questions

[William_T_Powers2]

[From Bill Powers (940714.1645 MDT)]

Paul George (940714.1140) --

I'm glad Rick asked about E, what we call "the disturbance." The
disturbance represents the class of independent environmental
variables that can act directly on the quantity you're trying to
control, independently of the outputs of your control system. Like
someone pushing on your elbow while you're aiming a pistol.

Your reply to Rick didn't clear up the point:

It is some facit of the process that we cannot directly sense, but
must rather infer via F.

This would partly fit, except that in general there is no way to
work backward from the state of F (actually, H) to the cause of a
disturbance, that is, to E. All the control system knows is that H
changed when the reference signal A did not change. In a properly
designed control system, the change will be strongly resisted
because of the closed-loop action. It isn't necessary to know E. But
a complete model of a control process has to include E; otherwise
you'll be designing for a disturbance-free environment instead of a
real one.

I know of no formal name for it.

Don't tell me I have invented something. Actually, I doubt it. I
learned about disturbances, after all, from reading control-system
texts. Maybe the subject has been dropped since the 1950s. If so, I
might be dubious about trusting my safety to any control system
designed since then.

One of the sources of misunderstanding we have had with control
engineers (especially concerning "open-loop control") is that many
of them seem to assume, in their designs, that all sources of
disturbance can be accounted for, so they can be incorporated into a
"world model." But that is a very unrealistic approach to designing
control systems, unless they are so simple and transparent that you
can anticipate everything the environment might do to them. Well,
that point aside, it is certainly not true of organisms that they
can identify all possible causes of disturbances and prepare to meet
them. MOST disturbances are known only in terms of an unexpected and
persistent change in the controlled variable. It takes a closed-loop
system to maintain control when that happens.

So am I to take it that in your process-control models, you do NOT
make an explicit provision for unpredictable disturbances to affect
the controlled variable?

Since we often cannot sense the
perception we really want to control, we often have to provide an
F5 and/or F3 inverse within F4 or F1 (or more usually between, this
is the 'world model' or 'mirrored object'.

We never assume that either F3 or F5 (function linking a disturbing
variable to the controlled variable F) is known to the control
system; the organic system has to be able to maintain good control
without knowing the form of either function.

It sounds odd to me to say that "we often cannot sense the
perception we really want to control." But then I realize that you
mean we, the engineers, don't have available a sensor that can be
put into the control system to let IT sense the variable we, the
engineers, really want to control.

I should think, though, that it would often be possible to sense
component variables and compute the state of the variable you really
want to control. Isn't that one solution? That counts as sensing it,
in PCT.

This is not a problem in modeling organisms. There is no engineer.
If the organism doesn't sense a variable, it can't be controlled by
that organism.

By convention and practice the system is seen as controlling D in
order to provide control of F (the 'process centric view').

Maybe I'll accept that. But I'd have to believe that the
conventional view is a hierarchical control view, which would
surprise me. A more accurate way of expressing this idea as it is
usually carried out is to say that the control system VARIES D in
order to control F. To control something means to bring it to a
reference state and keep it there, in my dictionary. If you're
driving a car, you can't simultaneously control the steering wheel
angle and control the car's position on the road. Not, that is,
unless you mean this hierarchically: you send a VARYING reference
signal to the wheel-angle control system, which controls wheel angle
to make it match the reference signal, and thus you control the
position of the car.

The reason you can't speak of controlling D is that D is determined
just as much by disturbances E as it is by the reference signal A.
You can't bring D to a predetermined state, because that may be the
wrong state for counteracting a disturbance. The output action of
the control system must NOT be controlled in some preferred state;
it has to be free to vary as required to counteract the effects of
disturbances. Just imagine trying to drive while holding the
steering wheel at your favorite angle.

You can see why disturbances are considered important in PCT.

Recall that in our system we are often triggering an industrial
machine (often with control capabilities) that may do a fairly
complex series of things to the process under control. They respond
to a fixed set of signal's that are usually interpreted as commands
or instructions.

"Our system?" Are you speaking of the organism, or about the process
control systems you-all design? If you're speaking about the human

system, I would dispute the claim that ANYTHING is done open-loop.
There are always disturbances. But in industrial control systems,
the system design is whatever the engineers decide it should be. If
they're confident that the process will always produce exactly the
result that the command specifies, without any feedback to check
that what happened is what was wanted, and without any means of
altering the command if the process strays off track, then they are
braver souls than I am. But it's their necks.

Don't tell me. Management says that sensors are too expensive.

I notice that NASA systems are quite variable in this regard.
Sometimes they are fully fed-back so that the result of every
command, even every switch-transition, is immediately observed, and
the command can be changed or canceled if the wrong result even
starts to occur (or if the contacts don't close). Others are
designed by engineers who seem never to have heard the term
"disturbance." When they send commands to those systems, Mission
Control sends an astronaut to stand by and observe the actual
result, and report it back immediately. Good thing, too.

Sometimes B may be a a program which is downloaded to replace an
F3. Similarly a F5 could represent a machine not under the direct
control of the control system (I know the phraseology flies in the
face of the PCT concept of what is controlled).

