PCT is about experience

[From Bill Powers (931206.0200 MST)]

I started a private correspondence on Martin Taylor's
930512.1230, but had some thoughts that might work better on the
net.

Mary pointed out the importance of this paragraph of Martin's:

Let's try a different tack. Early in my acquaintance with PCT,
I felt that the notion of "control system" was a metaphor for
something, but it was hard for me to see what it might be a
metaphor for. Gary, specifically, put me straight, that the
PCT "control systems" were exactly normal everyday engineering
control systems working in the real world. Now that made
everything make sense. One could bring to bear all the tools
that apply to control systems operating in the real world.
They really have to work. No fancy stuff. Just testing
against Boss Reality.

This was also my initial approach to control theory and behavior.
Everyday engineering, just checking out how living control
systems work in the real world. It took quite a few years for me
to see and begin to be able to say what is amiss in this normal
everyday engineering picture. I'm still searching for a good way
to say it, but the light began to dawn when I was working on BCP
and trying to get some definitions of the levels of control
worked out.

When you consider things like driving a car or pointing to a
target, the engineering model seems perfectly adequate. But as
you survey ALL the things that people control, you begin to
notice some anomalies: people control things that don't exist in
the environment in a way that an engineer can get hold of them.
In BCP I used the example of making lemonade. I like mine tart
and sweet, Mary likes it tart, period. What am I controlling when
I add lemon juice and sugar to make the lemonade taste right? If
you analyze the contents of the glass, you won't find any taste.
You might find correlates of what I'm controlling for, but you
won't find that single unified thing that I experience when I
take a sip. I'm controlling something that doesn't exist outside
me.

When I speak to a friend, I use a pleasant tone of voice. Where
is the pleasantness? Can you find it in the sound-waves that pass
from me to my friend? When the retirement checks come in each
month, I perceive a feeling of modest security that I gave some
effort to achieving. Where is that security -- is it floating in
the air in my house? Is it stored in my bank? When I open a door,
I create a space through which I can walk. Where is "through?"
You can search inside the doorframe forever, and you won't find
it there.

Once you begin to notice these physically nonexistent things you
perceive and control for, more and more of them show up. What,
exactly, is the "shape" of the person's face you are looking for?
What is the form of the "movement" with which you fill a spoon
with soup and convey it to your face? Where is the "sameness"
that you perceive between your checkbook balance and the bank's
balance (or the difference?). Where on the wallpaper is the
"pattern"? At what point in its trajectory, exactly, does a ball
"bounce"? Where, exactly, does the hill end and the valley begin?
What, exactly, is the relationship between where your car is in
its lane and what you actually see in the windshield? And where,
exactly, is that "relationship?"

These are not philosophical questions, but very practical ones,
because we can actually control all these kinds of experiences.
We can act so as to bring them into existence and maintain them
there; if we choose, we can alter them at will, by changing our
actions. Unless we pause to wonder, we are perfectly comfortable
in taking them as given aspects of the world that affects us and
on which we act. It's only when we try to make an engineering
model that controls those things that it beomes evident that
something is wrong. The engineer can control the measured
position of the car in its lane, but not the apparent position
that the driver experiences and controls.

The first temptation when thoughts like these crop up is to
divide the world into two parts: things that actually exist
outside of us, and things that we clearly make up through
perceptual interpretations: the objective aspects of the world,
and the subjective aspects. But that division can hold up only
for a while. The more carefully you look at the objective aspects
of the world, the clearer it becomes that there are none.

This is a concept that can't be communicated unless the listener
has independently discovered it, or takes time to search out
possible meanings for the words and is thus led to discover it.
It's all perception, the whole shebang, even the parts that seem
objective.

The point, before I stray too far off the subject, is that
control theory is really not about engineering or physics; it's
about a phenomenon of experience, the central aspect of our
relationship to the world. We act in order to affect what we
experience. We learn how to affect experience to make it come
closer to what we want it to be, and as long as that is what we
want, to keep it in that form. This is a simple and direct
observation about reality itself, which is the world of
experience, not the world of science or mathematics.

Science and mathematics are simply tools we have invented in our
attempts to describe order in the world of experience. But there
is nothing for those tools to work with unless we first notice
orderliness without their aid. PCT must first be grasped as a
real phenomenon of experience, _one's own_ experience. Reducing
it immediately to a mathematical representation and a physical
model, and then confining one's reasoning to the mathematics and
the model, is to turn away from the very phenomenon that PCT is
about. It is giving contemplation of the tools priority over
using them to explain phenomena.

PCT is not about controlling a plant in the external world. It is
about how we eat, walk, drive cars, find our way to a restaurant,
kiss our loved ones, argue with our colleagues, scratch an itch,
build our confidence, comb our hair, earn money. It's about
everything we do, every waking moment. That's the level where PCT
must first be understood, before there's any point in trying to
make a simplified idealized engineering or mathematical model of
control. It's a world view of behavior that can be checked
directly against experience, by anyone, without even drawing a
block diagram. When you grasp that world view, you then have
something to explain.

