Conflict; degrees of freedom

[From Rick Marken (2003.11.30.1200)]

Martin Taylor (2003.11.30 1249) --

The point where I differed with Rick, and where Bill agrees with me,
is that the "variable that must be in two different states at once"
need not be a perception, but "could be an external variable, a
perception, or a reference signal."

Well, then we no longer differ. I like this definition because it
means that my "Cost of conflict" demo
(Cost of Conflict) is a true
demonstration of conflict and I don't have to get rid of or change it.

In that demo, it is an external variable (the effect of the x,y
position of the mouse on the x,y position of the cursor) that must --
and _can't_ -- be in two different states a once. In order to control
the x,y position of the cursor, the subject must be able to set
references for the x,y position of the mouse. This can be done without
conflict since there are 2 separate perceptual degrees of freedom (x, y
position of the mouse) for 2 separate reference degrees of freedom
(where the mouse should be in x,y space). But the computer program
turns the two degrees of freedom of mouse position into a single degree
of freedom of effect on the cursor. That single degree of freedom of
variation in this "external variable" is all that is available to
control the cursor in two dimensions.

The conflict in the demo is created in one of the ways you (Martin)
suggested in an earlier post -- by changing output functions. There is
only one output degree of freedom available to control a variable
(cursor position) in two dimensions. The net result of this reduction
of degrees of freedom in an external variable is to reduce the degrees
of freedom of the perceptual variables.

Although the perceptual variable varies in x , y space it can really
only vary in a one dimensional subset of this space -- on the diagonal.
  So I think that a general definition of conflict can be given in terms
of degrees of freedom (number of dimensions in which variables can vary
independently of each other) of reference and perceptual variables:

Conflict exists in a set of N control systems when the number of
reference degrees of freedom is greater than the number of perceptual
degrees of freedom in the set.

I think this definition holds regardless of where the limitation in
perceptual degrees of freedom comes from: the perception itself,
external variables (as in the demo) or reference signals (as when two
higher level systems try to control the same perception).

What do you think?

Best

Rick

···

----
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

[Martin Taylor 2003.11.30.1655 EST]

[From Rick Marken (2003.11.30.1200)]

Martin Taylor (2003.11.30 1249) --

The point where I differed with Rick, and where Bill agrees with me,
is that the "variable that must be in two different states at once"
need not be a perception, but "could be an external variable, a
perception, or a reference signal."

Well, then we no longer differ.

Great! We usually don't when we deal face-to-face. It's nice that we
can come to agreement over e-mail, too.

So I think that a general definition of conflict can be given in terms
of degrees of freedom (number of dimensions in which variables can vary
independently of each other) of reference and perceptual variables:

That's good, but I'd make one small change. The number of perceptual
degrees of freedom is necessarily the same as the number of reference
degrees of freedom (one per Elementary Control Unit). So I'd reword
it thus:

Conflict exists in a set of N control systems when the number of
reference degrees of freedom is greater than the number of degrees of freedom
in the set available for perceptual control.

This seems to cover the case when two perceptions are the same
function of environmental variables, as well as the case when there
are limits on the environmental degrees of freedom available for
action.

I think this definition holds regardless of where the limitation in
perceptual degrees of freedom comes from: the perception itself,
external variables (as in the demo) or reference signals (as when two
higher level systems try to control the same perception).

What do you think?

I think it is right, with my slight amendment. At first I thought it
did not cover the case of Bill Williams's two-commodity conflict, but
it does.

When the available budget $T exceeds the dollars, $A + $B, available
to bring the two commodities A and B to their reference levels, the
system does have two degrees of freedom for control $A and $B both
can vary above and below their respective "ideal" values (i.e. when
both quantities are at their reference levels). But when $A(ideal) +
$B(ideal) > $T, then the best the two controllers can do lies
somewhere on the line $A(actual) + $B(actual) = $T, thus reducing the
degrees of freedom to one.

I can't offhand think of a case your definition does not cover.

I didn't make the statement as explicit as you have done, but I
discussed its implications for reorganization on the pages starting
at <http://www.mmtaylor.net/PCT/Mutuality/domains.1.html&gt;

Thanks.

Martin

from [Marc Abrams (2003.11.30.1731)]

[Martin Taylor 2003.11.30.1655 EST]

That's good, but I'd make one small change. The number of perceptual
degrees of freedom is necessarily the same as the number of reference
degrees of freedom (one per Elementary Control Unit). So I'd reword
it thus:

Conflict exists in a set of N control systems when the number of
reference degrees of freedom is greater than the number of degrees of

freedom

in the set available for perceptual control.

This seems to cover the case when two perceptions are the same
function of environmental variables, as well as the case when there
are limits on the environmental degrees of freedom available for
action.

How can we get this into, and part of, the current PCT lit?

Does anyone think it might be worthwhile to have an 'update' section on the
web page for these kinds of occasions?

Marc

[From Bill Powers (2003.11609 MST)]

Martin Taylor 2003.11.30.1655 EST--

I'd reword it thus:

Conflict exists in a set of N control systems when the number of
reference degrees of freedom is greater than the number of degrees of freedom
in the set available for perceptual control.

