how to do the Test

CSGnet Archive CSG1997
1

[From Stefan Balke (970528.1130 CET]]

Rick Marken (970527.0920 PDT)

Bill Powers (970527.0619 MDT) replying to Stefan Balke (970527.1100
CET):

When you say your statement is a disturbance, you aren't specifying >>

what perception it is supposedly disturbing. To do the Test, you

have to define a variable that you think is under control.

I would say that this is the most common misconception people have >about

doing the Test. Most people seem to get the idea that The Test

involves the application od "a disturbance". What is often forgotten or
ignored is the fact that the notion of a disturbance makes no
sense unless one has, in mind, some variable whose value _would_ be
changed by the disturbance _if it were not under control_.

The most essential (and difficult) aspect of the Test is _guessing_ >what

variable(s) an organism _might_ be controlling. Once you have

guessed what variable _might_ be under control, it is easy to think of
things that would change the value of this variable if it were _not_
under control.

Thanks, Bill and Rick,

now I've learned to make a more complete analysis at first (but I'm still
puzzling how to do that :-)). I remember the example of the car and the wind
(disturbance), blowing the car out of the lane (which surely would happen
without an attempt to steer the car in the center of the lane (reference
state) and the movement of the steering wheel (action)). Here we have
everything well defined and most of the variables are even measurable. But
in the social context it's more difficult to say A (which is an action of
one person) would have moved B (which is an action of person b) into this
direction, but it didn't, therefore we have a hint that a control process
could be involved.

One problem could be that we are confronted with a hierachie of lot's of
control systems, and the absense of any control is in contrast to the windy
road example not to be expected. So there always will be the possibility of
alternative explanations.

Best, Stefan

2

[From Bill Powers (970528.0648 MDT)]

Stefan Balke (970528.1130 CET]--

Thanks, Bill and Rick,

now I've learned to make a more complete analysis at first (but I'm >still

puzzling how to do that :-)).

The best way to start is to look at the only controlled perceptions you can
see directly -- your own. Suppose someone were to say to you, "Oh, by the
way, Stefan, the department's budget has been cut by 50%, so we're going to
have to eliminate some of our faculty and graduate students -- and you're
on the list." What variables that you're controlling might be disturbed by
this statement, if any? Would there be any changes in your behavior?

But in the social context it's more difficult to say A (which is an
action of one person) would have moved B (which is an action of >person b)

into this direction, but it didn't, therefore we have a >hint that a
control process could be involved.

Not the action of person B, the _perceptions_ of person B. The action of
person A, in telling you that you might have to leave school, affects your
perceptions of where you are relative to where you want to be. If you did
nothing, certain things would happen to affect your perceived position in
life. I assume that you would then take some sort of action to prevent as
many of these effects from actually occurring as you could. Of course what
you imagine that you would do might not be exactly what you would do in
practice, but by imagining this problem you can see how disturbances affect
your perceptions, and how your actions would be directed toward restoring
your perceptions as far as possible toward their undisturbed state.

One problem could be that we are confronted with a hierachie of >lots of

control systems, and the absence of any control is in >contrast to the
windy road example not to be expected. So there >always will be the
possibility of alternative explanations.

If you can think of two explanations, you can think of disturbances that
would affect one of the possible controlled variables but not the other.
This is how you eliminate hypotheses. The object of the game is to think up
as many reasonable alternatives as you can, and experimentally eliminate as
many of the explanations as possible (one hopes, of course, that you don't
eliminate ALL of them!).

An old piece of advice: "don't borrow trouble." That is, don't start
imagining all the difficulties that could arise, because that is a good way
to keep from ever trying anything. Just guess at a perception that someone
might be controlling, think up a way of applying a small disturbance to it,
and see if the person does something to oppose the effect of the
disturbance. If not, you've eliminated one hypothesis, and that is useful.

Consider this:

Stefan, you're not really trying to understand PCT.

Does that disturb any perception of yours, and do you feel the need to do
something that would correct the error (or would you, if you thought I was
serious)?

Best,

Bill P.

