Ruminations on Importance

[From Rick Marken (2003.06.27.1130)]

Bruce Gregory (2003.0627.1217)

This particular fire hydrant has always had trouble understanding how
the control hierarchy 'decides' which perception to control and, in
particular, how it rapidly shifts from controlling one perception to
controlling another without the delays associated with relying on upper
levels in the hierarchy to reset reference levels.

It depends on what you mean by "which perception to control". I think of the
hierarchy as always controlling a large (and basically fixed) set of perceptual
_variables_. The state in which any perceptual variable is maintained is varied
as the means of keeping other, higher level perceptions under control (see my
"Spreadsheet analysis..." paper in _Mind Readings_). So if the question "which
perception is controlled?" means "to which state is a perceptual variable brought
at any particular time?" then this question can be answered in terms the normal
operation of the hierarchy. What perception is controlled depends on the state to
which a perceptual variable is brought in order to maintain control of a higher
level perception. For example, I must change the perceived location of my finger
in order to type a particular letter. The higher level goal (typing "a", say)
determines which perception of finger position I must get.

If, on the other hand, "which perception is controlled?" means "which perceptual
variable is controlled at any time?" then my answer is "I don't believe that we
switch from controlling one perceptual variable to another". I think the PCT model
says that we are always controlling the same set of perceptual variables all the
time (until we learn to perceive new ones or lose old one). Perceptual changes
that seem like changes from one to another perceptual variable could simply be
changes in the states of perceptual variables that are always under control. For
example, changing from controlling for driving to controlling for going into the
house could be conceived of as a change in the state at which each of these
perceptual variables (driving, going in house) is controlled. The driving variable
is set to zero while the going in the house variable is set to maximum. The
changes are being made, presumably, as the means of keeping still higher order
variables, like "going home" under control.

When I look up from
tuning my car radio and see the brake lights of the car in front of me,
how does the perceptual hierarchy know enough to slam on my brakes?

I think you are continuously controlling for (among other things) avoidance of a
collision with the car in front of you when you are driving. The brake lights are
a sudden disturbance to this variable; a disturbance that is handled in whatever
way is appropriate (slamming on the brakes, swerving, etc).

When the traffic is dicey, how does the perceptual hierarchy know that
I can't afford to direct my attention to looking for a cassette to
play?

Attention is a consciousness phenomenon. So the hierarchy is not involved in
allocating attention. Moreover, attention changes don't affect which variables you
are controlling. What you are talking about is a resource allocation problem; your
eyes can't foveate the radio and the cars ahead at the same time. But you need to
control visual variables to control the radio and the car's relationship to
traffic. I think very high levels of the hierarchy (the program level) are
involved in this low level resource allocation. When you drive, you are probably
controlling for a program like "if traffic dicey then keep eyes on road, else
allow eyes to move from road for no more than 5 seconds".

It has always seemed from my lowly position that the hierarchy is
extremely capable but lacking in much sense of what is important at the
moment.

It can be made to have as much "sense" as you can imagine it to have, where
"sense" would probably be defined in terms of the program and principle
perceptions controlled by any particular hierarchy.

Best regards

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 (2003.0627.1217)]

This particular fire hydrant has always had trouble understanding how
the control hierarchy 'decides' which perception to control and, in
particular, how it rapidly shifts from controlling one perception to
controlling another without the delays associated with relying on upper
levels in the hierarchy to reset reference levels. When I look up from
tuning my car radio and see the brake lights of the car in front of me,
how does the perceptual hierarchy know enough to slam on my brakes?
When the traffic is dicey, how does the perceptual hierarchy know that
I can't afford to direct my attention to looking for a cassette to
play? It has always seemed from my lowly position that the hierarchy is
extremely capable but lacking in much sense of what is important at the
moment.

To the extent that this assessment is not wholly without merit, I offer
the following questions. Could it be that the limbic system functions
to tell the control hierarchy what is important 'right now'? Further,
could it be that the mechanism employed by the limbic system to convey
this information consists of increasing the gain on some control loops
and reducing the gain on others? These questions do not seem totally
implausible on the basis of what little I know about the wiring of the
brain.

[From Rick Marken (2003.06.27.1500)]

Bruce Gregory (2003.0627.1703)]

So the process involves conflict since the higher level goal of getting
to the meeting on time and the still higher level goal of continuing to
advance my career are still endeavoring to keep my foot on the
accelerator...

I think you should take Bill's [Bill Powers (2003.06.27.1517 MDT)] excellent
advice and try to figure this out for yourself. I will give you a little hint:
focus on the idea of conflict between systems at different levels.