Bingo! I think this not only defines conflict (of all degrees) but explains
it as well. Conflict is a state of impossibility: there is no solution of
the implied system of equations. Your definition opens the way to
recognizing many other "impossible" situations beside a simple opposition
of two forces. I suppose if we made these equations differential equations,
we could include oscillatory results, too -- limit cycles and all the rest
of the dynamic effects.

Well, that was fun. What shall we solve next? I guess Martin gets to take
the trophy home this year.

Best,

Bill P.

[From Kenny Kitzke (2003.12.01)]

<Bill Powers (2003.11609 MST)>

Conflict exists in a set of N control systems when the number of
reference degrees of freedom is greater than the number of degrees of freedom
in the set available for perceptual control.

<Bingo! I think this not only defines conflict (of all degrees) but explains
it as well.>

Breakthrough! With this definition, we can finally go forth and use PCT/HPCT to explain conflict that exists in and between human beings.

The teen boy, conflicted between using or not using cocaine, will be assuaged when we share this definition with him.

The husband and wife, in conflict over whether to continue or end their marriage, grasping for understanding, will have their eyes opened and rejoice to hear the correct explanation of their conflict.

Now, while this definition of “conflict” is perfectly clear to me (ah-ha), and so much more useful in addressing situations of “conflict” in myself, or between me and other control systems, could you or Martin explain it just a little bit more, perhaps in non-technical terms or with concrete examples, so the less intelligent folk will also be able to grasp it too?

Thanks. 8-))

Kenny,

Immagine a poor elderly person who has to make a choice between paying for the pills required to maintain life, or putting food on the table. If the pension isn't enough to pay for both goals then they have a conflict.

The pills, the required calories of food, and the level of expenditure are the subject of attempts to control by three systems. If the pension was large enough this person wouldn't have to face a choice ( a conflicted choice ) between pills and food.

Now that the theoretical problem has been solved and an adquate defintion obtained we can explain this symple relationship to our congressmen and women and the genuine problem facing old people in this circumstance cab be corrected. ( That is controlled by reference to somesort of humane reference level. )

Bill Williams
  UMKC

···

-----Original Message-----
From: Control Systems Group Network (CSGnet) on behalf of Kenneth Kitzke Value Creation Systems
Sent: Mon 12/1/2003 7:54 AM
To: CSGNET@listserv.uiuc.edu
Subject: Re: Conflict; degrees of freedom

[From Kenny Kitzke (2003.12.01)]

<Bill Powers (2003.11609 MST)>

Conflict exists in a set of N control systems when the number of
reference degrees of freedom is greater than the number of degrees of freedom
in the set available for perceptual control.

<Bingo! I think this not only defines conflict (of all degrees) but explains
it as well.>

Breakthrough! With this definition, we can finally go forth and use PCT/HPCT
to explain conflict that exists in and between human beings.

The teen boy, conflicted between using or not using cocaine, will be assuaged
when we share this definition with him.

The husband and wife, in conflict over whether to continue or end their
marriage, grasping for understanding, will have their eyes opened and rejoice to
hear the correct explanation of their conflict.

Now, while this definition of "conflict" is perfectly clear to me (ah-ha),
and so much more useful in addressing situations of "conflict" in myself, or
between me and other control systems, could you or Martin explain it just a
little bit more, perhaps in non-technical terms or with concrete examples, so the
less intelligent folk will also be able to grasp it too?

Thanks. 8-))

[From Bruce gregory (2003.12.01.1041)]

Kenny Kitzke (2003.12.01)

Now, while this definition of "conflict" is perfectly clear to me (ah-ha), and so much more useful in addressing situations of "conflict" in myself, or between me and other control systems, could you or Martin explain it just a little bit more, perhaps in non-technical terms or with concrete examples, so the less intelligent folk will also be able to grasp it too?

If you have to ask... Just kidding. I think this would be an excellent idea.

Bruce Gregory

"Everything that needs to be said has already been said. But since no one was listening, everything must be said again."

                                                                                Andre Gide

[Martin Taylor 2003.12.01.1059]

[From Kenny Kitzke (2003.12.01)]

<Bill Powers (2003.11609 MST)>

>Conflict exists in a set of N control systems when the number of

reference degrees of freedom is greater than the number of degrees of freedom
in the set available for perceptual control.

<Bingo! I think this not only defines conflict (of all degrees) but explains
it as well.>

Breakthrough! With this definition, we can finally go forth and use
PCT/HPCT to explain conflict that exists in and between human beings.

Explanation is only partway to solution. It takes a lot of precision
measurement and computation, not to mention engineering skill and
manufacturing effort, to get from Newton's explanation of planetary
orbits to an ability to land a lunar module in a particular crater.

Having said that, let's look at your examples.

The teen boy, conflicted between using or not using cocaine, will be
assuaged when we share this definition with him.

I imagine that there are at least two distinct perceptions being
controlled, one of which involves a reference for something relating
to cocaine use (physiological pleasure, approval of friends, or
whatever), and the other for something else related to cocaine use
(approval of parents, self-image as a law-abiding person, or
whatever). That's two degrees of freedom for perceptual reference
values, but only one for the environmental variable of cocaine use
that is important for the control of each perception.