3

[From Stefan Balke (970530.1030 CET)]

Bill Powers (970528.0648 MDT)--

The best way to start is to look at the only controlled perceptions you can
see directly -- your own. Suppose someone were to say to you, "Oh, by the
way, Stefan, the department's budget has been cut by 50%, so we're going to
have to eliminate some of our faculty and graduate students -- and you're
on the list." What variables that you're controlling might be disturbed by
this statement, if any? Would there be any changes in your behavior?

I actually got that message and I'm still trying to solve the resulting
income problem :-). (I try to receive a grant from the ministry of education
to introduce a german version of Ed Fords program. If this actually becomes
true I'll be able to work with PCT in a practical field, which is a great
chance for me.) Indeed this message changed a lot of my actions, but it
didn't change my ideas about scientific and practical work.

But in the social context it's more difficult to say A (which is an
action of one person) would have moved B (which is an action of
person b) into this direction, but it didn't, therefore we have a
hint that a control process could be involved.

Not the action of person B, the _perceptions_ of person B.

Agreed, but I spoke about the actions of person B, because I, as the
observer, can only perceive B's actions but not B's perceptions, therefore
the only influence I can notice is a possible influence to B's actions.

An old piece of advice: "don't borrow trouble." That is, don't start
imagining all the difficulties that could arise, because that is a good way
to keep from ever trying anything. Just guess at a perception that someone
might be controlling, think up a way of applying a small disturbance to it,
and see if the person does something to oppose the effect of the
disturbance. If not, you've eliminated one hypothesis, and that is useful.

Consider this:

Stefan, you're not really trying to understand PCT.

Why do you think so? - Did you notice any 'wrong' behavior?
Bill, this is really a great disturbance, didn't you just talk about not
doing this :slight_smile: To be more serious, this example was a very powerful lesson
in self-experienced PCT. I thought about this disturbance for two days and I
watched out for other natural disturbances and my reactions.

Best, Stefan

4

[From Bill Powers (970530.0857 MDT)]

Stefan Balke (970530.1030 CET)--

I actually got that message and I'm still trying to solve the >resulting

income problem :-). ...

Indeed this message changed a lot of my actions, but it
didn't change my ideas about scientific and practical work.

OK, so what were the disturbances, the controlled variables (and their
reference levels), and the actions involved? I can guess at some controlled
variables and their reference levels:

  Perception Reference

Perceived income --- desired income
Perceived scientific work --- desired scientific work
Perceived practical work ---- desired practical work

Of course only you know the details.

The message carried news of something that would affect all three of these
perceptions, as well as others (I would guess). You then took actions,
again guessing, that would tend to counteract the effects of this
disturbance. Can you organize these things in the PCT pattern to show how
the parts of the PCT diagram fit what is going on?

But in the social context it's more difficult to say A (which is >>>an

action of one person) would have moved B (which is an action of

person b) into this direction, but it didn't, therefore we have a
hint that a control process could be involved.

Not the action of person B, the _perceptions_ of person B.

Agreed, but I spoke about the actions of person B, because I, as the
observer, can only perceive B's actions but not B's perceptions,
therefore the only influence I can notice is a possible influence to >B's

actions.

Yes, but the trick in doing the Test is to make an educated guess as to the
nature of the controlled variable. A's action does not affect B's action
directly. A's action affects some controlled variable, and B's action
affects it in the opposite direction. This creates the illusion that A's
action is somehow causing B's action , but B is really acting to oppose the
change in B's controlled variable, no matter what caused it to change. Once
you know what the controlled variable is, you can predict B's action for
ANY disturbance that can affect the same controlled variable. At least, if
you can't predict the exact action that is taken, you can say that it will
be one that counteracts the effect of the disturbance.

Consider this:

Stefan, you're not really trying to understand PCT.

Why do you think so? - Did you notice any 'wrong' behavior?
Bill, this is really a great disturbance, didn't you just talk about >not

doing this :slight_smile: To be more serious, this example was a very >powerful lesson
in self-experienced PCT. I thought about this >disturbance for two days and
I watched out for other natural >disturbances and my reactions.