Happy modeling.

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 Rick Marken (2003.06.27.1512)]

Fred Nickols wrote:

Bill's response to Bruce G prompts a question on my part:

>I think that if you put your mind to it, you could answer all these
>questions just as well as Rick or I could. They all propose hypothetical
>problems, and the only answers expected are deductions or extrapolations
>from PCT or HPCT. This game doesn't work very well if one person is
>confined to asking probing questions while the other person has nothing to
>do but think up plausible answers. Both can obvious go on doing this
>indefinitely. It is much more productive if each person thinks up probing
>questions and then tries to answer them as far as possible, and failing
>that modifies the theory to correct the problem.

Bill's response seems to me to be predicated on having spotted a pattern, a
"game" as Bill calls it.

Where do "games" such as this fit into the PCT hierarchy.

Great question. I think a game is a good example of a program level perception.
Perceiving a game involves perceiving someone carrying out a contingent set of
actions, within constraints, to achieve the goal of "winning" in some sense.

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 (2003.0627.1703)]

Rick Marken (2003.06.27.1130)

When I look up from
tuning my car radio and see the brake lights of the car in front of
me,
how does the perceptual hierarchy know enough to slam on my brakes?

I think you are continuously controlling for (among other things)
avoidance of a
collision with the car in front of you when you are driving. The
brake lights are
a sudden disturbance to this variable; a disturbance that is handled
in whatever
way is appropriate (slamming on the brakes, swerving, etc).

So the process involves conflict since the higher level goal of getting
to the meeting on time and the still higher level goal of continuing to
advance my career are still endeavoring to keep my foot on the
accelerator. Why does the foot move from the accelerator to the brake?
Why doesn't the foot wind up halfway between the two as a result of the
conflict?

When the traffic is dicey, how does the perceptual hierarchy know that
I can't afford to direct my attention to looking for a cassette to
play?

Attention is a consciousness phenomenon. So the hierarchy is not
involved in
allocating attention.

Since there are observables associated with attention, are these
observables, such as changes in fMRI and eye fixation, not controlled
by the hierarchy?

Moreover, attention changes don't affect which variables you
are controlling. What you are talking about is a resource allocation
problem; your
eyes can't foveate the radio and the cars ahead at the same time. But
you need to
control visual variables to control the radio and the car's
relationship to
traffic. I think very high levels of the hierarchy (the program level)
are
involved in this low level resource allocation. When you drive, you
are probably
controlling for a program like "if traffic dicey then keep eyes on
road, else
allow eyes to move from road for no more than 5 seconds".

Again, conflict arises since slamming on the brakes interferes with
these programs which attempt to continue despite the impending
accidents. Does this conflict persist until the traffic accident clears?

This particular fire hydrant has
always had trouble understanding how

the control hierarchy ‘decides’ which perception to control and, in

particular, how it rapidly shifts from controlling one perception to

controlling another without the delays associated with relying on
upper

levels in the hierarchy to reset reference
levels.
[From Bill Powers (2003.06.27.1407 MDT]

Bruce Gregory (2003.0627.1217)–

I don’t think we control just one variable at a time – or am I
misreading what you mean?

When I look up
from tuning my car radio and see the brake lights of the car in front of
me, how does the perceptual hierarchy know enough to slam on my
brakes?

I don’t understand this way of asking the question. It knows everything
you know, doesn’t it? If a higher system in you has temporarily removed
visual inputs from the distance-keeping control system, that system won’t
perceive any brake lights until you restore the visual inputs. If you
look up after the distance has decreased significantly below the range
you normally maintain, there will be an abnormally large distance error
which should lead to a more energetic braking action than usual. If there
is still plenty of distance to the car ahead, you’ll just brake normally,
won’t you? That’s what I do.

When the traffic is
dicey, how does the perceptual hierarchy know that

I can’t afford to direct my attention to looking for a cassette to

play? It has always seemed from my lowly position that the hierarchy
is

extremely capable but lacking in much sense of what is important at
the

moment.

I liked Rick’s answer to this. Since we’re dealing with a hypothetical
case, it has exactly as much sense as you choose to give it by proposing
details for your model.