Would that help him be assuaged? Probably not until the actual
perceptions being controlled through that environmental variable have
been discovered, and the possibility of finding another degree of
freedom through which one or other (or both) can be controlled is
discovered. For example, suppose that the controlled perceptions are
"approval of friends" (use cocaine) and "approval of parents" (no
cocaine).

Can a way be found to retain both approvals, at whatever level of
cocaine use? He would have to look for other environmental variables
that might affect one or other approval. Perhaps he could show
affection for parents that would enhance their approval sufficiently
to allow for cocaine use that would gain approval of friends. Perhaps
he could demonstrate his value to the friendship group in ways other
than by using cocaine, and thus maintain their approval without
jeaopardizing parental approval. Either way, an extra degree of
freedom has been added to the environmental path, so that the number
of degrees of freedom for perceptual control now equals the number of
references for perceptions being controlled.

Obviously, in any actual situation, there will be a more complex set
of controlled perceptions. But this should suffice to show how the
notion of degrees of freedom might apply in that situation.

The husband and wife, in conflict over whether to continue or end
their marriage, grasping for understanding, will have their eyes
opened and rejoice to hear the correct explanation of their conflict.

Here, as you describe the problem, one wants to end the marriage, and
the other does not. For each, this is a mechanism for controlling
some other perception (or in this case, often a plethora of other
perceptions, though often it comes down to one at some high level).
As with the internally conflicted boy, it is necessary to discover
just what perceptions are being controlled through the existence or
non-existence of the marriage, and whether some other means can be
found for one or both to control their perceptions using some other
means.

Again, it is necessary to add degrees of freedom sufficient to allow
the various perceptions to be controlled--or to relinquish attempts
to control one or more of those perceptions. In the "marital dispute"
case, the latter is a distinct possibility, since at its root, the
perceptions being controlled get their reference values from
something like control of a perception of self-image, which uses many
different perceptual control systems at lower levels (I imagine Dick
Robertson might have something to say on this). So it may be the
"self-image" controlled perceptions that could be controlled by using
different lower-level perceptions that do not depend on
ending/maintaining the marriage for successful control.

So, the resolution of a conflict couold occur by increasing the
environmental degrees of freedom available for control, or by
reducing the perceptula degrees of freedom being controlled through a
particular environment.

However, when the degree of freedom bottleneck is the control of "the
same" perception at two levels (e.g. to have a Jwwish state covering
the whole of Palestine, or to eliminate Jewish authority over any of
Palestine), that conflict is irreconcilable so long as the
perceptions being controlled don't change. There's only one degree of
freedom for perceptual control, with two perceptual signals being
controlled (representing the same function of environmental
variables). So the ONLY solution is for one or other of the
controlled perceptions to alter the function of environmental
variables it represents (e.g. control for the kind of authority
exercised over disparate parts of Palestine). That increases the
degrees of freedom available for perceptual control to match the
number of references (two) for controlled perceptions.

could you or Martin explain it just a little bit more, perhaps in
non-technical terms or with concrete examples, so the less
intelligent folk will also be able to grasp it too?

Does this help?

Martin

[From Rick Marken (2003.12.01.0930)]

Kenny Kitzke (2003.12.01) --

Breakthrough! With this definition, we can finally go forth and use PCT/HPCT to
explain conflict that exists in and between human beings...

Now, while this definition of "conflict" is perfectly clear to me (ah-ha), and
so much more useful in addressing situations of "conflict" in myself, or between
me and other control systems, could you or Martin explain it just a little bit
more, perhaps in non-technical terms or with concrete examples, so the less
intelligent folk will also be able to grasp it too?

I think Bill [Bill Powers (2003.11609 MST)] gave the best non technical
explanation of what the definition of conflict means: it means that "Conflict is a
state of impossibility". What is impossible is achieving (controlling for) all
one's goals. You can't always get what you want! You can't because the number of
degrees of freedom in the set available for perceptual control is smaller than the
number of reference degrees of freedom (which doesn't scan nearly as well as
Mick's version).

The definition of conflict says that there is no solution to conflict because
conflict is a state of impossibility; if there is a possible solution, then there
is no conflict. The only real "solution" to a real conflict is to dismantle it.
Since conflict is created by a situation where there are more goals than means to
achieve them, the only way to dismantle conflict is to change one's goals,
functionally reducing the number of independent perceptions you want to control
simultaneously.

So the solution to conflict involves recognizing that "You can't always want what
you want"!

Best

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

[From Bruce Gregory (2003.12.01.1247)]

Rick Marken (2003.12.01.0930)

The definition of conflict says that there is no solution to conflict
because
conflict is a state of impossibility; if there is a possible solution,
then there
is no conflict.

This strikes me as a bit odd. Just because there is a possible
solution, doesn't mean that the "parties" involved have found it, does
it?

The only real "solution" to a real conflict is to dismantle it.
Since conflict is created by a situation where there are more goals
than means to
achieve them, the only way to dismantle conflict is to change one's
goals,
functionally reducing the number of independent perceptions you want
to control
simultaneously.