ยทยทยท

---------------------------------------------------------------
Best,

Bill P.

[From Rick Marken (931206.0830)]

Bill Powers (931206.0200 MST) --

WOW!

You might find correlates of what I'm controlling for, but you
won't find that single unified thing that I experience when I
take a sip. I'm controlling something that doesn't exist outside
me.

Beautifully put.

What, exactly, is the relationship between where your car is in
its lane and what you actually see in the windshield? And where,
exactly, is that "relationship?"

Why, it's in the information in perception, of course. But only
some people can find it (information in perception seems to be
a bit like a boojum -- those of us who can't find it seem to be
stuck with the snarks of our experience)>

Martin Taylor (931205 12:30) --

Me:

If I'm wrong, I'm happy to be corrected.

Martin:

Oh, would that it were true!

Test it out! Show me where I'm wrong. There are several possibilities.
You could correct me be providing evidence of any of the following:

1) That there IS information in the perceptual signal.

2) That when you have two or more interacting control systems without
a hierarchic relationship, you DO have to use dynamical analysis
beyond simple control.

3) That purpose arises ONLY in a negative feedback system with a
changeable reference.

Best

Rick

[Martin Taylor 931206 15:00]
(Bill Powers 931206.0200)

I'm not sure if you are trying to say that PCT control systems are NOT
physical control systems. I hope not, but some of your wording seems to
have that connotation. Take away the connotation, and there's an awful
lot to like in your posting (as usual). I'd like to build on it, if you
don't mind.

When you consider things like driving a car or pointing to a
target, the engineering model seems perfectly adequate. But as
you survey ALL the things that people control, you begin to
notice some anomalies: people control things that don't exist in
the environment in a way that an engineer can get hold of them.

Yes, this is a point that I have tried to make on quite a few occasions,
in dealing with the nature of a CEV. (Remember how strongly I tried to
make that point in Durango, in the context of communication?).

Unless the situation is tightly constrained, as in Rick's Mind-Reading
demo(s), no external observer can be sure of discovering ANY CEV
corresponding to another's controlled perception. Even in the M-R demo,
the computer could not determine that the subject is trying to keep
the "3" directly above the "5" (there's no action that would allow the
user to do that), though it might be able to determine that the subject
is trying to keep the "3" centred vertically and the "5" centred
horizontally.

Since all CEVs of interest to the subject are defined entirely by the
subject's PIFs, the engineer can "get hold of them" only by determining
those PIFs. This is impossible in detail. The Test determines
only that the CEV influenced by the tester is positively connected with
(I was going to say "correlated with") the CEV controlled by the testee.
It can never show that the tester's CEV precisely corresponds to a
perception controlled by the subject.

However, this fact does not, I think, justify:

...PCT "control systems" were exactly normal everyday engineering
control systems working in the real world. Now that made
everything make sense. One could bring to bear all the tools
that apply to control systems operating in the real world.
They really have to work. No fancy stuff. Just testing
against Boss Reality.

This was also my initial approach to control theory and behavior.
Everyday engineering, just checking out how living control
systems work in the real world. It took quite a few years for me
to see and begin to be able to say what is amiss in this normal
everyday engineering picture.

I don't see the connection between, on the one hand, the agreed
position that the world of the CEVs is a mirror world defined by
the interconnection of PIFs, and, on the other, that normal science
and engineering methods do not apply. In other words, I don't see
what is amiss.

What am I controlling when
I add lemon juice and sugar to make the lemonade taste right? If
you analyze the contents of the glass, you won't find any taste.
You might find correlates of what I'm controlling for, but you
won't find that single unified thing that I experience when I
take a sip. I'm controlling something that doesn't exist outside
me.

This is true of every single instance of control, not only in humans,
but also in machine control.

... Where on the wallpaper is the
"pattern"? At what point in its trajectory, exactly, does a ball
"bounce"? Where, exactly, does the hill end and the valley begin?
What, exactly, is the relationship between where your car is in
its lane and what you actually see in the windshield? And where,
exactly, is that "relationship?"

These are not philosophical questions, but very practical ones,
because we can actually control all these kinds of experiences.

Yes, indeed. How better can one describe the mirror world?

It's only when we try to make an engineering
model that controls those things that it beomes evident that
something is wrong. The engineer can control the measured
position of the car in its lane, but not the apparent position
that the driver experiences and controls.

This is where you begin to lose me. Isn't the "measured position"
the very "apparent position" for the mechanical system that acts on the
car to affect its perceived position in the lane? Of course it isn't
the "apparent position" that the driver experiences. Are you saying any
more than that one person's CEV is not available to another except by
an outrageous fluke, and that in any case neither would know if the fluke
had occurred?