Great! I took a bit of a risk, but I figured you would know that this was
a thought-experiment and not a real disturbance. When you look back on your
reactions, can you see that they were really _actions_ intended to restore
whatever variables I disturbed to their reference states?

Best,

Bill P.

5

[From Stefan Balke (970604.1000 CET)]

Bill Powers (970530.0857 MDT)--

The message carried news of something that would affect all three of >these

perceptions, as well as others (I would guess). You then took >actions,
again guessing, that would tend to counteract the effects of >this
disturbance. Can you organize these things in the PCT pattern to >show how
the parts of the PCT diagram fit what is going on?

Yes, if I reflect about my thoughts, feelings and actions concerning the job
problem I can easyly see an underlying HPCT pattern. Most of the work is
going on in the imagination mode and even leads to 'sleepless nights'.

This creates the illusion that A's action is somehow causing B's action,
but B is really acting to oppose the change in B's controlled variable,
_no matter what caused it to change_.

That's great! Until now, I didn't really notice this _essential_ idea.
Thanks for opening my eyes! Of course I wonder why I didn't really look at
this idea which seems so obvious since I have looked at it now.

Great! I took a bit of a risk, but I figured you would know that this >was

a thought-experiment and not a real disturbance. When you look back >on your
reactions, can you see that they were really _actions_ intended >to restore
whatever variables I disturbed to their reference states?

Both, thoughts and real actions. One of them was to write an email to CSGnet.

Best, Stefan

6

[From Bill Powers (970604.0640 MDT)]

Stefan Balke (970604.1000 CET)--

This creates the illusion that A's action is somehow causing B's
action, but B is really acting to oppose the change in B's >>controlled

variable,_no matter what caused it to change_.

That's great! Until now, I didn't really notice this _essential_ >idea.

Thanks for opening my eyes! Of course I wonder why I didn't >really look at
this idea which seems so obvious since I have looked >at it now.

That's the story of PCT, isn't it? Once you open your eyes and look at it,
you wonder why you didn't see how obvious it is. Believe me, while I was
struggling to apply control theory to behavior none of these ideas was
obvious to me, either. Sometimes I felt very stupid when I finally figured
out how something worked -- what took me so long to see it?

The answer, I think, is in what we believed as an explanation of the same
relationships _before_ we got the great insight. Many people can't accept
PCT because they simply can't believe that the explanation they already
have is incorrect or inadequate. They hold on so hard to what they think
they already know that they can't relax long enough to see the simple sense
that PCT makes. Or else they have never realized that the explanation they
believe isn't really an explanation, so they think that PCT is just another
"perspective."

I think that people would be interested in your desciption of how you
explained these phenomena before you saw how the PCT explanation works.
It's still fresh in your mind -- you can do a better job of that than I
could, because all this happened to me so long ago.

Best,

Bill P.

P.S. to everyone contemplating coming to the CSG meeting August 6th: It is
now June 4. The deadline for registrations ($20 fee) at the regular rates
is June 15th. After that date, the total fee is $30 more. The balance has
to be paid by July 30, because we have to put down a deposit. Also, Mary,
as usual, is beginning to despair -- should we cancel the meeting at Fort
Lewis College and just have the few people who are going to attend meet at
our house? I don't care if we have 6 or 60 people at the meeting, but it
does make a difference in the planning. Can't you-all put off
procrastinating until tomorrow?

7

[From Stefan Balke (970609.1130 CET)]

Bill Powers (970604.0640 MDT) to Stefan Balke (970604.1000 CET)--

Bill:

This creates the illusion that A's action is somehow causing B's
action, but B is really acting to oppose the change in B's controlled
variable,_no matter what caused it to change_.

Me:

That's great! Until now, I didn't really notice this _essential_ idea.
Thanks for opening my eyes! Of course I wonder why I didn't really look at
this idea which seems so obvious since I have looked at it now.

Bill:

I think that people would be interested in your desciption of how you
explained these phenomena before you saw how the PCT explanation works. It's
still fresh in your mind -- you can do a better job of that than I could,
because all this happened to me so long ago.