Let's look at the mid-east (I know, I know. Who wants to look at the
mid-east...) It does not help to say that the two parties must change
their goals. (Even though this is true.) It seems to me that the two
parties must agree that the same outcome would meet the majority of
their goals. If they cannot agree to this, the outcome will be
perpetual conflict. This "analysis" does not conflict with your
statement above.

Bruce Gregory

"Everything that needs to be said has already been said. But since no
one was listening, everything must be said again."

                                                                                Andre Gide

[From Bill Powers (2003.12.01.0859 MST)]
Kenny Kitzke (2003.12.01)–
Here’s an attempt at an explanation, and a few free added
comments.
The basic situation is that a person is trying to control N variables,
but for some reason can’t affect all of them independently. Either
several of them are related so they change together, or your actions
affect more than one of them in the same way, or for some other reason
you can’t control all of them any way you want. It turns out that if you
leave out one or more variables, you can control the rest relative to any
reference levels you want, but the remaining ones will be uncontrolled.

Suppose you are trying to control 10 variables. And say 7 of them at a
time – any 7 – can vary independently. But if you try to change 8 or
more of them at a time, you can’t get all of them to their reference
levels. You could make any 7 of them match their reference levels
exactly, but then the others would have whatever values they had; you
couldn’t control them. Or, you could try to reduce the overall error as
much as possible, in which case none of the variables would be at its
reference level.
A simple way to see this is in a buying situation, where the amount of
each thing you buy depends not only on how much or what kind of it you
want, but on how much money you can spend. If you’re buying ice-cream
cones for a birthday party of 10 children, and cones cost $1.00 each
including tax, and you have $9.50, then any 9 of the children can have an
ice-cream cone, but not all 10.
This gets more complicated when different children want different
flavors, and each flavor costs a different amount. There will be
many combinations of flavors that you can pay for with $9.50, and many
that would run you over the limit. So it’s simply not possible, in
general, for each child to have a totally free choice of flavors (no
credit at this ice-cream store). You might get lucky, but you might
not.
Instead of 7, 8. or 9 children getting what they want and 3, 2, or 1
getting nothing, you could ask the children for their first, second, and
third choices of flavors. Now you have a much better chance of being able
to find combinations you could afford. In effect, you’re spreading the
error out over all the children instead of assigning it all to a few
while the rest have zero error. Maybe there’s a combination in which all
the children get their second choice or better; then nobody has whatever
error 3rd choice, or no ice cream, means.
This is probably what happens in a system made of many control systems
when, for any reason, the practical range of values of all the controlled
variables is less than the maximum possible. Each control system adjusts
its output to bring its own reference signal as close as possible to its
own reference level. The interactions among the systems then determine
how much error will be left in each system. You can figure that there
will be no system with exactly zero error, because that system would stop
trying to change its input, and all the other systems would then
determine the result. It would be most unlikely for that result to be
zero error in any one system. On the other hand, if some systems can
exert much greater maximum influences on the environment than others, or
have much higher loop gain, they can enjoy much smaller errors than the
other systems. In other words, systems with more power and stronger
desires than other systems will get closer to what they want than the
rest will.
In mathematics, problems like these boil down to sets of simultaneos
equations in many unknowns, one unknown for each variable. In order for a
solution to exist when there are N variables, there must be N
independent equations, which means what Martin meant by the term
orthogonal, and which would take more discussion than appropriate
here to make clear. It’s usually said that unless there are N independent
equations, there is no solution for the set of equations; that is, it’s
not possible to solve for the values of all tha
variables.
When we try to apply this idea to the real world, we find that systems
described by N equations in N unknowns are ALWAYS solvable in fact even
if not in theory. Suppose we have the following two equations, set up by
measuring the sum of two experimental variables with weights applied to
them:
2x + 4y = 7
5x + 10y = 9
The set of equations is unsolvable if the “determinant” of the
coefficients is zero. The determinant is (210 - 54), or 0. So there is
no solution for these equations. If you plotted these equations you would
see that they describe two parallel lines, which never cross and thus
have no point in common.
However, suppose we find that the number 4 in the first equation is
experimentally measured as 4.0000000000000000001. Now there is a
solution, because the determinant is now 210 - 54.0000000000000000001,
or,
-0.0000000000000000001. That is not zero, so there is a
solution.
In short, when the numbers used in equations are measured experimentally
and expressed with the full accuracy of the measurements, just about any
set of N equations in N unknowns will have a solution. The mathematical
criterion for not having a solution is meaningless: there’s no
significant chance that the determinant will be exactly
zero.

The only limitation on solutions of N equations in N unknowns is the
practical one: what values of the variables are required to reach this
solution? In the equations above, the solution for x (going with
4.00000000000000001) is

x = (70 - 36)/-0.0000000000000000001.

Not even bothering to count zeros, we can see that x is an absolutely
enormous number if it’s a measurement of any physical variable. If x
represented the required output from a control system, we could forget
it: no real control system could produce that much x, no matter what x is
(aside from particle physics, don’t quibble).

