledge. So I found some interesting thoughts about
"behavioral illusion".

Ross A. (1960) : As the book is on internet (Arthur Dijkstra), we can all
read it. I interpreted Ashby's thinking (speccialy on pages 2-12) as
"mainframe of behavioral illussion". I think the whole book is representing
the "behavioral illusion". And as there are some similarities in Ashby'y
book that I also met in Bill's, I concluded that Bill upgraded Ashby's
theory of "adapted behavior". For example :

BP (2005) : What an organism senses affects what it does, and what it does
affects what it senses (page 41).

Ross A. (1960) : The organism affects the environment, and the environment
affects the organism. Such a system is said to have "feed-back" (page 37).
Example with riders and birds.

Ross A. : It's really important to point out to the existence of feed-back
in the relation between the free-living organism and it's environment
because most physiological experiments are deliberatelly arranged to avoid
feed-back�(page 38).

Boris H: I could say this is an aspect of "behavioral illusion".

BP: The behavioral illusion is the illusion that stimuli are causing responses, when the stimuli are actually disturbances that affect controlled variables, and the responses are actions intended to oppose the effects of the disturbances. It was the behavioral illusion that led early biologists and psychologists down the wrong track, so they came up with stimulus-response theory instead of control theory.

BH: There are some other similar heuristics as for example Ashby mentioned
"Great machine" and Bill started the book with it. The diagram of immediate
effects and diagram in Bill's book 2005, page 191 include "practically the
same" heuristics with double feed-back and so on. Did you read Ashby's book ?

BP: If you read Ashby carefully, you will see that I use a great many words that Ashby also used, though whether he used them before I did is unknown. He says "the" a great deal; also "and." And of course the term "feedback" shows up in his books frequently, though not as frequently as it should. I use that word, too. I'm pretty sure, though, that I got the term "great machine" from someone who wrote long before Ashby -- possibly from a newspaper account of the exposition in the 1890s in which the Corliss steam engine was referred to that way.

BH: So I start wondering how much is in PCT really Bill's knowledge and how much is integrated knowledge or contribution of others.

BP: I would say that most of it arose from seeing how the work of others could be put together in new ways. I certainly didn't invent control theory itself; that was done by electrical engineers in the 1930s while I was in grade school. The original idea that control systems could be found in living organisms is credited to Arturo Rosenblueth: he was a student of Walter B. Cannon who took up Claude Bernard's ideas and made them into the idea of "homeostasis." Rosenblueth passed his idea onto Norbert Wiener, who made it into cybernetics, and that was where Ashby got involved. I looked over all I knew about control systems, and all that Weiner and Ashby wrote, and saw, in 1952 and 1953, that these ideas could be developed further into a complete theory of human behavior. And so it goes in science; one idea in one person leads to the next one in someone else.

BH: P.S. But something is still puzzling me. The term reorganization. Could we say it's similar to meaning such as "re-structuring" or changing
(modification) of organization in the moleculs in the body while energy is
acting on them or biochemical reactions are going on. Or the term means
tendency to reach the same organization as it was in the state of stability
? Probably there are some others meanings beside these.

In PCT, the word reorganization is shorthand for the more technical term, "E. coli reorganization." That term refers to a very specific algorithm for a method of self-adaptation that is highly efficient, yet uses only random changes of organization to achieve convergence to the final form. Of course molecules and energy are involved, as they are in all interactions among objects, but the critical aspects of reorganization -- meaning E. coli reorganization -- are not visible at that level of description.

What does it mean organization in PCT sense ? The same as in Maturana's
autopoiesis ?

No, autopoiesis is "The process whereby a system reproduces itself". Maturana did not propose any mechanism by which a system reproduces itself. He just described and named the process of self-reproduction.

I assume you were asking about Reorganization, not "organization" as you wrote above. I mean by the term "reorganization" the alteration of properties of neural networks to change the forms of the functions they carry out.

