Beyond the Fringe (was An Opportunity for PCT PR)

[BUSINESS_GAVIN_HERMA]

( Gavin
Ritz 2010.05.19.20.10NZT)

[From
Rick Marken (2010.05.18.2330)]

Bill Powers
(2010.05.18.0911 MDT)–

Rick says he doesn’t care if anyone else accepts PCT, but I do

Well, for the record, what I said is: I am not interested in getting people to
understand that PCT is revolutionary. I am interested in getting people to
understand PCT. And I certainly would like to see people accept PCT. But I’ve
seen too many people accept PCT only to find that they accept something quite
different than what I understand PCT to be. The only problem with that (from my
perspective) is that I end up without anyone to play PCT with (well, do
research with). It’s like having people accept being your partner at
“bridge” only to find that what they call “bridge” is what
you call “gin rummy”. This makes for bridge partnerships that are as
unpleasant as some of the “partnerships” we experience with people
who “accept” PCT.

Actually Rick you should take full accountability
for this situation. You chase people away who want to understand, and then you moan
about the situation as above.

Further you complicate the situation to
such an extent that I sometimes think this is exactly what you really want. What exactly is your
controlled variable in this situation only you would know?

I have gone back to some dialogue you had
with us on another list, this is more than 10 years ago, and it wasn’t much
different.

I have taken a real interest in PCT
and have plumbed some of its depths and limits but you have shown no interest other
than a whole lot of negative responses (I cannot say what they really are
because I do not understand what your controlled variables are).

I have brought some real interesting
things to the table e.g. it is impossible to have a zero reference signal at
the highest level, and provided a very good reason why? Your responses were not
very positive, well actually very unimpressive.

PCT will not grow if this behaviour continues;
in fact a lot of behavior on this list is much to be desired for whatever
reason. I cannot hope to fathom.

I have tried to be positive and look at
many aspects of PCT; I have probably bought every book on this subject and let
me tell you some of it very very good. Runkel’s casting nets and testing
of Specimens is one of the most robust open books I have read in years, but a
theory needs more than just great books, it needs people with certain characteristics,
those that are inclusive, helpful, friendly, supportive, open, firm, understanding,
I can go on.

I use a mix of both methods of information
gathering and use, in many business situations, but I don’t like sharing
them here, because of the negative responses.

What you need to do is to change your perceptual
view.

I see a lot of people who are chased away from
PCT, pity.

Kind regards

Gavin

···

[Bruce_Gregory6]

[From Bruce Gregory (2010.05.19.2010.0650 EDT)]

[From Rick Marken (2010.05.18.2030)]

I would add that this difference would require a different approach to understanding behavior (and the mind), one oriented toward the discovery of controlled variables. I would then give a brief description of what experimental psychology would look like if it were based on the closed-loop control model of behavior.

BG: The point has been made from time to time that in most psychological experiments it is reasonable to assume that the subjects are controlling the perception “I am following the experimenter’s instructions.” In a typical learning experiment, the task presents a disturbance to controlling this perception. The outcome of the experiment is then a measure of each subject’s success in controlling the perception. What would an experiment involving the learning of pairs of nonsense syllables, say, but based on the control-loop model of behavior, look like?

Bruce

[Bruce_Gregory6]

[From Bruce Gregory (2010.05.19.0750 EDT)]

(Gavin Ritz 2010.05.19.20.10NZT)

To RM: I have taken a real interest in PCT and have plumbed some of its depths and limits but you have shown no interest other than a whole lot of negative responses (I cannot say what they really are because I do not understand what your controlled variables are).

BG: I do not wish to be critical, but I find it hard to believe that you cannot make a very informed guess as to what some of Rick’s controlled variables are. Let me suggest a few candidates:

1. I am right.

2. Only knaves and fools fail to appreciate my wisdom, which comes from my close reading of B:CP.

3. I am a humble servant of the truth surrounded by the acolytes of error.

Feel free to suggest others.

Bruce

[rsmarken]

[From Rick Marken (2010.05.19.0840)]

Martin Taylor (2010.05.18.14.36)–

Rick points out that when the reference is constant, qo is a function
of d, and the function is the inverse of the feedback loop function.

This is also true if r is not constant, in which case the equation has to explicitly include r as a variable:

qo = r - 1/g()d

The functions [feedback and system, I presume] and concepts would be conceptually and mathematically
distinct if instead of qi = qo + d, the environmental feedback path
were more complicated than a 1-1 correspondence. Then one would write
qi = E(qo) + d, and the function E() would enter into the inverse of
the loop feedback function but would not enter into the forward “mental
process” function. I think this may be why Bill’s “square-circle” demo
is so convincing.

Your function E() is equivalent to the feedback function, g(), in the equations I wrote. You might have seen this if I had written the complete derivation, as Bill suggested. My derivation of the “behavioral illusion” equation – qo = r - 1/g()d – assumes that qi = g(qo+d). Actually, in my paper describing the power law as a possible example of a behavioral illusion (http://www.mindreadings.com/BehavioralIllusion.pdf), I assume that the perceptual function itself is part of the feedback path so that p = g(qo+d).

I see Bill’s “square circle” demo as a good demonstration of the fact that we control input and not output. But I think the clearest demonstration of the behavioral illusion, where g() is very non-linear – indeed it’s a cubic – function of output, is in Experiment 4 in Bill’s 1978 Psych Review paper “Quantitative analysis of purposive systems” (reprinted in LCS I). Another pretty good demo of this “illusion” is at (http://www.mindreadings.com/ControlDemo/Illusion.html).