The way I would answer this question would be to look at how I know
that I should pay more attention to traffic when it looks bad. When I see
a lot of cars nearby, I watch other cars and the road more, and look only
very briefly at objects or people inside the car, if at all. I think I
tense up a little, too, which moves the motor systems toward the centers
of their operating ranges as well as making the muscles stiffer. It may
be that I also raise the gain of the control systems involved in
controlling spatial relationships. The result is that I control with less
error, counteracting smaller disturbances and continuing to control until
the errors are smaller than they usually are. Why do I do this? Because
the spaces between cars are smaller than usual and my normal control
actions allow the car to get too close to others or to obstacles. The
errors actually get bigger, so I have to raise my control gain to keep
them as small as I want them. It’s not hard to imagine a higher-order
control system that does this sort of thing, is it?

To the extent that
this assessment is not wholly without merit, I offer

the following questions. Could it be that the limbic system
functions

to tell the control hierarchy what is important ‘right now’?
]

I’m not guessing about which parts of the brain are involved. Whatever
brain structure we propose as containing the needed control systems, it
would have to be able to perceive changes in the organism’s relationship
to the environment and detect whatever errors those changes create, and
automatically alter reference signals and/or system parameters in motor
systems in the way required to make the errors smaller. If the limbic
system can do all those things, it may be where that sort of control
takes place. I wouldn’t know.

Further, could it be
that the mechanism employed by the limbic system to convey

this information consists of increasing the gain on some control
loops

and reducing the gain on others?

Certainly. I have long said that higher systems may well act by varying
the parameters of lower systems as well as their reference signals. An
early demo of this effect was offered by Tom Bourbon at my suggestion, 10
or 15 years ago. He set up a model in which a higher-level system
monitored the average absolute value of error signal in a control system,
and varied the output gain in that system to achieve minimum error.
Actually, he set this up as a reorganization task, so the gain variations
were done through a random walk. More recently, I proposed a model in
which an auxiliary control system (whether you should consider it
“higher” or not is debatable) changes the weightings in an
output function in a way that emulates the convolution theorem. It worked
pretty well when embedded in the Little Man model. I called this model
the “artificial cerebellum,” because of some resemblances of
the algorithm to processes known to happen in the cerebellum. Of could
this doesn’t mean that the amygdala could not do something similar.
However, my modeling efforts focus on what kind of control process is
done, which doesn’t depend on guessing which part of the brain does
it.

These questions
do not seem totally implausible on the basis of what little I know about
the wiring of the brain.

They seem plausible to me, too.

Best,

Bill P.

So the process involves conflict
since the higher level goal of getting

to the meeting on time and the still higher level goal of continuing
to

advance my career are still endeavoring to keep my foot on the

accelerator. Why does the foot move from the accelerator to the
brake?

Why doesn’t the foot wind up halfway between the two as a result of
the

conflict?
[From Bill Powers (2003.06.27.1517 MDT)]

Bruce Gregory (2003.0627.1703)–

Since there are
observables associated with attention, are these

observables, such as changes in fMRI and eye fixation, not
controlled

by the hierarchy?

Again, conflict arises
since slamming on the brakes interferes with

these programs which attempt to continue despite the impending

accidents. Does this conflict persist until the traffic accident
clears?

I think that if you put your mind to it, you could answer all these
questions just as well as Rick or I could. They all propose hypothetical
problems, and the only answers expected are deductions or extrapolations
from PCT or HPCT. This game doesn’t work very well if one person is
confined to asking probing questions while the other person has nothing
to do but think up plausible answers. Both can obvious go on doing this
indefinitely. It is much more productive if each person thinks up probing
questions and then tries to answer them as far as possible, and failing
that modifies the theory to correct the problem.

Best,

Bill P.

Bill's response to Bruce G prompts a question on my part:

I think that if you put your mind to it, you could answer all these
questions just as well as Rick or I could. They all propose hypothetical
problems, and the only answers expected are deductions or extrapolations
from PCT or HPCT. This game doesn't work very well if one person is
confined to asking probing questions while the other person has nothing to
do but think up plausible answers. Both can obvious go on doing this
indefinitely. It is much more productive if each person thinks up probing
questions and then tries to answer them as far as possible, and failing
that modifies the theory to correct the problem.

Bill's response seems to me to be predicated on having spotted a pattern, a
"game" as Bill calls it.

Where do "games" such as this fit into the PCT hierarchy. The first thing
that occurred to me was "configuration" but, upon a moment's reflection,
that seems wrong (because configurations, as I understand them, have to do
with "sensed" or "felt" patterns (i.e., what I would ordinarily call
"proprioreceptive" sensations).

Anyway, where does "pattern spotting" fit in the PCT hierarchy?