The point here is that even when mathematics says a set of N equations in
N unknowns has a solution, it may not actually have a solution in any
practical terms. If constants in our set of control systems are anywhere
near to the values that tell us there is no solution, they will be in
conflict and there will be no practical solution even if mathematics
tells us there is one. In the plots of the above equations, this is the
case when the lines are near enough to parallel that their point of
intersection is way off the chart. The point only has to be extreme
enough to lie outside the range that any real system could produce. Then
there must be conflict to some degree.

When the number of equations is less than the number of variables, there
is no solution, period. What this says is that the control systems won’t
work: one or more variables is going to be uncontrolled. Or to put this
differently, for every value of the left-over variables there will be a
different solution for the others. This basically tells you that you
haven’t got a complete enough model, and you can’t make any predictions
at all. It’s like saying that your bank account consists of total
deposits minus total withdrawals plus an unknown mistake. You can account
for two of the variables, but not the third, so you don’t know what your
bank account is. Models have to be complete even if they’re not correct;
if they’re not complete, you’ll never know if they’re wrong or
right.

Thanks for asking.

Best,

Bill P.

Melding
[From Bjorn Simonsen(2003.12.02.09;50, EuST)]

[Martin Taylor 2003.12.01.1059]

Martin

If I understand PCT and Bill correct we will not get a solution given the condition we have a conflict. Reorganization will set in and maybe make a solution. But if Reorganization still is accidental, I have a problem with your:

<Can a way be found to retain both approvals, at whatever level of cocaine use? He would 
have to look for other environmental variables that might affect one or other approval. >

I understand this as a kind of Planning . Here we have a problem with Rick’s categorical statement:

<[From Rick Marken (2003.08.25.0910)]
Yet when we look at the basic phenomenon of control we see that
planning can't really "allow" us to control in situations where there are
unpredictable changes in the environment (variations in disturbances to the
variables we control) and/or our connection to it (variations in functions    
connecting us to the variables we control) -- that is, planning can't help us
control in any real world situation>

And

[From Bill Powers (2003.08.26.0814 MDT()]

<Basically, planning is (as Tom Bourbon indirectly said) like a control
system pretending to be a stimulus-response system. If this happens, do
that. Unfortunately, the world is continuously changing, so what you need
to do to have a certain effect on Monday will not be quite right on
Tuesday.> 

Do I mistake your explanation to Kenny, Martin? Do you really think that … “He would
have to look for other environmental variables that might affect one or other approval.” ? Is this planning and is this control of perceptions?

I think we have a complete model regarding internal conflicts. We have the definition and we have the prescription how to get a solution (if there is).

Bill indicate:

<[From Bill Powers (2003.11.27.0659 MST)]
If a higher order system or set of systems exists that can perceive the existence 
of the conflict and has learned how to resolve such conflicts (say, by momentarily 
turning off the goal of getting to the meeting on time), the conflict will immediately 
be resolved. Most conflicts are resolved in this way and cause only momentary 
inconveniences. But if the situation prevents any easy solution or there is no 
already-known solution, the conflict will simply persist. If the result of its persistence 
is to create significant large errors in other control systems, reorganization will 
probably start, and there is then no way to predict what the solution will be, if a 
solution is found.>

Some times we go up a level and some times Que Sera …Sera.

bjorn

Re: Conflict; degrees of
freedom
[Martin Taylor 2003.12.02 1016 EST]

I apologize for the extensive quote with which this
starts…

[From Bjorn
Simonsen(2003.12.02.09;50, EuST)]
[Martin Taylor
2003.12.01.1059]
Martin

<Conflict exists in a set

of N control systems when the number of

reference degrees of freedom is greater than the

number of degrees of

freedom in the set available for perceptual

control.>
If I understand PCT
and Bill correct we will not get a solution given the condition we
have a conflict. Reorganization will set in and maybe make a
solution. But if Reorganization still is accidental, I have a problem
with your:

<Can a way be found to retain both approvals, at whatever level of
cocaine use? He would

have to look for other environmental variables that

might affect one or other approval. >
I understand this as
a kind of Planning . Here we have a problem with Rick’s
categorical statement:

<[From Rick Marken (2003.08.25.0910)]

Yet when we look at the basic

phenomenon of control we see that

planning can't really

“allow” us to control in situations where there
are

unpredictable changes in the

environment (variations in disturbances to the

variables we control) and/or

our connection to it (variations in
functions

connecting us to the variables

we control) – that is, planning can’t help us

control in any real world

And
[From Bill Powers
(2003.08.26.0814 MDT()]

<Basically, planning is (as

Tom Bourbon indirectly said) like a control

system pretending to be a

stimulus-response system. If this happens, do

that. Unfortunately, the world is continuously changing, so what you
need

to do to have a certain effect

on Monday will not be quite right on

Tuesday.>
Do I mistake your
explanation to Kenny, Martin? Do you really think that … "He
would

have to look for other environmental variables that might affect one
or other approval." ? Is this planning and is this control of
perceptions?

This is a very shrewd question. I think the answer lies in time
scales, the relative stability of the “rules of the world”
and the definition of planning.

First, let’s consider Bill’s statement: “Unfortunately, the world is continuously changing, so
what you need to do to have a certain effect on Monday will not be
quite right on Tuesday”

Different things about the world change at different rates. What
we call “Laws of nature” are unlikely to change between
Monday and Tuesday, but whether it is true that “Today is Aunt
May’s birthday” must change.