Best

Rick

···

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

[MartinT]

[Martin Taylor 2010.05.19.14.45]

[From Rick Marken (2010.05.19.0840)]

Martin Taylor
(2010.05.18.14.36)–

Rick points out that when the reference is constant, qo is a function
of d, and the function is the inverse of the feedback loop function.

This is also true if r is not constant, in which case the equation has
to explicitly include r as a variable:

qo = r - 1/g()d

Well, yes, it is a function of d, but it’s a function of r also, which
would make perceptual control distinguishable from S-R in an experiment
in which r was variable in some measurable way. So your comment isn’t
really relevant to Bruce’s issue that the S-R and PCT mathematics are
indistinguishable if r is constant (to which I added the constraint
that the environmental feedback function should be linear).

The functions [feedback and
system, I presume] and concepts would be conceptually and
mathematically
distinct if instead of qi = qo + d, the environmental feedback path
were more complicated than a 1-1 correspondence. Then one would write
qi = E(qo) + d, and the function E() would enter into the inverse of
the loop feedback function but would not enter into the forward “mental
process” function. I think this may be why Bill’s “square-circle” demo
is so convincing.

Your function E() is equivalent to the feedback function, g(), in the
equations I wrote.

No, it isn’t. E() is the part of the loop between qo and qi. g() is the
feedback function of the entire loop.

You might have seen this if I had written the complete
derivation, as Bill suggested.

Actually, in my first draft of the message to which you are responding,
I did write the entire derivation, but I decided that the detail
obscured the main point, so I deleted it.

My derivation of the “behavioral illusion” equation – qo
= r - 1/g()d – assumes that qi = g(qo+d). Actually, in my paper
describing the power law as a possible example of a behavioral illusion
(http://www.mindreadings.com/BehavioralIllusion.pdf),
I assume that the perceptual function itself is part of the feedback
path so that p = g(qo+d).

I’m assuming that you have a typo there, since as you have written it,
g() comes after qo and d are combined, and has nothing to do with
feedback or the loop.

Either way, your g() is not the same as my E(). If you write the
feedback loop function F() as two functions applied in sequence, an
internal h() and an external E(), so that F() = E(h()), and qi =
E(h(qo)+d), h() is precisely what an S-R experiment would measure.
Also, h() would not be distinguishable from kh()+c, since the linear
form only changes the scale and origin of the function. So if E() ==
k*()+c, you can’t tell between S-R and perceptual control using
experimental data. That was Bruce’s point.

I see Bill’s “square circle” demo as a good demonstration of the fact
that we control input and not output. But I think the clearest
demonstration of the behavioral illusion, where g() is very non-linear
– indeed it’s a cubic – function of output, is in Experiment 4 in
Bill’s 1978 Psych Review paper “Quantitative analysis of purposive
systems” (reprinted in LCS I). Another pretty good demo of this
“illusion” is at (http://www.mindreadings.com/ControlDemo/Illusion.html ).

Yes, it’s another case of E() being nonlinear. But I don’t think it
comes close to being as convincing as does the square-circle demo.
Maybe that’s just me, and other people might find the cubic
environmental feedback function more convincing. Any nonlinear function
E() will allow S-R to be distinguished from perceptual control, and
that’s what we are talking about, not which function is psychologically
most convincing.

It’s almost always true, I think, that p() is nonlinear. Usually,
low-level perceptual functions are taken to be pretty close to
logarithmic (Stevens’ power law, which you and Bill showed would be the
result if all the perceptual input functions were logarithmic, if I
remember correctly). If p() is nonlinear, the loop function would
almost necessarily be nonlinear, but unless E() is nonlinear, this
doesn’t affect the ability of an experiment to distinguish S-R and
perceptual control.

Martin

[Bill_Powers1]

[From Bill Powers (2010.05.19.1725 MDT)]

Martin Taylor 2010.05.19.14.45]

RM:This is also true if r is not
constant, in which case the equation has to explicitly include r as a
variable:

qo = r - 1/g()d

MMT : Well, yes, it is a function of d, but it’s a function of r
also, which would make perceptual control distinguishable from S-R in an
experiment in which r was variable in some measurable way.

BP: Only if r were sychronized with d. Too much is being said here
without reproducing the equations and showing the derivations.

RM: Your function E() is
equivalent to the feedback function, g(), in the equations I wrote.

MMT: No, it isn’t. E() is the part of the loop between qo and qi. g() is
the feedback function of the entire loop.

RM: You might have seen this if
I had written the complete derivation, as Bill suggested.

MMT: Actually, in my first draft
of the message to which you are responding, I did write the entire
derivation, but I decided that the detail obscured the main point, so I
deleted it.

My derivation of the
“behavioral illusion” equation – qo = r - 1/g()d – assumes
that qi = g(qo+d). Actually, in my paper describing the power law as a
possible example of a behavioral illusion
(
http://www.mindreadings.com/BehavioralIllusion.pdf), I assume that
the perceptual function itself is part of the feedback path so that p =
g(qo+d).

You guys are going all over the place. Derive the equations from the
following:

p = qi

o = f(r - p)

qi = g(o) + d

You will find that g is the function connecting the output to the
controlled variable qi, which Martin objected to, and that p is NOT g(qo

• d), as Rick said.