Regards,

Fred Nickols
"Assistance at a Distance"
Distance Consulting
nickols@safe-t.net
www.nickols.us

Where do “games” such as
this fit into the PCT hierarchy.
The first thing

that occurred to me was “configuration” but, upon a moment’s
reflection,

that seems wrong (because configurations, as I understand them, have to
do

with “sensed” or “felt” patterns (i.e., what I would
ordinarily call

“proprioreceptive” sensations).
Anyway, where does “pattern
spotting” fit in the PCT hierarchy?
[From Bill Powers (2003.06.27.1917 MDT)]

Fred Nichols (2003.06.27)-[-

I’ll go with Rick’s answer: a game is an interaction governed by rules,
which suggests the logic level (9).

I haven’t intended that configurations be confined to kinesthesia. There
are visual configurations (shapes, objects), auditory configurations
(chords and other things), taste configurations, and so forth. A
configuration is a static pattern in any sensory modality.

I think there are patterns at every level. Temporal patterns might belong
at what I call the event level, patterns of symmetry at the relationship
level, patterns of reasoning at the logic level … Is there something
you mean by “spotting patterns” that isn’t covered by
“perceiving?”

Best,

Bill P.

[From Bruce Gregory (2003.0628.0714)]

Bill Powers (2003.06.27.1517 MDT)

I think that if you put your mind to it, you could answer all these questions just as well as Rick or I could. They all propose hypothetical problems, and the only answers expected are deductions or extrapolations from PCT or HPCT. This game doesn't work very well if one person is confined to asking probing questions while the other person has nothing to do but think up plausible answers. Both can obvious go on doing this indefinitely. It is much more productive if each person thinks up probing questions and then tries to answer them as far as possible, and failing that modifies the theory to correct the problem.

Fair enough. Since I told you that HPCT has not helped me to arrive at answers to these questions, I will assume that this is equally true for you and Rick. The model has limitations and ignoring them is unlikely to lead to progress.The most obvious problem form my perspective is that controlling higher level perceptions does little to explain how we are able to control low level perceptions on very short time scales. I will leave it to the real modelers to solve this problem when and if they recognize it as a problem.

[From Rick Marken (2003.06.28.0955)]

Bruce Gregory (2003.0628.0714)--

>Bill Powers (2003.06.27.1517 MDT)

> I think that if you put your mind to it, you could answer all these questions just as well as Rick or I could...

Fair enough. Since I told you that HPCT has not helped me to arrive at answers to these questions, I will assume that this is equally true for you and Rick.

That's like saying that algebra hasn't helped you arrive at the answer the the question "What is the the value of x in 5 = 3x/7"? If HPCT hasn't helped you answer your questions then you have not yet mastered HPCT. Bill was simply pointing out that you could answer your own questions as well as Bill or I or anyone who understands HPCT could if you would "put your mind to it", i.e. try running the HPCT model (as differential equations or as a computer program) and see what happens.

The model has limitations and ignoring them is unlikely to lead to progress.

The model certainly does have limitations -- though it has far fewer limitations than other models of behavior with which I am familiar. But you have to learn the model before you can know what those limitations are.

The most obvious problem form my perspective is that controlling higher level perceptions does little to explain how we are able to control low level perceptions on very short time scales.

Controlling higher level perceptions is not meant to explain how we are able to control lower level perceptions on very short time scales. The model's control of higher level perceptions is meant to explain the observed control of higher level perceptions by people like chess players and disgruntled denizens of CSGNet. The model's control of lower level perceptions is meant to explain control of lower level perceptions, such as the perception of the location of chess pieces or of the shapes of letters typed in an e-mail. The idea that there are different levels of perceptions (higher and lower level) is itself a
theoretical notion meant to explain complex behaviors (such as balancing a broom on your hand) where many simultaneous actions are used as the means of achieving particular ends.

I will leave it to the real modelers to solve this problem when and if they recognize it as a problem.

We can't solve this "problem" any more than we can solve any of your other personal problems. PCT is aimed at solving the problem of explaining observations of control phenomena. PCT explains these observations to the extent that the model produces behavior that is very similar to the observed behavior. The problem you describe (how controlling higher level perceptions can explain how we are able to control low level perceptions on very short time scales) is your own personal problem that comes from your unwillingness (or, perhaps, inability) to learn HPCT. The problem you describe is not a problem for those who
understand HPCT because we know that controlling higher level perceptions is not meant to explain how we are able to control low level perceptions on very short time scales.