In mainstream psychology, memory has a variety of types: memory
of how to do something (in PCT, largely embodied in the reorganization
of the structure of the hierarchy), memory for the momentarily
unsensed states of the environment that are currently part of the
perceptions in the hierarchy (currently part of “speculative
PCT”), and memory for past events (conventionally called
“episodic memory”), among others.

The only member of this triad susceptible to change between
Monday and Tuesday is memory for states of the environment not
currently observed–and that is a great deal of the environment within
which we control. Just look around your room. I expect that you
“see” a more or less rectangular box (or some other shape),
and that shape stays unchanged as you look around again. You perceive
yourself to be in a rectangular box. But you NEVER actually see the
rectangular box. You see in some detail a tiny fraction of the box
directly in front of your eyes, and in ever-decreasing detail less
than a third of the rest of the box. However, the perception of the
box is there, and is usable (in principle) for planned actions.

I can plan to put a book on the bookshelf that is (I imagine)
still at my back. But someone may have moved the bookshelf when I was
not looking. My perception of the room may be wrong–but it is still
my perception of the room.

So it is incorrect to say “planning can’t help us control in
any real world situation”. It is more accurate to say that
disturbances may require continuous observation to control well,
despite good planning.

On the other hand, a lot of real-world disturbances don’t have a
white noise-like spectrum. They don’t happen continuously pushing
perceptions this way and that without cease. Many are discrete events,
such as someone jogging the elbow as you pick up a glass. Most often,
the plan works.

In CSGnet discussions many years ago, we talked about
“ballistic” or “fire and forget” control. Once the
bullet has left the gun, nothing more can be done to make it hit the
target, and if the target moves before the (accurately aimed) bullet
arrives, it will miss. Only because real-life bullets tend to move a
lot faster than real-life targets do gunshots usually hit the target
if the aim is good.

Planning is like a high-level fire-and-forget, and can work
without on-line control only if the environment changes slowly in
respect of the planning objective (I include here the possibility of
imagining future developments of the environment as part of the
perceived environment, just as predictive fire control allows a gun to
hit a moving target).

But if planning is at a relatively high perceptual level, the
relevant actions are executed by providing reference values for
control systems at a lower level. These need not be planned. The
high-level perceptual environment may be (and often is) much more
stable than the immediate sensory environment.

I may make a plan when I am in school that by the time I am 50 I
will be President. That plan involves taking certain school courses
and then getting involved in local politics, and so on and so forth.
But the execution of the plan involves lots of variation within that
plan, and over months and years even the high-level enviroment may
change. I may realize that I am not approaching my goal, or, in PCT
terms, the error in my “perceive myself to be achieving the
Presidency” control system is not decreasing. I may choose
different methods, such as organizing a military coup. Again, planning
comes in, but on-line control at lower levels is required to deal with
the faster disturbances that usually occur at the lower perceptual
levels.

Even without reorganization, a control system is likely to have
various actions available to it, and if one is blocked, another will
be used. A plan can provide stable reference values for an on-line
control system, but on-line control may well be necessary for the
plan’s success.

There’s a quite separate aspect of this time-scale issue, and
that is the difference between what we might call “the way the
world works” and “the things in the world.” The way the
world works is likely to be pretty stable, whereas things come and go.
The result of reorganization (or even evolution), the structure of the
organism, is more or less fitted to the way the world works. It could
have been designed, given sufficient knowledge on the part of the
designer (which is why creationism appeals to some people). The
“things in the world” change on varying time scales, and it
is they that make “the best laid plans of mice and men gang aft
agley.”

Time scales and the discreteness of rare events are what matter
when we deal with whether planning and PCT are compatible.

I don’t know if I answered the question, but it’s a valuable
topic.

Martin

Melding
[From Bjorn Simonsen (2003.12.04 13,25 EuST)]

Martin Taylor 2003.12.02 1016 EST

<.............I think the answer lies in time scales, the relative stability of the "rules of the world" 
and the definition of planning.

Interesting.I will say something about time and about definition.

I have discussed the concept planning one or two times earlier. But I got a negative response and answers indicating that Planning is a NO interesting concept in PCT. Shall we try to go a step forward? (or is it back again?). We need to exemplify the concept, define the concept and describe what is happening in the hierarchy.

You described an example where you used the concept Planning. This is good enough:

Example

<I may make a plan when I am in school that by the time I am 50 I will be President. That plan involves

taking certain school courses and then getting involved in local politics, and so on and so forth. But the

execution of the plan involves lots of variation within that plan, and over months and years even the

high-level environment may change. I may realize that I am not approaching my goal, or, in PCT terms,

the error in my "perceive myself to be achieving the Presidency" control system is not decreasing. I

may choose different methods, such as organizing a military coup. Again, planning comes in, but

on-line control at lower levels is required to deal with the faster disturbances that usually occur at

the lower perceptual levels.>

Before I hint at a definition for the concept Planning, I will comment parts of your mail.