Do all the steps. Show your work. By this time, the best either of you
will get is a B.

Note that f(r - p) is not f(r) - f(p).

Best,

Bill P.

[rsmarken]

[From Rick Marken (2010.05.19.1700)]

Martin Taylor (2010.05.19.14.45) –

Rick Marken (2010.05.19.0840)–

qo = r - 1/g()d

Well, yes, it is a function of d, but it’s a function of r also, which
would make perceptual control distinguishable from S-R in an experiment
in which r was variable in some measurable way.

I think the only way to distinguish perceptual control from S-R – that is, the only way to determine whether the system under study is a control or an S-R system – is to test for a controlled variable. If the system under study is a control system, then the test will reveal this fact whether the reference for the controlled variable is fixed or variable.

So your comment isn’t
really relevant to Bruce’s issue that the S-R and PCT mathematics are
indistinguishable if r is constant (to which I added the constraint
that the environmental feedback function should be linear).

Your function E() is equivalent to the feedback function, g(), in the
equations I wrote.

No, it isn’t. E() is the part of the loop between qo and qi. g() is the
feedback function of the entire loop.

Actually, your E() is the same as g() in my disturbance-output equation for a control system: qo = -g-1(d) (that’s the way I should have written it). That equation is derived (well, a linear version of it; see More Mind Readings, p. 36 and 37; unfortunately I use different symbols for the variables and functions) under the assumption that qi = g(qo) - d. That is, it assumes that the function relating the disturbance to the controlled variable (qi) is just the unity multiplier. I was wrong to say that my g() was qi = g(qo+d), as I said here:

My derivation of the “behavioral illusion” equation – qo
= r - 1/g()d – assumes that qi = g(qo+d).

I’m assuming that you have a typo there, since as you have written it,
g() comes after qo and d are combined, and has nothing to do with
feedback or the loop.

It was worse than a typo. It was a purposefully produced mistake. I see that Bill caught it. Oops.

Either way, your g() is not the same as my E(). If you write the
feedback loop function F() as two functions applied in sequence, an
internal h() and an external E(), so that F() = E(h()), and qi =
E(h(qo)+d), h() is precisely what an S-R experiment would measure.

Well you lost me. I presume an S-R experiment is the conventional IV-DV experiment done in psychology, where the IV is an environmental variable and DV is a behavioral variable. PCT identifies the IV as a disturbance, d, to a controlled variable and the DV as an output, qo, which affects the controlled variable. So what is observed in an S-R experiment (according to PCT) is a relationship between qo and d. Conventional psychology takes the observed relationship between and IV and DV (qo and d) to be a reflection of characteristics of the organism. PCT shows that, if the organism under study is a control system, the relationship between IV and DV (qo and d) reflects characteristics of the organism’s environment. That is what the “behavioral illusion” equation is about. qo = g-1(d) says that the observed relationship between qo and d in what you call an S-R experiment is a reflection of (the inverse of) g(), the environmental feedback function, rather than f(), the function that represents characteristics of the system itself.

A tangible example of this “illusion”, provided to me by bike enthusiast Gary Cziko, is given in my “Revolution” paper. The IV (disturbance, d) is the appearance or non-appearance of an obstruction while driving your bike; the DV (output, qo) is the pressure exerted on the hand brakes. There will be a clear effect of the disturbance on hand brake force: With the obstruction the brake force is much higher than with no obstruction. qo (brake force) will be observe to be related to the amount of obstruction (d).

Now let’s do the same experiment on a wet street. There will still be a relationship between qo and d but the biker will appear to be much more responsive to the obstruction, squeezing the brakes with much more force when the obstruction occurs. The relationship between qo and d will have changed from what it was when the experiment was done in dry conditions.

The change in the relationship between qo and d on a wet street looks like a change in the responsiveness of the biker; in fact the change results from the change in the feedback connection between the biker’s output (braking) and desired input (stopping). Of course, in this example, one can see that there is a change in the environment that is responsible for the apparent change in responsiveness. But the point is to show that, in all cases, when you are dealing with a control system, observed relationships between IVs (disturbances) and DV (outputs) reflect characteristics of the organism’s environment, not of the organism itself.

Also, h() would not be distinguishable from kh()+c, since the linear
form only changes the scale and origin of the function. So if E() ==
k*()+c, you can’t tell between S-R and perceptual control using
experimental data. That was Bruce’s point.

It’s not that you can’t distinguish S-R from control using experimental data under these (or any other) conditions. What PCT shows is that you can’t distinguish S-R from control by simply looking at an observed relationship between qo and d. You have to also determine whether control is actually occurring, and that requires some kind of test to determine whether a controlled variable is involved.

Best

Rick

···

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

[rsmarken]

[From Rick Marken (2010.05.19.0925)

Gavin Ritz (2010.05.19.20.10NZT)--

Rick Marken (2010.05.18.2330)]

Well, for the record, what I said is: I am not interested in getting people to >>understand that PCT is revolutionary. I am interested in getting people to >>understand PCT. And I certainly would like to see people accept PCT...

Actually Rick you should take full accountability for this situation. You chase
people away who want to understand, and then you moan about the situation
as above.