Bill, I was a few days offline and will answer now:

In fact I used and still use somewhat vague explanations because most of
them are ex post explanations. From time to time it happens that I'm
convinced by a new idea because it gives an explanation that fits with my
everyday experiences. So I was convinced about the ideas of Fritz Heider
that we analyse the source of the disturbance to find out what will happen
next. In the same way I'm convinced about the fact that living systems are
control systems. Now I have in mind a block diagram with only three elements
inside a person: the reference level, the perceptual signal and the output
function - that's enough for the control system to work! The rest, the inner
nature of the disturbance is unknown to the special control system and
sometimes even to the controlling person as a whole and irrelevant for the
sake of control. Paul Feyerabend once pointed out that the ancient warriors
started to fight because they thought that the god of war told them to do
so. And they really fought.

The 'aha' experience was concerning the last part of your sentence: '_no
matter what caused it to change_'. The beginning of the sentence was already
clear for me. The idea that a control system operates in the same way
independent of the source and/or context of the disturbance opens a new
perspective for me, because I still looked for the 'cause and nature of the
disturbance' to find out what will be to expect from it in the future. So my
focus was to look at the behavior/development of the disturbances. Now my
focus changes to the control system which lays inside myself.

Best, Stefan

8

[From Bill Powers (970609.0727 MDT)]

Stefan Balke (970609.1130 CET)--

From time to time it happens that I'm
convinced by a new idea because it gives an explanation that fits >with my

everyday experiences. So I was convinced about the ideas of >Fritz Heider
that we analyse the source of the disturbance to find >out what will happen
next.

I think we do this, but it's done at the higher cognitive levels and so is
too slow and inaccurate for real-time lower-level control processes.
Sometimes we CAN actually see what is causing a perturbation of a
perception, and in some subset of that subset of cases, knowing the cause
might enable us to control better. If you can see a car approaching on a
side road, you don't have to wait for a near-miss to try to save yourself
from a collision. You might even be able to judge whether the other driver
is braking sufficiently to stop before you get to the intersection, but
such judgments take time to form and they are not necessarily correct.
It's safest to keep the relationship between cars under control, which can
be done without prediction.

There are many cases, however, the great majority, in which the first you
know about a disturbance is a departure of a perception from its reference
level. The resulting error is the most immediate possible indication of a
disturbance, but since the error directly drives the output, there's no
need to "understand" what caused the disturbance. You're already acting
against it, and your cognitive systems are following along behind the
action, explaining it perhaps but not doing anything about it.

In the same way I'm convinced about the fact that living systems are
control systems. Now I have in mind a block diagram with only three
elements inside a person: the reference level, the perceptual signal >and

the output function - that's enough for the control system to >work!

But sooner or later, you're going to have to ask why that reference signal
is set as it is at that time, and then you'll have to get into hierarchical
control systems anyway.

Best,

Bill P.

9

[Hans Blom, 970609b]

(Stefan Balke (970609.1130 CET))

Bill: This creates the illusion that A's action is somehow causing
B's action, but B is really acting to oppose the change in B's
controlled variable,_no matter what caused it to change_.

Me: That's great! Until now, I didn't really notice this
_essential_ idea. Thanks for opening my eyes! Of course I wonder
why I didn't really look at this idea which seems so obvious since
I have looked at it now.

So I was convinced about the ideas of Fritz Heider that we analyse
the source of the disturbance to find out what will happen next. In
the same way I'm convinced about the fact that living systems are
control systems. Now I have in mind a block diagram with only three
elements inside a person: the reference level, the perceptual signal
and the output function - that's enough for the control system to
work!

You miss an aspect, maybe. Generally, there are a large number of
ways in which a disturbance can be eliminated. Say that somebody
keeps pushing you, whereas it is your goal to remain standing and in
balance. What can you do? You can tense your muscles in order to
provide a counter-momentum. Or you can walk away. Or you can hit the
pusher and hope he'll go away. You can think of a lot of possible
additional solutions (control strategies).