<In mainstream psychology, memory has a variety of types: memory of how to do something

(in PCT, largely embodied in the reorganization of the structure of the hierarchy), >

For me “memory of how to do something” is the same as B:CP names Remembering. I don’t disregard Remembering involve reorganization, but that is when Remembering is unsuccessful. I think Remembering in essentials is control at the Reorganizations level with non switches vertical.
If I go into Imagination mode and try to remember how to come from A to B, I may not succeed. The error provokes a perception signal which after output initiate new perceptual signals also going to the levels above. This may (after some time (slowing factor)) bring me to control the perception “finding a map”. I have a control system for finding and reading a map. Reorganization is unnecessary. You said the same: "Even without reorganization, a control system is likely to have various actions available to it, and if one is blocked, another will be used.

<memory for  the momentarily unsensed states of the environment that are currently part of the perceptions 
in the hierarchy (currently part of "speculative PCT"),>

I agree. That form for memory is what I will emphasize unconscious perceptions. I think Bill described it nearly as you:

[From Bill Powers (2003.11.27.1543 MST)]
<…When you examine an object, you can perceptually analyze it into smaller objects, in the way
that a chair is seen to be made of arms, legs, rungs, decorations, and whatever other parts there are. It can
also be seen as part of a display of furniture in a store, and the store can be seen as part of the shape of a
city, and so forth. Going to the smallest possible parts, like atomic particles, or the largest, like galaxies,
requires imagination.

I don’t visualize planning as connected with that form for memory or unconscious perceptions.

<So it is incorrect to say "planning can't help us control in any real world situation". It is more 
accurate to say that disturbances may require continuous observation to control well, despite 
good planning.>

I am inclined to agree. Disturbances is transformed in the input function to perceptual signals, and they are controlled.

<But if planning is at a relatively high perceptual level, the relevant actions are executed by

providing reference values for control systems at a lower level. These need not be planned. 
The high-level perceptual environment may be (and often is) much more stable than the 
immediate sensory environment.>

I imagine the planning in your example above as a high perceptual level, higher than events. If one plans something to happen in a month or a year I imagine it is a kind of relationship level planning. Happenings are perceived relative to other happenings. I think all planning is at levels above the Events level.

<The "things in the world" change on varying time scales, and it is they that make "the best laid 
plans of mice and men gang aft agley.">

I think time scales has something to do “in the real world”. There must be space and change for time to be perceived. When we make Plans, we omit time. There are seldom time scales in Plans, there are Relationships.

I imagine Planning as a kind of imagining. Planning is imagining (both switches are disconnected).

I think the lowest level for planning is the Relationships level, but Planning/imagining may send perceptions to higher levels. I think the Programs level very often is active. I think the example of planning above is a Planning/Imagining at the levels between the Relationships level and the Principles Level. This Planning is more or less stable, it can be fragmentary, difficult to express with words and I can say more.

It is accepted that we in Imagination mode can imagine an elephant with a horse’s head. Something not existing. In the same way it must be accepted that we can imagine a combination of perceptions that are existing. This combination of perceptions may be perceived at a later moment of time when Plans are flipped in.

And here is my central point. When it is time to manipulate Planned perceptions we flip the memory switch to the reference position and initiating the lower-order behaviors. Now we are back in business. There certain is an uncertainty which control systems are initiated. I know Bill says “Unfortunately, the world is continuously changing, so what you need to do to have a certain effect on Monday will not be quite right on Tuesday”

But I don’t think he is a Behaviorist. I think he will also say: "There are many disturbances in the real world, but remember Behavior is Control of Perceptions and the Reference value is an important value. If the control of perceptions not results in an error with value zero, Reorganizing may happen.

I remember Rick [From Rick Marken (2003.08.25.0910)]

<Yet when we look at the basic phenomenon of control we see that
    planning can't really "allow" us to control in situations where 
    there are unpredictable changes in the environment (variations 
in disturbances to the variables we control) and/or our connection 
to it (variations in functions connecting us to the variables 
    we control) -- that is, planning can't help us control in any real 
    world situation.>

I am inclined to say that Planning is imaging what to do. I accept Rick Marken (2003.05.07.1230) “Memory is the process of accessing previously experienced perceptions”. But in Imagining mode one at the Relationships level can combine previously experienced perceptions in a new Relationship. There is no guarantee that the Reference values which are put into operation provoke perceptions which result in an Error value at Zero. But this is “Business as usual”. Behavior is control of perceptions and if the error doesn’t approach zero, reorganization sets in.

Rick expresses the same in other words [From Rick Marken (2003.05.07.0910)]
<How much of our perception is built from
memory-produced perceptions is
an empirical question. My guess is relatively little for normal people,
quite a bit more for so-called schizophrenics.>

I may agree with Rick. Also I think people most control their perception. And that’s life.But when I use the concept Planning I hypothesize that normal people plan perceptions which they in Control mode can control.
I expect Rick then will say that those Plans are perceptions already existing in the hierarchy, and I think I still agree. But there has been a process in Imagining mode. And this process has activated the control of new (earlier existing?) perceptions. The result is that I control other perceptions I would have controlled if NO Planning had taken place.

So what is Planning comparative to PCT?
An example of planning is mentioned above.

Planning (a suggestion to definition): Planning is the process a person does in Imagination mode, perceiving at the Relationships level and higher perceptions which can be controlled in control mode.