I take full responsibility (I presume that's what you mean by
"accountability") for what I say about PCT and how I say it. I don't
take responsibility for what other people do because I am not them. I
cannot set their references; I can only set my own. And I am
reasonably comfortable with how I do that, though I know I'm not
perfect. I've done and said some things that I regret either because
they were factually wrong or unintentionally hurtful. I am open to
informative criticism. If there are things I have said that you think
I shouldn't have said, or that I should have said differently, please
feel free to tell me what they are and what you would rather I had
said. However, I can't work to improve my performance (from your
perspective) unless I know what _specifically_ you think is wrong with
it. It doesn't help, for example, if you just tell that I "drive
people away". Tell me exactly what it is I said that drove people away
and what I could have said instead that would not have driven them
away. But it must be specific; exact quotes from what you consider
offensive posts. OK?

Thanks

Rick

···

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

[BUSINESS_GAVIN_HERMA]

(Gavin Ritz 2010.05.19.23.28NZT)

[From Rick Marken
(2010.05.19.0925)

Gavin Ritz (2010.05.19.20.10NZT)–

Rick Marken
(2010.05.18.2330)]

Well, for the record, what I said is: I am
not interested in getting people to >>understand that PCT is
revolutionary. I am interested in getting people to >>understand PCT. And
I certainly would like to see people accept PCT…

Actually Rick you should take full accountability for this situation. You chase

people away who want to understand, and then you
moan about the situation

as above.

I take full responsibility (I presume
that’s what you mean by

“accountability”)

No it’s not what I
mean “accountability” always involves role relationships, the very essence
of control systems with each other.

for what I say about PCT and how I say it. I don’t

take responsibility for what other people do because I
am not them. I

cannot set their references; I can only set my own.

Yes but you know very
well what key emotional references individuals have, you’ve been in the people
game for years Rick.

And I am

reasonably comfortable with how I do that,

that’s fine then,
so there’s no need for you respond any further to my comments.

though I know I’m not

perfect. I’ve done and said some things that I regret
either because

they were factually wrong or unintentionally hurtful.
I am open to

informative criticism. If there are things I have said
that you think

I shouldn’t have said, or that I should have said differently,
please

feel free to tell me what they are and what you would
rather I had

said. However, I can’t work to improve my performance
(from your

perspective) unless I know what specifically you
think is wrong with

it. It doesn’t help, for example, if you just
tell that I "drive

people away". Tell me exactly what it is I said
that drove people away

and what I could have said instead that would not have
driven them

away. But it must be specific; exact quotes from what
you consider

offensive posts. OK?

Well for one in your last
email you’ve made it quite clear. Look deeper into your intentions and
find the answers for yourself. As I said I don’t know your references,
maybe you do, maybe you don’t.

Kind regards

Gavin

···

[Bruce_Gregory6]

[From Bruce Gregory (2010.05.20.0756 EDT)]

[From Rick Marken (2010.05.19.0925)

I take full responsibility (I presume that's what you mean by
"accountability") for what I say about PCT and how I say it. I don't
take responsibility for what other people do because I am not them. I
cannot set their references; I can only set my own.

BG: Not really. At least if you believe PCT. The great majority of your references are established by the hierarchy. The remainder by reorganization. Life would be a lot simpler if we could set our own reference levels.

Bruce

[BUSINESS_GAVIN_HERMA]

(Gavin Ritz
2010.05.19.23.41)

[From Bruce
Gregory (2010.05.19.0750 EDT)]

(Gavin
Ritz 2010.05.19.20.10NZT)

To RM: I
have taken a real interest in PCT
and have plumbed some of its depths and limits but you have shown no interest
other than a whole lot of negative responses (I cannot say what they really are
because I do not understand what your controlled variables are).

BG: I do not wish to be
critical, but I find it hard to believe that you cannot make a very informed
guess as to what some of Rick’s controlled variables are. Let me
suggest a few candidates:

1. I am right.

2. Only knaves and fools
   fail to appreciate my wisdom, which comes from my close reading of B:CP.

3. I am a humble
   servant of the truth surrounded by the acolytes of error.

I’m
sorry Bruce I don’t see it that way.

I really
don’t know what Rick’s reference signals are for his controlled variables, but
for someone who is supposed to understand conflict between control systems, he’s
not good at it. The essence of trust and accountability is being very sharp and
understanding another’s references. Anyway that’s my opinion.

Your
comments above are just a further spiral into negatively which I’m not
keen in descending. He does understand PCT really well and he needs to be a
leader one with wit, intelligence, empathy, caring, support, accountability and
all the effective stuff leaders need in their relationships with other
individuals. If he can’t well that’s his baby and just another mill
stone for PCT.

For me
personally I have only two interests in understanding the nature of humanity
and the nature of nature (don’t think I ever will), PCT falls into
somewhat both camps. So I’ll endure and see what happens in this space.

Regards

Gavin

[Bruce_Gregory6]

[From Bruce Gregory (2010.05.20.0903 EDT)]

(Gavin Ritz 2010.05.19.23.41)

Your comments above are just a further spiral into negatively which I’m not keen in descending. He does understand PCT really well and he needs to be a leader one with wit, intelligence, empathy, caring, support, accountability and all the effective stuff leaders need in their relationships with other individuals. If he can’t well that’s his baby and just another mill stone for PCT.

For me personally I have only two interests in understanding the nature of humanity and the nature of nature (don’t think I ever will), PCT falls into somewhat both camps. So I’ll endure and see what happens in this space.