Best,

Bill P.

···

At 04:18 AM 3/18/2010 -0500, Boris Hartman wrote:

[_Boris_Hartman]

Hi, Bill !

BP : I looked over all I knew about control systems, and all that Wiener and
Ashby wrote, and saw, in 1952 and 1953, that these ideas could be developed
further into a complete theory of human behavior. And so it goes in science;
one idea in one person leads to the next one in someone else.

BH : Nicely said. Now we have to explain this to Rick :)). O, you forgot the
"Old Anatomy book" which was also integrated in your PCTďż˝

Best,

Boris

[rsmarken]

[From Rick Marken (2010.03.18.1330)]

Hi, Bill !

BP : I looked over all I knew about control systems, and all that Wiener and
Ashby wrote, and saw, in 1952 and 1953, that these ideas could be developed
further into a complete theory of human behavior. And so it goes in science;
one idea in one person leads to the next one in someone else.

BH : Nicely said. Now we have to explain this to Rick :)).

Not really, I already knew all about the history of the development of
PCT. You had originally asked if someone before Bill had come up with
the idea of the "behavioral illusion". I said there might have been
someone but, if there was, I didn't know who it was. And you haven't
told me about anyone describing the "behavioral illusion" before
Bill. Ashby certainly didn't know about it. I'm pretty sure that Bill
Powers is the first person to have described the "behavioral illusion"
and pointed to its implications for scientific psychology.

Best

Rick

···

2010/3/18 Boris Hartman <boris.hartman@masicom.net>:

--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[_Boris_Hartman]

Hi Rick !

RM : Not really, I already knew all about the history of the development of
PCT. You had originally asked if someone before Bill had come up with
the idea of the "behavioral illusion". I said there might have been
someone but, if there was, I didn't know who it was. And you haven't
told me about anyone describing the "behavioral illusion" before
Bill. Ashby certainly didn't know about it. I'm pretty sure that Bill
Powers is the first person to have described the "behavioral illusion"
and pointed to its implications for scientific psychology.

BH : I don't understand right whether you are joking or you are serious. But
if you read Ashby's book I think it's impossible to say such a statement.

BP (earlier) : The behavioral illusion is the illusion that
stimuli are causing responses, when the stimuli
are actually disturbances that affect controlled
variables, and the responses are actions intended
to oppose the effects of the disturbances. It was
the behavioral illusion that led early biologists
and psychologists down the wrong track, so they
came up with stimulus-response theory instead of control theory.

BH : Consireding the citation :

BP (1998) : When the intrinsic error signal are restored to zero, all the

associated critical variables are once again near the built-in reference
states.I took these idea, incidentally from the cyberneticist W. Ross Ashby
because I thought he was right though I couldn't prove it. All that remains
is to turn the set if intrinsic error signals into learning. And here we

run into wall, not because we can't think of ways to do this, but because
there are many ways and we have no data that will tell us which of these
ways the human system actually uses. (page 50-51).

BH : And what's the main idea of Ashby ? "Without making any enquiry at this
stage into what has happened to the kitten's brain, we can at least say that
whereas at first the kitten's behavior was not homeostatic for skin
temperature, it has now become so. Such behavior is adapted : it preserves
the life of the animal by keeping the essential variables within limits. The
same thesis can be applied to a great deal, if not all, of the normal human
adult's behavior" (Ashby 1960, p.62). "As the kitten proceeds with trial and
errors, it's final behavior will depend on the outcome of the trials, on how
the essential variables have been effected" (p. 83). Obviously the fire is
disturbing controled variables in kitten and kitten is adapting (behaving)
as the result of stability in essential variables. Is this S-R theory ? The
cat is obviously controling essential variables what is the main goal of
living organisms. Only Ashby put it in his own words. But heuristics is the
same. PCT is controling the perception of intrinsic "errors" while Ashby is
"goal seeking" stability of essential variables when they are displaced from
a state of equlibrium (out of their physiological limits).
How can the Ashby's definition of adaptive behavior be S-R theory ? "A form
of behavior is adaptive if it maintains the essential variables within
physiological limits" (Ashby, 1960, p. 58).