What is happening in the hierarch when Planning: Look at B:CP page 221.

>I don't know if I answered the question, but it's a valuable topic.>

You did. And also I find Planning as a valuable topic. But I expect somebody saying Planning is a NON existing concept in PCT. The perceptions controlled after “a Planning” process are the perception controlled if “the Planning” process had not happened. … Who can verify that?

bjorn

I have discussed the concept
planning one or two times earlier. But I got a negative response and
answers indicating that Planning is a NO interesting concept in
PCT.
I may make a plan when I am
in school that by the time I am 50 I will be President.
[From Bill Powers (2003.12.04.1024 MST)]
Bjorn Simonsen (2003.12.04 13,25
EuST)–
My main idea about planning is that at whatever level we make plans,
we’re planning results, not behavior. – that is, perceptions, not
actions. So planning is already part of HPCT: plans are reference
conditions for perceptions.
That is an outcome, a result, not an action. The actions you will have to
perform to achieve this result are only partly predictable, and then only
to the extent that the environment remains the same so the action will
continue to produce the same result. Your reference-perception is “I
am president and I am no more than 50 years old.”
“President” is part of a system concept, so to extablish that
perception, you must vary reference signals for principles (get a good
education), which require setting reference-programs, which requires
setting reference-sequences., and so on down to tensing the muscles that
cause the hand to pick up the pen to write your name on the application
to Yale.
So “planning” is simply another word for how the hierarchy
works. It means setting reference conditions.
When we plan, we’re really planning what we intend or want to perceive.
This even applies to physical actions. “I’m going to hammer that
nail in” is a plan about a physical process that is to be perceived,
but not about how hard, in which direction, or how many times to swing
the hammer. The actions that are required depend on details of the
environment that are either hidden from view (the knot hidden inside the
plank) or that haven’t come into existence yet (the neighbor who soon
rushes in to borrow your hammer for an emergency repair).
I repeat what I keep saying while on this subject: that there are far
more disturbances in the environment and our relationship to it than
people seem to realize. We see ourselves “doing the same thing”
over and over, which means perceiving a repeated result or effect, and
simply assume that our action must have been the same since the result
repeated. But a close look, even with the unaided eye, will show that the
actions are always varying – and if the actions didn’t vary, the
results would vary.
The same thing applies to plans as preselected outcomes of actions. If
the actions did not vary with every environmental disturbance, with every
change in our relationship to the environment, the planned outcome could
not possible happen, at least in any realistic environment. So there is
nothing about planning that is different from anything else we do, except
the specific perceptions for which we are setting reference
conditions.
The stickiest part of this subject involves contingency planning. Here,
surely, we are trying to anticipate all the disturbances that might
arise, and selection actions that will oppose each possible kind of
disturbance. NASA has a book of contingency plans for refuelling the main
tank of the Shuttle; it’s a foot thick. But when you read parts of it,
you do not see
Step 1.1.3.7a; Separate thumb and forefinger by 8 cm.
b:
Move hand until nut is centered between thumb and forefinger
c.
Bring thumb and forefinger together by 4 cm each
That book would be 40 miles thick. And even then, all you
would find in it would be descriptions of perceptions, as
above.

A contingency plan is an attempt to say how one set of perceptions
depends on other perceptions. IF the nut has fallen on the floor (which
usually doesn’t happen, but could happen) THEN a series of other
perceptions must be created to bring the perception of the nut back to
its former condition. So the corrective “actions” involved are
really perceptions that affect our perception of the position of the nut.
The person constructing a contingency plan begins with the assumption
that some disturbance has occurred, some change in a perception, some
error that needs to be corrected. Then that person has to think of some
physical process that could be carried out so as to restore the disturbed
variable to its original state, or closer to it.

Obviously, imagination is involve. Pick up the nut with a magnet, one
imagines. But where is a magnet to be found? In a cabinet, nonexistent as
yet, marked “EMERGENCY TOOLS FOR PICKING THINGS UP,” which one
then jots down in the list of equipment that is to be on hand for filling
the shuttle’s fuel tank.

One must have a serviceable mental model of properties of the world in
order to make contingency plans. If the nut has fallen on the floor, one
must realize that the contingency plan cannot be "“Pick up nut
and replace on dolly”, if this is the nut that holds the engines
onto the back of the Shuttle, and it weighs 400 pounds (I made that up,
in case anyone is amazed that one nut holds the engines on). Contingency
plans have to be possible to carry out, and when carried out they must
actually have the required effect.

Anyway, no matter how detailed the plan, it can never be detailed enough
to permit the blind execution of physical actions to succeed in carrying
out the plan. “Never” is an exaggeration, since a determined
obsessive-compulsive engineer can arrange the environment so that
practically any variant of an action will create a predestined result,
but in the natural world no such engineers are standing around to help
make plans work.

I guess I have remarked from time to time that I think planning is
overrated. But I don’t really mean we can do without it. It’s really a
good idea for a plumber to bring his tools with him, and so forth. I’m
just reacting against excesses in the other direction, produced by people
who think they can plan for every possible problem that might arise, and
are therefore helpless when anything unexpected happens.

Best,

Bill P’