BG: I commend you for staying above the fray. My comments were not intended to be negative, however. Quite the contrary. On my website you would find:

The author has been described (admittedly by himself) as, “a humble servant of the truth, surrounded by the acolytes of ignorance.”

Bruce

[rsmarken]

[From Rick Marken (2010.05.20.1100)]

Gavin Ritz (2010.05.19.23.41)

I really don�t know what Rick�s reference signals are for his
controlled variables

It should be pretty obvious from my posts. One variable I am
controlling for (with rather high gain) is demonstrating how to study
purposeful behavior and my reference setting for it is "by testing for
controlled variables". I'm also controlling for playing Bach like
Glenn Gould. I'm apparently not doing particularly well at controlling
for either of these. So I think I'll go control a variable that I can
control successfully: the toastiness of a couple pieces of rye bread.

Best

Rick

···

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

[_Boris_Hartman]

Hi !

BP : This means backing up a bit before getting to the equations. We need to
find a starting place where the reader can agree and understand without
knowing anything about control theory (even though we know that many readers
will have some understanding of it). One approach would then be to introduce
feedback concepts by showing first that a closed loop exists, then that the
feedback in organisms is concurrent with the perceptions, then how
disturbances relate to behavior, and finally show the two equations as
representing behavior without feedback and with feedback. The important
thing is to show how the two ways of seeing behavior differ, without saying
that one is better than the other (we want the reader to make that decision).

BH : This is a really good idea and it's really new approach. And I have
some suggestions.

1. If you want to find a starting place I suggest you start as Ashby did.
Find a good life example (like kitten and the heat source). With this
example you can show all characteristics of PCT and feed-back. You can even
show the main problem of control. How to change behavior to better. I think
cat is better problem then driving car, because driving car I think is
rather complex and "secondary" control of people's safety. Dealing with
heat, I think you come to the heart of the problem : direct control of
variying esential variables due to disturbances.
2. Then you can explain how other theories explain the same phenomenon
(casual psychology, Ashby, Glasser, Carver, and so on).
3. Find the difference between theories and show it to the reader.
4. Hope that readers wil "swallow" the bait.

Orďż˝

1. Find life example as starting point.
2. Try to explain the example with diagram on page 191 (B:CP, 2005). That
will be a real contribution to a science, because as I see it, the diagram
consists of knowledge from other people and your original contribution
(hierachical control sub-system inside the diagram). If the diagram
(organism) on page 191 "survive" or compensate the real inside and outside
environmental disturbances (whatever the example is) then you have made it.
PCT is right in this case. Try another case.
3. Then you can show how other theories deal with the same problem and to
what solution they came, and so on�The model of "organism" on page 191, with
great luck, could be proven as right when comparing with other theories on
the same example.
4. The real problem is, that I doubt the "organism" on page 191 will
"survive" in life circumstances (examples). But I see it as a good start to
make a real PCT "organism", which could be better than other aproaches. I
think the fundamental statement of PCT is right : Behavior is control of
perception and goal seeking with feed-back. You just have to prove it with
right and complete model of "organism", which will be working in real
circumstances. "Homeostat" could be a beginning.

I will be glad if you convince me, that I'm wrong about actual model on
page 191. But according to knowledge which I get lately from other sciences
and compare it to PCT knowledge, it's announcing, that you will have quite a
hard work. Other sciences show some important "dificultes" and mistakes in
your theory and in "organism" on page 191.
So here I see a problem. Your accusations that other sciences (or other
beleivings) are wrong by my opinion don't hold. There must be a middle way.

PCT "organism" on page 191 is quite some steps from being properly defined
in the light of other sciences. So the question isn't only who's right, but
also can we make a proper whole model of living organism that will really
work in "real circumstances" ?

You've made quite a progress on definition of control loop itself (I assume
it's reaction system in Ashby's theory) with reference signal coming from
anywhere or who knows what is forming it. But we know that the whole
"organism" is constituted of partial control systems which are somehow
connected and represent the living wholeness, which in PCT by my opinion
don't function yet. Only some parts of your system or "organism" on page 191
are defined well and are functioning mathematicaly O.K., but some parts
specially genetic control unit is totally unexplored.

Approach with comparing theories could result in interesting conclussions.

So I think Bill, that you think that your model is perfect and you have to
teach others. You showed me that you can be a good teacher. You taught me a
lot. But lately I've got the impression that you are working just for your
fame, like a real control unit.

And I have shown you at least five times that PCT model have "holes" and
have to be improved and you acknowledged it, because your answers were, that
you've changed your mind about your theory. But you never admitted that
you've learned also something from me or from the others on CSG net. It's
obvious that you and Rick changed your view on some subjects of PCT, since
I'm on CSG net.

I still think that some parts of PCT model needs serious modifications. We
are learning from you and you are (admitting or not) learning from us. I
still think that good teacher is only the teacher whose students are finaly
better than teacher.

I remember when you invited me to the PCT forum with words, that I can join
PCT family. Well that was quite long ago. We never were real family with
trust and everything what comes with family and we'll never be. We are just
a group of individuals following control rules. And I think that's the main
problem why we don't want to share important informations with you or with
Rick as Gavin pointed out : I use a mix of both methods of information
gathering and use, in many business situations, but I don�t like sharing
them here, because of the negative responses.