Best regards

Rick

···

--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313

Fair enough. Since I told you that
HPCT has not helped me to arrive at answers to these questions, I will
assume that this is equally true for you and
Rick.
[From Bill Powers (2003.06.28.1016 MDT)]

Bruce Gregory (2003.0628.0714)–

Not at all. The particular problem you posed – why we jam on the brakes
after having looked away from the road for a while, and see a car’s brake
lights on ahead of us – seems particularly easy to deal with. But it
will not help you to understand HPCT if some one else sees a solution and
supplies it for you.

HPCT doesn’t help you arrive at answers to questions any more than
algebra does. Some assistance is required from your brain. A theory is
not a list of answers to questions. It’s a tool that can be used to work
your way to answers.

Your assumption that there must be a conflict between higher-order goals
(getting to a meeting) and stopping forward progress (braking) leads to a
prediction contrary to observation (we do not usually freeze halfway
between braking and not braking), so it must be incorrect. For example,
the idea of getting to the meeting might include not colliding with
anything, which would probably at least make you late. Or it is possible
that the braking incident is over before the higher system can come
anywhere near doing anything about it one way or the other. True conflict
between different levels of control is unlikely. Do you understand how
there can be one control system that tries to move a person to a goal,
and another that avoids collisions, without any conflict? Study the
“Crowd” simulation – that’s what happens there.

It’s easy to think of ways to modify your assumptions and make the model
behave correctly. You didn’t offer any alternative assumptions and
discuss how they would modify the predictions you make, so the appearance
is that you stopped at the first sign of difficulty and decided that the
problem was with the theory and not your assumptions. It is far more
likely, in my opinion, that the assumptions are at fault, at least with
respect to finding assumptions that will, when plugged into the theory,
lead to a correct prediction.

I don’t mean to imply that once we find a set of assumptions and an
instance of the model that solve the problems, we should just accept the
solution as being correct. All we have at that point is a seemingly
plausible explanation. The next step is to verify the assumptions and
convert the verbal approximations into a simulation that will really test
it. That is not going to happen by means of further discussion on the
internet, or by waiting passively for HPCT to tell us what to do
next.

The model has
limitations and ignoring them is unlikely to lead to
progress.

What limitations? So far the only limitation I can see is on the amount
of effort you’re willing to put into finding a solution. Your interest
seems to be mainly in creating difficulties, or at least what you see at
first glance to be difficulties. I don’t want to treat your objections
any less seriously than they deserve, but so far you haven’t posed any
serious problems or objections, and you’ve proposed no solutions at
all.

If I sound a little ticked off, you’re reading me correctly.

Best,

Bill P.

The most obvious
problem form my perspective is that controlling higher level perceptions
does little to explain how we are able to control low level perceptions
on very short time scales. I will leave it to the real modelers to
solve this problem when and if they recognize it as a
problem.

But the answer to that “problem” is inherent in HPCT. In the
two-part paper by Clark, MacFarland, and me there is described a
“portable demonstrator” which includes a demonstration of how a
lower-order system can correct an error before a higher-level
system can tell it not to. That paper was published 43 years ago, and
I’ve been pointing out the same feature of hierarchical control ever
since. If you’ve missed it every time, is that my fault?

[From Bruce Gregory (2003.0628.1440)]

[From Rick Marken (2003.06.28.0955]

Bruce Gregory (2003.0628.0714)--

Bill Powers (2003.06.27.1517 MDT)

I think that if you put your mind to it, you could answer all these
questions just as well as Rick or I could...

Fair enough. Since I told you that HPCT has not helped me to arrive
at answers to these questions, I will assume that this is equally
true for you and Rick.

That's like saying that algebra hasn't helped you arrive at the answer
the the question "What is the the value of x in 5 = 3x/7"? If HPCT
hasn't helped you answer your questions then you have not yet mastered
HPCT. Bill was simply pointing out that you could answer your own
questions as well as Bill or I or anyone who understands HPCT could if
you would "put your mind to it", i.e. try running the HPCT model (as
differential equations or as a computer program) and see what happens.

The model has limitations and ignoring them is unlikely to lead to
progress.

The model certainly does have limitations -- though it has far fewer
limitations than other models of behavior with which I am familiar.
But you have to learn the model before you can know what those
limitations are.

When I ask questions I am told that I can answer them as well as the
experts. Then I am told that I cannot answer them because I do not
understand the model. The PCT approach to pedagogy seems to leave
something to be desired.

We can't solve this "problem" any more than we can solve any of your
other personal problems.