The first temptation when thoughts like these crop up is to
divide the world into two parts: things that actually exist
outside of us, and things that we clearly make up through
perceptual interpretations: the objective aspects of the world,
and the subjective aspects.

Doesn't tempt me. I don't think it has ever tempted me since I first
appreciated what PCT was about. Might have done, once upon a time, long
ago in fairy-tale days. Not now.

The point, before I stray too far off the subject, is that
control theory is really not about engineering or physics; it's
about a phenomenon of experience, the central aspect of our
relationship to the world.

Isn't that precisely what engineering and physics and the rest of
science is about? I wouldn't say that control theory is ABOUT engineering,
but it is an aspect of engineering, and of physics. It deals with the
world as the world is perceived to be in the mind of the theorist. No
science can do otherwise.

Science and mathematics are simply tools we have invented in our
attempts to describe order in the world of experience. But there
is nothing for those tools to work with unless we first notice
orderliness without their aid.

Sure. We build on what we _and others_ have observed to be the regularities
of nature. I don't know when this started, but it must be at least as
old as writing. Nobody creates the entire edifice of science anew from
their own individual observation. Some people tear down and rebuild parts
of the edifice based on what they have observed, but they don't start
from scratch.

PCT must first be grasped as a
real phenomenon of experience, _one's own_ experience. Reducing
it immediately to a mathematical representation and a physical
model, and then confining one's reasoning to the mathematics and
the model, is to turn away from the very phenomenon that PCT is
about. It is giving contemplation of the tools priority over
using them to explain phenomena.

"Priority?" There's nothing to stop one thinking about how to make a
more effective hammer while one is driving in nails. In fact, it's
the best time for such thoughts. Leaving the analogy, the best way
to see how the mathematics and fundamental physics work is to think
of them _while_ dealing with one's own perceptual control experience.

PCT is not about controlling a plant in the external world. It is
about how we eat, walk, drive cars, find our way to a restaurant,
kiss our loved ones, argue with our colleagues, scratch an itch,
build our confidence, comb our hair, earn money. It's about
everything we do, every waking moment.

Yes. It's hard to get people to believe that this is the claim, let
alone that it must be so. But it is so, provided you don't conflate
"PCT" with a specific design of the control hierarchy.

When you grasp that world view, you then have something to explain.

Which you started to do some 40 years ago, and have continued to do,
with the help or hindrance of others with whom you communicate so ably.
And whatever explanations there may be, they must always be consistent
with whatever else we think we "know," such as the laws of thermodynamics
and the rest of physics. We are not explaining magical systems. We
are explaining everyday phenomena, phenomena of that amusing complex, life.

Thanks for that posting.

Martin

[Martin Taylor 931206 17:55]
(Rick Marken 931206.0830)

You knew I wouldn't resist, didn't you, you old S-R conniver, you!

What, exactly, is the relationship between where your car is in
its lane and what you actually see in the windshield? And where,
exactly, is that "relationship?"

Why, it's in the information in perception, of course. But only
some people can find it (information in perception seems to be
a bit like a boojum -- those of us who can't find it seem to be
stuck with the snarks of our experience)>

Are you really and truly offering to vanish if you ever get information
from your perception?

Test it out! Show me where I'm wrong. There are several possibilities.
You could correct me be providing evidence of any of the following:

1) That there IS information in the perceptual signal.

2) That when you have two or more interacting control systems without
a hierarchic relationship, you DO have to use dynamical analysis
beyond simple control.

3) That purpose arises ONLY in a negative feedback system with a
changeable reference.

(1) I did, but you refused to accept the direct demonstration. I am at
present preparing what I hope will be a demonstration that even you
will accept.

But PLEASE, read enough of my stuff to understand that there never is
information IN a signal. The information measure is a measure of
uncertainty reduction, and is ordinarily information available FROM
a signal ABOUT a situation. The concept of information IN a signal
is meaningless.

(2) As well you know, I retracted that one immediately, and noted my
stupidity in ever having said it, as soon as Tom mentioned his work.

(3) You are going further out on a limb here than you have before.
What are you proposing as an alternative to control systems as
purposive? That vortices in a stream have purpose? or that there
are other purposive systems than negative feedback systems? What
do you mean by "purpose"? In the past, you have argued that the
purpose of a control system is in its reference signal. That's
what I think is the case, but I can't prove it to be true.

I can't tell what I am supposed to show. If you want to argue that there
are structures other than control systems that have purpose, I guess it is
up to you to show it. For me it is partly a matter of faith, and
partly a matter of verbal definition, as to whether purpose implies
control. I believe in PCT, but I can't prove I'm right.

Sorry. I have to decline number 3 unless you want to define purpose in a way
that allows your sentence to be tested for correctness.

Martin