Best,

Boris

[BUSINESS_GAVIN_HERMA]

(gavin Ritz 2010.05.21.10.04NZT)

[From Rick Marken
(2010.05.20.1100)]

Gavin Ritz (2010.05.19.23.41)

I really don’t know what Rick’s reference signals are for his

controlled variables

It should be pretty obvious from my posts.
One variable I am

controlling for (with rather high gain) is
demonstrating how to study

purposeful behavior and my reference
setting for it is "by testing for

controlled variables". I’m also
controlling for playing Bach like

Glenn Gould. I’m apparently not doing
particularly well at controlling

for either of these. So I think I’ll go
control a variable that I can

control successfully: the toastiness of a
couple pieces of rye bread.

At least you have lost
your humor, Rick.

Regards

Gavin

[BUSINESS_GAVIN_HERMA]

(Gavin Ritz 2010.05.21.1000NZT)

[From Bruce
Gregory (2010.05.20.0903 EDT)]

(Gavin Ritz
2010.05.19.23.41)

Your
comments above are just a further spiral into negatively which I’m not
keen in descending. He does understand PCT really well and he needs to be a
leader one with wit, intelligence, empathy, caring, support, accountability and
all the effective stuff leaders need in their relationships with other
individuals. If he can’t well that’s his baby and just another mill
stone for PCT.

For me
personally I have only two interests in understanding the nature of humanity
and the nature of nature (don’t think I ever will), PCT falls into
somewhat both camps. So I’ll endure and see what happens in this space.

BG: I commend you for
staying above the fray. My comments were not intended to be negative, however.

Fair enough.

Quite the contrary. On my
website you would find:

The author has been
described (admittedly by himself) as, “a humble servant of the truth,
surrounded by the acolytes of ignorance.”

O well, I wonder what this controlled
variable with its associated reference signal is. I have some ideas.
Normally when I feel like diminishing people ( in most cases I do not, it’s
a fight against the powerful chemical forces in my neurons) I don’t feel
so good about myself with a particular controlled variable (quite few). Maybe
that could a potential candidate.

Regards

Gavin

Bruce

[MartinT]

Martin Taylor 2010.05.20.14:30]

[From Bill Powers (2010.05.19.1725 MDT)]

Martin Taylor 2010.05.19.14.45]

RM:This is also true
if r is not
constant, in which case the equation has to explicitly include r as a
variable:

qo = r - 1/g()d

MMT : Well, yes, it is a function of d, but it’s a function of r
also, which would make perceptual control distinguishable from S-R in
an
experiment in which r was variable in some measurable way.

BP: Only if r were sychronized with d. Too much is being said here
without reproducing the equations and showing the derivations.

I think you mean “unsynchronized”. If r were synchronized with d (a
near impossibility) you couldn’t tell. But I think you are right about
showing the equations. As I said, I had originally done so in my first
draft of my first message on the topic, but deleted them as distracting
from the point at issue, which was the apparent but unreal conflict
between Rick and Bruce as to whether experimental data could
distinguish between S-R (or IV-DV) and perceptual control when p = o+d
and r is constant. Bruce asked how you could tell between them if the
reference remained constant, and Rick acknowledged that the math was
indistinguishable but that the difference could be told because the
concepts between the two made the meanings of the variables different
(or that’s how I interpreted his strange message). Somehow the sequence
of messages looked like an unnecessary escalating conflict, since they
both said the same thing technically, but used different language to
say it.
So, to the equations. We start at the point where the influence of the
control system’s output to the environment joins the influence of the
disturbance to form the input quantity, and for simplicity we assume
that “join” means “add to”. I’ll use uppercase to symbolize functions,
lowercase to symbolize time-varying values. I will try to remember to
use == to symbolize “means” or “is defined as”.
Symbols:
qi == input quantity
qo == output quantity
e == error value
r == reference value
p == perceptual value
P( ) == perceptual input function
G( ) == output function
E( ) == Environmental feedback path function