  I am not asking you to solve my personal problems. I am asking you to
explain how the model you love so much resolves an obvious problem. I
am relieved to hear that it doesn't. That was my conclusion as well. As
long as we don't deal with problems the model cannot solve, I'm sure we
will have no differences of opinion. Can I safely conclude that the
famous spreadsheet analysis described in _Mind Readings_ is therefore
irrelevant in this case?

PCT is aimed at solving the problem of explaining observations of
control phenomena. PCT explains these observations to the extent that
the model produces behavior that is very similar to the observed
behavior. The problem you describe (how controlling higher level
perceptions can explain how we are able to control low level
perceptions on very short time scales) is your own personal problem
that comes from your unwillingness (or, perhaps, inability) to learn
HPCT.

Have you noticed how often you fall back on this avoidance mechanism?
If I can understand general relativity and quantum mechanics, have you
ever wondered why I can't understand PCT? Evoking some Freudian
explanation is a good way to avoid confronting the issue. It must be
because I am 'resisting' some unpleasant truth.

The problem you describe is not a problem for those who
understand HPCT because we know that controlling higher level
perceptions is not meant to explain how we are able to control low
level perceptions on very short time scales.

I knew there must be a good reason why you are ignoring the issue. It's
a non-issue. Like the WMD in Iraq.

[From Bruce Gregory (2003.0628.1600)]

Bill Powers (2003.06.28.1016 MDT)

Bruce Gregory (2003.0628.0714)--

Fair enough. Since I told you that HPCT has not helped me to arrive at answers to these questions, I will assume that this is equally true for you and Rick.

Not at all. The particular problem you posed -- why we jam on the brakes after having looked away from the road for a while, and see a car's brake lights on ahead of us -- seems particularly easy to deal with. But it will not help you to understand HPCT if some one else sees a solution and supplies it for you.

Apparently little I do helps me to understand HPCT. I'm sure there's a perfectly valid explanation for this. Perhaps its my unwillingness to see the truth because on an unresolved Oedipal complex.

HPCT doesn't help you arrive at answers to questions any more than algebra does. Some assistance is required from your brain. A theory is not a list of answers to questions. It's a tool that can be used to work your way to answers.

It's amazing that I didn't learn this from physics, isn't it? Thanks for pointing it out to me.

Your assumption that there must be a conflict between higher-order goals (getting to a meeting) and stopping forward progress (braking) leads to a prediction contrary to observation (we do not usually freeze halfway between braking and not braking), so it must be incorrect. For example, the idea of getting to the meeting might include not colliding with anything, which would probably at least make you late.

How doe the perception "getting to the meeting on time" 'include' 'not colliding with anything'. Does the former also 'include' "not eating anything that might give me the runs"? Clearly testing for so elaborate a controlled variable would tax the patience of Job.

Or it is possible that the braking incident is over before the higher system can come anywhere near doing anything about it one way or the other. True conflict between different levels of control is unlikely.

I do not understand the last statement. Conflict, I thought, results when two higher level systems are trying to maintain the same lower level system at difference reference levels. My only question was why does the higher level perception being controlled not continue to attempt to keep my foot on the accelerator?

Do you understand how there can be one control system that tries to move a person to a goal, and another that avoids collisions, without any conflict? Study the "Crowd" simulation -- that's what happens there.

It's easy to think of ways to modify your assumptions and make the model behave correctly. You didn't offer any alternative assumptions and discuss how they would modify the predictions you make, so the appearance is that you stopped at the first sign of difficulty and decided that the problem was with the theory and not your assumptions. It is far more likely, in my opinion, that the assumptions are at fault, at least with respect to finding assumptions that will, when plugged into the theory, lead to a correct prediction.

I asked for alternative assumptions and was told, apparently mistakenly, that I was just as capable as the experts of coming up with them. Then I was told that either wouldn't or could understand the model. As I suggested to Rick, your pedagogy might leave a wee bit to be desired. Have you ever considered simply answering questions? Just a suggestion.

I don't mean to imply that once we find a set of assumptions and an instance of the model that solve the problems, we should just accept the solution as being correct. All we have at that point is a seemingly plausible explanation. The next step is to verify the assumptions and convert the verbal approximations into a simulation that will really test it. That is not going to happen by means of further discussion on the internet, or by waiting passively for HPCT to tell us what to do next.

I'm sure this a valid point, but its relevance to my questions escapes me.

The model has limitations and ignoring them is unlikely to lead to progress.