BP (2005) : "INTRINSIC ERROR : A discrepancy between any intrinsic quantity
and it's intrinsic reference level. The driving force for reorganization"
(p. 297).

BH : I presume that intrinsic "error" is the same as displacement from a
state of equilibrium and reorganization is the same as step-mechanisms. It's
only play with the words. I'm not sure Rick, that you read Ashby's book. And
please consider also Bill's confession that he "took" some ideas directly
from Ashby.
And maybe you could tell me what's the main difference between Ashby's
ULTRASTABILITY "diagram of immediate effects" and Bill's diagram on page 191
(Powers, 2005) ? If anybody did that before Ashby, who was that. Please give
me the source.

I'm still standing on the position, that Bill was strongly affected by
Ashby's theory of "adapted behavior" and that Ashby was considering
"behavioral illusion" before Bill did. Also from Bill's answer I could
conclude, that he upgraded many ideas and put pieces together, not inventing
all of them as you try to present it. We can not deny Bill's huge
contribution to behavior as perceptual control. But that's not the reason
that we can subscibe to him everything what falls on your mind. Such an
excluding statements are by my opinion obstacles to development of PCT and
to PCT "revolution".

ABOUT REORGANIZATION :

BP : In PCT, the word reorganization is shorthand for the more technical
term, "E. coli
reorganization."

BP (1998) : "What we will do is to introduce an ordinary term used in a
special way, "reorganization". Reorganization is a blanket term that means
changing the way the nervous system is internally connected" (p. 51).

BH : What is reorganizing in E.colli ? What is happening in E.colli that we
can say it is reorganizing ? Are some molecules in it cell continuously
"re-structuring" and we are talking about reorganization ? What does it mean
reorganization in the E. colli cell ?

Best,

Boris

[rsmarken]

[From Rick Marken (2010.03.18.1600)]

Hi Rick !

RM : I'm pretty sure that Bill
Powers is the first person to have described the "behavioral illusion"
and pointed to its implications for scientific psychology.

BH : I don't understand right whether you are joking or you are serious. But
if you read Ashby's book I think it's impossible to say such a statement.

I read Ashby years ago and saw nothing in there about the behavioral
illusion -- the fact that the disturbance to a controlled variable
appears to be the cause of action when an observer assumes that a
control systems is actually an input-output device. I would appreciate
it if you could just copy (or paraphrase) Ashby's description of it.
It would be really interesting to see whether Ashby really did get it.

Best

Rick

···

2010/3/18 Boris Hartman <boris.hartman@masicom.net>:
--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[_Boris_Hartman]

HI, Rick

RM : I read Ashby years ago and saw nothing in there about the behavioral
illusion -- the fact that the disturbance to a controlled variable
appears to be the cause of action when an observer assumes that a
control systems is actually an input-output device. I would appreciate
it if you could just copy (or paraphrase) Ashby's description of it.
It would be really interesting to see whether Ashby really did get it

BH : Ha,ha you are very clever guy Richard. When you don't know the answers,
you start with counter questions (opposing disturbances). I've helped you
before quite some times to "reorganize" your thoughts (baseball catch,
school systemďż˝) but I'm sorry now. I've learned some lessons on CSGnet since
then, so this time you will have to find answers for yourself. O.K. I'll
give you a hint : if you will try to answer at least how bacteria E.colli
works (how it tries to maintain stability or almost constant internal
environment according to physiology, Darwin, Ashby's "diagram of immediate
effects", Maturana's autopoiesis and PCT), you would come to interesting
conclusions. Only with PCT it won't work. I'm just trying to tell you that
PCT can't answer everything. Of course it's on you whether you'll beleive me
or not. After all you are PCT, autopoietic, self-regulating,
autoregulating�being.