Equations
qi = E(qo) + d
qo = G(e)
e = r - p
p = P(qi)
which gives the loop equation
qo = G(r - P(E(qo) + d))
in which the “loop gain function” is G(P(E( )))
If r is constant, we can ignore it in the functional form, since we can
write G’(x) == G(r - x)
so, for r constant
qo = G’(P(E(qo) + d))
In this formulation, G’( ) and P( ) are internal to the organism while
E is the environmental feedback function. We can write the sequence of
functions G’(P( )) as I( ), and the loop gain function G’(P(E( ))) as
L( ). (We won’t use L( ) in what follows, but noting its nature could
be helpful.)
qo = I(E(qo) + d))
If I( ) is invertible
I^-1(qo) = E(qo) + d,
which is straightforward, since I^-1(qo) (backtracking through the
internal part of the loop) is just qi. We didn’t need to go through all
the rigmarole to come back to qi = E(qo) + d, but in the above form
there are only two variables, qo and d, both of which are observable by
an experimenter.
This last equation can be rewritten
d = I^-1(qo) - E(qo),
making d a function only of qo (or, if that function is invertible, qo
is a function of d, which we can write qo = D(d)). Well, we know that
this is obviously true if r is constant, since d is the only other
input into the loop, but having the equations to show that it is true
can be psychologically more satisfying than just pointing it out.
What an experimenter can see is d (“stimulus”) and qo (“response”). The
question is whether the form of the last equation can be distinguished
from the S-R formulation, qo = F(d). And since we have just shown that
when the reference is constant, qo is a function of d, the answer is
that they cannot be distinguished at that level.
In earlier messages I claimed that it would be possible to tell the
difference between S-R and perceptual control if E( ) were nonlinear.
This seems not to be the case, so writing the equations has helped to
correct one error.
Now the question becomes one of what the function qo = D(d) might tell
the experimenter.
Rick has emphasised the behavioural illusion. Using the notation above,
the behavioural illusion is due to the fact that both I( ) and E( )
come into the relationship between d and qo.
Rick put this very well in [From Rick Marken (2010.05.19.1700)]:"
*A tangible example of this “illusion”, provided to me by bike
enthusiast Gary Cziko, is given in my “Revolution” paper. The IV
(disturbance, d) is the appearance or non-appearance of an obstruction
while driving your bike; the DV (output, qo) is the pressure exerted on
the hand brakes. There will be a clear effect of the disturbance on
hand brake force: With the obstruction the brake force is much higher
than with no obstruction. qo (brake force) will be observe to be
related to the amount of obstruction (d).
Now let’s do the same experiment on a wet street. There will still
be a relationship between qo and d but the biker will appear to be much
more responsive to the obstruction, squeezing the brakes with much more
force when the obstruction occurs. The relationship between qo and d
will have changed from what it was when the experiment was done in dry
conditions.
The change in the relationship between qo and d on a wet street
looks like a change in the responsiveness of the biker; in fact the
change results from the change in the feedback connection between the
biker’s output (braking) and desired input (stopping)."*In this example, E( ) changes from one situation to the other. It
is presumed that I( ) does not, though nothing in the data assures that
this is so. The cyclist may well be more responsive and pay more
careful attention to the road when it is wet and slippery than when it
is dry and clear. All one can tell from the data is that the relation
qo = D(d) has changed, and by the analysis above we do not know whether
the change includes change in I( ) along with the measureable change in
E( ). To determine whether I( ) has changed, the data would have to
include physical measurements of E( ) under the two conditions.

Now let’s consider another experimental situation, in which qo has no
influence on qi, so that qo = I(d). This is the typical psychophysical
experiment, in which qo is a report after a trial as to whether a tone
was in the first or the second interval of the trial. Neither a report
of “1” nor a report of “2” can influence the input quantities on the
that trial. Of course, control is involved in the act of deciding on
the choice of which to report, and on the actual execution of the
report, but that’s not the issue. The issue is usually how well the
subject’s choice matches the interval that actually contained the tone,
and since the subject’s choice cannot influence the earlier selection
of the interval, E( ) is null, and qo = I(d). In this situation, the
observations of qo and d do depend on I( ), and from them some
properties of I( ) can be inferred.

Now we have the nub of the issue. Let’s suppose that there was an
experiment in which the experimenter believed that the setup precluded
qo having any influence on qi, but was wrong, and the subject could
control (perhaps poorly). Could an experimenter who understands PCT
discover from the experimental data, without examining the experimental
setup, that the S-R (IV-DV) analysis was wrong? I think the equations
above suggest that the data do not allow it. The discrimination between
S-R and perceptual control must be made on other grounds.

And that is all that Bruce was asking about, was it not?

Martin

[rsmarken]

[From Rick Marken (2010.05.19.2300)]

Martin Taylor (2010.05.20.14:30)–

This last equation can be rewritten

d = I^-1(qo) - E(qo),

making d a function only of qo (or, if that function is invertible, qo
is a function of d, which we can write qo = D(d)).

What an experimenter can see is d (“stimulus”) and qo (“response”). The
question is whether the form of the last equation can be distinguished
from the S-R formulation, qo = F(d).

Your derivation produces results that are quite different than mine (using a linear approximation) and Bill’s (using functional notation).In the derivations with which I am familiar, d is not a function of I^-1(qo) (the organism function) in a control loop. Bill’s derivations are on pp. 145-146 of LCS I; for the closed-loop case he gets (using your notation):

qo = 1/E (r - d)

Your version of the open loop equation is the same as Bill’s:

qo = F(d)

The
question, it seems to me, is not whether the form of closed-loop equation can be distinguished
from the S-R (open loop) formulation. The equations can be distinguished quite easily. The question is whether one who observesan S-R relationship (the relationship between d and qo) will conclude that this relationship reflects F(), the organism function, or 1/E(), the inverse of the feedback function. Obviously, conventional researchers take an observed relationship between d and qo to be a reflection of F(). But if the system under study is closed loop, the the observed relationship between d and qo actually reflects 1/E(). That’s the behavioral illusion.

Rick has emphasised the behavioural illusion. Using the notation above,
the behavioural illusion is due to the fact that both I( ) and E( )
come into the relationship between d and qo.

I don’t think I( ) comes into it. The behavioral illusion comes about for the reasons I described above; it comes from thinking that the system under study is open loop when it is actually closed loop. I highly recommend that you read Bill’s “Quantitative analysis of purposive systems” paper, which is reprinted in LCS I. Note especially pp. 145-146, which gives the equations that describe the illusion, and pp.156-158, which shows a nice experimental demonstration of it.

Rick put this very well in [From Rick Marken (2010.05.19.1700)]:"

Thanks

In this example, E( ) changes from one situation to the other. It
is presumed that I( ) does not, though nothing in the data assures that
this is so.