What limitations? So far the only limitation I can see is on the amount of effort you're willing to put into finding a solution. Your interest seems to be mainly in creating difficulties, or at least what you see at first glance to be difficulties. I don't want to treat your objections any less seriously than they deserve, but so far you haven't posed any serious problems or objections, and you've proposed no solutions at all.

I'm sure this is a valid statement of your experience. It seems that only proposals that fall within your way of thinking are even noticed by you. It's not that you ignore them, I suspect that you literally do not see them, in much the same way that those who constructed the case for invading Iraq literally could not see any evidence that conflicted with their foregone conclusions. It does make for a very focussed approach, but one that does have certain limitations.

If I sound a little ticked off, you're reading me correctly.

O.K. I won't ask any more questions and you won't get ticked off. Sorry to have imposed on you. I realize that CSGnet has a very different purpose than the one I assumed it did. I should have realized this from the fact that so few join the list and even fewer post.

[From Bruce Gregory (2003.0628.1600)]

Bill Powers (2003.06.28.1016 MDT)

Bruce Gregory (2003.0628.0714)--

Fair enough. Since I told you that HPCT has not helped me to arrive at
answers to these questions, I will assume that this is equally true for
you and Rick.

Not at all. The particular problem you posed -- why we jam on the brakes
after having looked away from the road for a while, and see a car's brake
lights on ahead of us -- seems particularly easy to deal with. But it
will not help you to understand HPCT if some one else sees a solution and
supplies it for you.

Apparently little I do helps me to understand HPCT. I'm sure there's a
perfectly valid explanation for this. Perhaps its my unwillingness to see
the truth because on an unresolved Oedipal complex.

No, I don't think that is the explanation. Tell you what. Why don't you
outline the way you arrived at the conclusion that there had to be a
conflict between two control systems in the situation you have set up, and
I'll try to help you work out some possible answers. This won't settle the
issue, but at least it might show how one sets up potential experiments.
Start from scratch by describing the exact situation you have in mind, the
principles from HPCT you are applying, and the preliminary conclusions you
have reached (including the way you reached them, if practicable).

Note that I am taking your request at face value and not assuming that
you're just pulling my chain.

Best,

Bill P.

[From Rick Marken (2003.06.28.1645)]

Bruce Gregory (2003.0628.1440)--

When I ask questions I am told that I can answer them as well as the
experts. Then I am told that I cannot answer them because I do not
understand the model.

This must be very difficult for you. Have you ever thought of trying to
learn the model?

I am not asking you to solve my personal problems. I am asking you to
explain how the model you love so much resolves an obvious problem.

It's not obvious to me. What you present as a problem is a non-problem, as
far as I can tell. The "problem" you presented -- why we jam on the brakes
after having looked away from the road for a while and see a car's brake
lights on ahead of us -- is easy to deal with in PCT. But, as Bill said, it
will not help you to understand HPCT if someone else supplies the solution
for you. You've been on the net for years and we (Bill and I) have
explained how the model handles this kind of control situation and many
others that you find problematic. It's clear that you cannot learn how PCT
explains this kind of behavior by having us tell you over and over again
how it's done. You really have to sit down now and work through the
equations (or programs) yourself. If you find that you can't make the
equations handle the situation you find to be problematic, then post your
work to the net and we'll help you out, if we can.

Can I safely conclude that the
famous spreadsheet analysis described in _Mind Readings_ is therefore
irrelevant in this case?

It's precisely relevant.

If I can understand general relativity and quantum mechanics, have you
ever wondered why I can't understand PCT?

You bet I have. My guess is that your understanding of general relativity
and quantum mechanics must be as good as your understanding of PCT.

Best regards

Rick

···

--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313

[From Bruce Gregory (2003.0628.2125)]

Rick Marken (2003.06.28.1645)

You bet I have. My guess is that your understanding of general relativity
and quantum mechanics must be as good as your understanding of PCT.

Should I take that as a "fuck you"? I find it difficult to interpret in
any other way.

···

--
Bruce Gregory lives with the poet and painter Gray Jacobik in the future
Canadian Province of New England.

www.joincanadanow.org

[From Bruce Gregory (2003.0628.2136)]

Bill Powers wrote:

No, I don't think that is the explanation. Tell you what. Why don't you
outline the way you arrived at the conclusion that there had to be a
conflict between two control systems in the situation you have set up, and
I'll try to help you work out some possible answers. This won't settle the
issue, but at least it might show how one sets up potential experiments.
Start from scratch by describing the exact situation you have in mind, the
principles from HPCT you are applying, and the preliminary conclusions you
have reached (including the way you reached them, if practicable).