All I was trying to say is somehow adjusted to yours and Martin's
conclusions that PCT knows very little about genetic influence on structure
of perceptual control.

Earlier conversation :
MT : Let's consider for a moment what we know about reorganization and the

genetic influence on the structure of perceptual control. Actually, what we
know is not much.

RM : That's correct. We don't know much about it because no one is doing
any empirical research on it.

MT : > There are a few necessary constructs, such as that

intrinsic variables exist, and the PCT structure performs in such a way as
to maintain those variables within non-lethal bounds for long enough in
enough individuals that the genetic components can be passed on to following
generations. Beyond that, we know little about reorganization; that it is
done by an e-coli mechanism is somewhere between an assumption and a guess.

RM : It's called a theory.

Well then you can try at least with theoretical "mental experiments".

And by the way : did PCT find out where is coming reference signal to
eleventh level from ?

Best,

Boris

P.S. We have wonderful sunny day here in Ljubljana.

[rsmarken]

[From Rick Marken (2010.03.19.0805)]

HI, Rick

RM : I read Ashby years ago and saw nothing in there about the behavioral
illusion... I would appreciate it if you could just copy (or paraphrase) Ashby's
description of it.

BH : Ha,ha you are very clever guy Richard. When you don't know the answers,
you start with counter questions (opposing disturbances).

It doesn't seem all that clever. I always ask questions of a person
(like you) who presumably knows something (Ashby's description of the
behavioral illusion) that I want to find out about. Seems pretty
basic.

I've helped you
before quite some times to "reorganize" your thoughts (baseball catch,
school system�) but I'm sorry now. I've learned some lessons on CSGnet since
then, so this time you will have to find answers for yourself.

Why all the hostility?

O.K. I'll
give you a hint : if you will try to answer at least how bacteria E.colli
works (how it tries to maintain stability or almost constant internal
environment according to physiology, Darwin, Ashby's "diagram of immediate
effects", Maturana's autopoiesis and PCT), you would come to interesting
conclusions.

This hint suggests that we are not talking about the same "behavioral
illusion". What I mean by the "behavioral illusion" is described in
this paper: http://www.mindreadings.com/BehavioralIllusion.pdf.

And by the way : did PCT find out where is coming reference signal to
eleventh level from ?

That's not something to be "found out" but "proposed" (since that is
part of a theory). I think it has been proposed that the references
for the eleventh (highest) level control systems in the hierarchy are
fixed (there are no higher level systems to vary them) by
reorganization.

Best

Rick

···

2010/3/19 Boris Hartman <boris.hartman@masicom.net>:
--
Richard S. Marken PhD
rsmarken@gmail.com
www.mindreadings.com

[_Boris_Hartman]

Hi Rick !

RM : Why all the hostility?

BH : Well, this is the right question (it's only relevant if you saw film "I
robot") :)). We've always talked quite nicely and friendly. There was really
no reason for my attitude to you. But you are so close to Bill. It's
impossible to talk to you as Richard. It's either you or Bill who includes
in communication which is directed only to you. I don't know who is who. You
usually act like "one person". He answers instead of you and so onďż˝
I know it would be right that my hostility is directed to him. But for the
reasons I mentioned above I treat you as "one person". Sorry. We must reach
trust again if that is possible. But as I experienced life, that is usually
impossible. So the cooperation and help to each other, mutual thinking
reorganization in constructive conversation is quite an illusion if there is
no trust.

RM : That's not something to be "found out" but "proposed" (since that is
part of a theory). I think it has been proposed that the references
for the eleventh (highest) level control systems in the hierarchy are
fixed (there are no higher level systems to vary them) by
reorganization.

BH : And how reorganization is defined on 11th level ? I still have doubts
that organism on page 191 (Powers 2005) could survive if we put him in the
real environment.

Best, Boris