What often seems necessary is, as Heider says, to find the source (I
would say "cause") of the disturbance, and not so much to find out
what _will_ happen next but what _would_ happen next if we ourselves
performed certain (corrective) actions. What is the cause of me
loosing my balance? There are lots of "causes" that I could select: A
physical force (that it is delivered by a person is incidental),
which I must counteract with an equal but opposite force. A clumsy
movement of a person much stronger than me. Willful malice of an
enemy. In each case, I "select" (however unconsciously and, maybe,
erroneously) some "cause" which will guide my subsequent behavior.
Push back? Go away? Punch in the nose? Kill? In the complexities of
life, there is usually no single unique path from perceptual input
function to output function. The low-level perceptions may result in
very different high-level perceptions (amongst which are "causes")
and thus also to the employment of very different output functions
(action patterns).

We humans employ the -- extremely useful -- concept of "cause", I
think, not in the first place to determine how something inevitably
leads to something else, but to understand how we can break this
chain of inevitabilities by our own personal actions. In short, I
believe that our attribution of what constitutes a cause has a very
important function in us living control systems: it allows us to
react not only at the most proximal level but at higher levels as
well. If the pusher keeps pushing, use a more effective and/or longer
lasting method of disturbance elimination than mere pushing back.

Paul Feyerabend once pointed out that the ancient warriors started
to fight because they thought that the god of war told them to do
so. And they really fought.

An attribution that we do not accept anymore. Yet, many people even
now accept attributions like "willful malice" where "clumsy behavior"
would be more correctly indicated. It's all perception -- or should I
say imagination or attribution?

The idea that a control system operates in the same way independent
of the source and/or context of the disturbance opens a new
perspective for me, because I still looked for the 'cause and nature
of the disturbance' to find out what will be to expect from it in
the future.

The "context" of a disturbance would certainly contribute to behavior
if the disturbance is attributed to be somehow "caused". If such a
cause can be discovered (or invented), elimination of the disturbance
may come to mean, most regretfully, not elimination of somebody's
action having an undesired effect but ultimately the elimination of
that somebody himself.

Some AIer once said that a computer would only be truly intelligent
if it had a machine gun which it could effectively employ to prevent
someone from pulling the plug (from the power socket). If we equate
intelligence with being in control, that's a reasonable angle.

So my focus was to look at the behavior/development of the
disturbances. Now my focus changes to the control system which
lays inside myself.

Here you speak of having a choice, at least of the ability to see
things in a different light. A different perceptual input function? A
different output function? Or a different idea of cause and effect?

Greetings,

Hans

10

[From Stefan Balke (960613 CET)]

Me:

Now I have in mind a block diagram with only three elements >>inside a

person: the reference level, the perceptual signal >>and the output function
- that's enough for the control system >>to work!

Bill:

But sooner or later, you're going to have to ask why that >reference signal

is set as it is at that time, and then you'll >have to get into hierarchical
control systems anyway.

Yes, of course. By the way, I remember that you once posted a text about the
genesis of new control systems. I looked in my archive but couldn't find it.
Do you remember which text I mean? Could you post it again? I'd like to
reread it. Thanks in advance and best regards,

Stefan

11

[From Bill Powers (970613.0750 MDT)]

From Stefan Balke (960613 CET) --
By the way, I remember that you once posted a text about the
genesis of new control systems. I looked in my archive but couldn't >find

it. Do you remember which text I mean? Could you post it again? >I'd like
to reread it.

I don't have my archive indexed any more. The indexing program developed a
problem, and to replace it I would now have to spend about $200. I've just
sent off about 60 megabytes (compressed) of archives and other materials to
Dag Forssell, and he is going to make a CD ROM disk containing an indexer
(boolean search). That will take some time, of course.

Briefly, I reason that new control systems probably develop in this order:

1. Perceptual function, to create organized perceptions.

2. Memory, accumulating memories of states of the new perception that have
actually occurred.

3. Extraction of reference signals from past values of the perception: here
is an experience I would like to have again (or avoid having again).

4. Development of the comparison: detecting the difference between the
reference signal and the perceptual signal.

5. Development of the output function: converting the error into an action
that has a negative feedback effect on the error.

6. Tuning the system for accuracy and stability.

Best,

Bill P.