If the biker is a closed loop system then I() is not involved. E(), along with possible variations in r, determines the observed relationship between d and qo.

The cyclist may well be more responsive and pay more
careful attention to the road when it is wet and slippery than when it
is dry and clear.

Actually, the equations say that the d - qo relationship for a closed loop system depends only on E() and r. I’ll have to check this out in simulation. It does seem like a change in control parameters should affect it too.

All one can tell from the data is that the relation
qo = D(d) has changed, and by the analysis above we do not know whether
the change includes change in I( ) along with the measureable change in
E( ). To determine whether I( ) has changed, the data would have to
include physical measurements of E( ) under the two conditions.

Since I() doesn’t occur in my derivations I’m not sure that this is correct. Maybe Bill can answer this before I do the simulations.

Now let’s consider another experimental situation, in which qo has no
influence on qi, so that qo = I(d). This is the typical psychophysical
experiment, in which qo is a report after a trial as to whether a tone
was in the first or the second interval of the trial.

A person doesn’t become open loop when they are in a psychophysical experiment. My “Power Law” paper (http://www.mindreadings.com/BehavioralIllusion.pdf) shows how the closed loop might work in a magnitude estimation experiment. And I am currently working with Bill on a demonstration of the closed-loop nature of behavior in a choice reaction time experiment.

Now we have the nub of the issue. Let’s suppose that there was an
experiment in which the experimenter believed that the setup precluded
qo having any influence on qi, but was wrong, and the subject could
control (perhaps poorly). Could an experimenter who understands PCT
discover from the experimental data, without examining the experimental
setup, that the S-R (IV-DV) analysis was wrong? I think the equations
above suggest that the data do not allow it. The discrimination between
S-R and perceptual control must be made on other grounds.

It must always be made on other grounds, those being tests to determine that a variable is under control.

Best

Rick

···

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

[Richard_Kennaway]

[From Richard Kennaway (2010.05.21.1150 BST)]

[From Bruce Gregory (2010.05.18.1402 EDT)]

[From Rick Marken (2010.05.18.1045)]

According to PCT equation (2) is wrong even if f() = -1/g(); that is, even if f() and -1/g() are mathematically equivalent. Equation 2 is wrong (as a description of behavioral organization) because f() represents a different aspect of the behaving system than -1/g(). As I said, f() represents the "mental processes" that exist inside the organism, between sensory input to behavioral output. -1/g() is the inverse of the environmental feedback function, which represents physical properties of the environment that exist outside of the organism, linking behavioral output to sensory input.

This raises an interesting point. If two theories are mathematically identical, how is it possible to make sense of a claim that one is "wrong" and the other is "right"?

That is a good question. The reason is that the theories here omit some essential details which are the reason that one works and the other does not.

As a simpler example, consider the equation dx/dt = x. One solution of this is x=0 for all t. But if this equation is a description of a real physical system, then there will always be other effects going on, perhaps very small ones, but they will be there. And if you add to that equation an extra terms to represent all the stuff that was left out: dx/dt = x + epsilon, where epsilon is some small random variable, then you will find that x=0 is not even an approximately correct solution. Instead, x will fly off to plus or minus infinity.

In general, you have to analyse such models for stability, considering not merely the models precisely as specified, but a neighbourhood around them in the space of models.

I shall have to leave application of this to the SR and PCT models as an exercise for the moment, as I'm about to leave for a conference and I'll be offline for a few days. Back on Tuesday.

···

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

[rsmarken]

[From Rick Marken (2010.05.20.0800)]

Rick Marken (2010.05.19.2300)

Martin Taylor (2010.05.20.14:30)--

The cyclist may well be more responsive and pay more careful
attention to the road when it is wet and slippery than when it is
dry and clear.

Actually, the equations say that the d - qo relationship for a closed

loop system depends only on E() and� r.� I'll have to check this
out in simulation. It does seem like a change in control
parameters should affect it too.

Well, I checked it out in simulation (after figuring it out in my head
anyway) and a change in the cyclist's responsiveness, which is a
change in the cyclists system function, does not show up in the
observed relationship between d and qo if the cyclist is a control
system. So the "behavioral illusion" equation -- qo = 1/E (r - d)-- is
exactly right: IF the system under study happens to be a closed-loop
control system, then an observed relationship between d and qo, such
as that observed in conventional psychological experiments, tells you
nothing about the the psychological characteristics of the system
under study -- represented by the function F() -- but only about
characteristics of the system's environment -- represented by the
feedback function, E().

Of course, I cannot think for you, you'll have to decide whether this
observation has God on its side...oops, is (dare I say it)
revolutionary. But I will quote from a letter to some behaviorists
regarding the "behavioral illusion" written back in 1973 by Bill
Powers and reprinted in LCS I (p. 84):

"In the section on controlled quantities in my article [the 1973
Science paper, RM] there appears an approximation, g(d) = - h(o),
which says that the cause-effect relationships that can be observed
between stimulus events and consequences of nervous system outputs --
responses -- are expressible wholly in terms of the physics of the
local environment, containing almost no information about the behaving
system at all. I see no way in which behaviorism [or cognitive
psychology, which is based on the same open-loop model, RM] can
survive a full understanding of the derivation and significance of
this harmless expression."

Best

Rick

···

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