I'm afraid I can't do any better than my last two posts. I suggest that
we drop the issue.

···

--
Bruce Gregory lives with the poet and painter Gray Jacobik in the future
Canadian Province of New England.

www.joincanadanow.org

[From Rick Marken (2003.06.28.2150)]

Bruce Gregory (2003.0628.2136)--

Bill Powers wrote:

> No, I don't think that is the explanation. Tell you what. Why don't you
> outline the way you arrived at the conclusion that there had to be a
> conflict between two control systems in the situation you have set up, and
> I'll try to help you work out some possible answers. This won't settle the
> issue, but at least it might show how one sets up potential experiments.
> Start from scratch by describing the exact situation you have in mind, the
> principles from HPCT you are applying, and the preliminary conclusions you
> have reached (including the way you reached them, if practicable).

I'm afraid I can't do any better than my last two posts.

Why, then, don't you re-post those posts or, better, just the relevant
passages from them that outline the way you arrived at the conclusion that
there had to be a conflict between two control systems in the situation you
have set up. Then we can go from there.

I suggest that we drop the issue.

Why? Seems like this is a nice opportunity to show how the model can (or, if
you are right, cannot) be used to explain a particular example of behavior.

Best

Rick

···

--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313

I’m afraid I can’t do any better
than my last two posts. I suggest that we drop the
issue.
[From Bill Powers (2003.06.29.0637 MDT)]

Bruce Gregory (2003.0628.2136)–

As you wish. I hope that you won’t be referring to this issue in the
future as a case that HPCT can’t handle. If you’re tempted to do that, my
offer still stands: spell out the problem as you see it, and whatever
solutions you attempted to find, and I’ll try to work with you to find a
solution if a reasonable one exists. I don’t think there’s anything to be
gained by my offering possible solutions if (a) I don’t know exactly what
the problem is as you see it, and (b) you end up with an answer but no
understanding of how I arrived it (so I will have to keep answering the
same question over and over in different disguises). I think that if you
tried to state the problem clearly and as completely as possible, you
might see something that explains the difficulty, or see a lead to
answering your own questions. That kind of answer will stick with you for
a long time.

[Mary says: “At the blah-blah-blah level, HPCT is no better than any
other theory”]

It’s the same old story. Give a man a fish and he’ll eat for a day. Teach
him to fish, however, …

I don’t agree with Rick that the spreadsheet model, or any simulation,
could be directly applied here. Most likely he didn’t mean that. We can’t
write the equations for the sorts of control systems we’re talking about.
There are some useful principles there, such as the one saying that
lower-order goals are set so as to help satisfy many higher goals at the
same time, but that’s about as close as we can get to equations.

It is possible to see in the spreadsheet (or equivalent) model that
interactions among control systems at the same level, due to sharing of
lower-order systems, do not amount to conflict unless they require that
the higher systems produce their maximum possible outputs (which puts an
end to controlling because the outputs can no longer vary in response to
disturbances). Short of that extreme situation, what we have is a group
of control systems keeping their own error signals as small as possible
with the result that that total error across all systems is as low as it
can be given the current system parameters and environment. That is
the normal state of the hierarchy: not zero error, but as little error as
possible.

Add to this the principle that higher-order systems tell lower ones what
to sense (by adjusting their reference signals), while the lower systems
act autonomously and rapidly to bring their own perceptions to the
requested levels and counteract disturbances while keeping the
perceptions at those (constant or varying) levels. It might also be
appropriate to suppose that higher systems can adjust parameters like
gain in those lower systems, although those higher systems would be
concerned with different controlled variables.

Those are the main principles I would apply in arriving at an answer to
the questions you raised. There may be others that I would be reminded of
while working out a solution. Also, you might find it necessary to add
assumptions: for example, what has to be done, when the input is
completely removed from a control system (by looking away), to keep a
large error signal from appearing, or from causing an abnormally large
output?

HPCT itself supplies only the main framework of principles. Each specific
application introduces unique detailed considerations that have to be
worked out. This is true of any theory when it’s applied to a specific
situation. That’s what I meant when I said that HPCT can’t provide
answers without assistance from your brain, any more than algebra
can.

As you can see, it’s hard for me to accept the proposal that we just drop
the issue. I’m willing to help, but I don’t think it’s unreasonable to
expect that a learner will put forth as much effort as needed. If it’s OK
with you just to drop the subject, how important was the question in the
first place?

Best,

Bill P.