I don't care about phraseology as long as I can figure out what
you're talking about. I suppose that a control system could be built
that senses which program is running, which program should be
running, and if there is an error downloads a replacement program to
a lower-level system. So far we haven't done any experiments
complicated enough to call for that model. But there's a place
reserved for it, at the program and principle levels in HPCT.

Actually, since engineers aren't constrained by any evolutionary or
survival considerations, they can design systems any way they want,
as long as they do the job at hand. They aren't designing general-
purpose control systems like organisms. They don't have to worry
about ad-hoc patches to an existing design in terms of how it is
going to affect the whole system in a different situation. If it
works, screw the cover down and ship it.

I've been intermittently seeing the TEMPUS downlink video from STS-
65. This experiment includes a levitation device for "containerless
processing" of small spherical alloy samples. Maybe there's some
basic problem about stabilizing the sample spheres, but after five
days of watching I'm going crazy from seeing that sample jiggling
and bouncing around with every little disturbance from the
spacecraft, or from heating or from evaporation or from who knows
what. Jiggle, jiggle, jiggle. At least three samples that I know of
have hit the cage while molten and stuck to it, ruined. And then the
PI says "Thanks, Columbia, it's nice and stable now." Jiggle,
jiggle, jiggle. Jiggle, jiggle, jiggle. I'm probably doing the
designer a terrible injustice; the control problem might be all but
unsolvable. But I want to grab the phone and yell at the PI, "For

God's sake, haven't you ever seen a REAL control system?" I have
enough hubris to think that if someone asked me to stabilize a
levitated sphere, they'd think it was nailed in place, like the
parrot's feet. Sometimes being old and retired is hell.

···

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

Bill P.

[Paul_George]

[Paul George 940715 11:45]

[From Bill Powers (940714.1645 MDT)]
Don't tell me I have invented something. Actually, I doubt it. I
learned about disturbances, after all, from reading control-system
texts. Maybe the subject has been dropped since the 1950s. If so, I
might be dubious about trusting my safety to any control system
designed since then.

There probably is a formal name, the problem is that I am a software engineer
by training, not a control engineer. I work on development processes and
techniques not process control software. Unfortunately, most people developing
control systems these days aren't trained in control theory. Bailey Controls
used to have a kind of apprentice program for all engineers for that purpose,
but it was dropped about 15 years ago. Fear and trembling is appropriate. Read
the RISKS forum (ACM SIGSOFT Transactions or comp.risks. is also a mailing list
if you are interested).

One of the sources of misunderstanding we have had with control
engineers (especially concerning "open-loop control") is that many
of them seem to assume, in their designs, that all sources of
disturbance can be accounted for, so they can be incorporated into a
"world model."...
So am I to take it that in your process-control models, you do NOT
make an explicit provision for unpredictable disturbances to affect
the controlled variable?

Yup. We are forced to try to anticipate all possible problems and deal with
them. In fact for safety critical systems we are legally required to. That is
why requirements analysis and design are so difficult. If something
unanticipated does occur and something bad happens, we can be held legally
accountable if we 'knew or should have known' that it could ocurr. The big
problem is that if an E (disturbance) ocurrs we usually must have a D to affect
it (making it another F). Disturbances (E) which may be corrected via D are not
a problem. Our task is to insure we actually have all the Fs and D's that are
required to keep the process under control. Fortunately the task is eased by
having a human operator as part of the System. They are a bit more flexible and
don't just 'fly IFR' (instrument flight rules).

But then I realize that you mean we, the engineers, don't have available a
sensor that can be put into the control system to let IT sense the variable
we, the engineers, really want to control.

Or more likely, the customer wouldn't pay for it, or there is no place to put
it. Sensors and wireing are expensive and kept to a minimum. One of the reasons
that continious control isn't used is that message passing networks are used
for communication between controllers and instruments to minimize cableing.
Point to point wire connections, nerve style, would be prohibitivly expensive.
There are also transmission problems with analog control signals and long
cables. Practicality raises its ugly head.

By convention and practice the system is seen as controlling D in
order to provide control of F (the 'process centric view').

Maybe I'll accept that. But I'd have to believe that the
conventional view is a hierarchical control view, which would
surprise me.

These days it is, at least in companies who build plant control systems.
Control engineers working at the PLC loop level is probably what you are
familiar with. Hierarchical systems emerged in the 70's. The car example is
exactly how we usually design it.

"Our system?" Are you speaking of the organism, or about the process
control systems you-all design?

My whole post dealt with process control systems, or more acurately
hierarchical distributed plant control systems. Organisms work differently,
though system architectures are patterned on various understandings of how
organisms function.

Don't tell me Management says that sensors are too expensive.
I notice that NASA systems are quite variable in this regard.

NASA doesn't _sell_ control systems. Their contractors (system developers) are
in the business of selling engineering hours. In the industrial world, system
proposals are evaluated mostly on price. The number of I/O points and cable
lengths are the major cost drivers.

I suppose that a control system could be built
that senses which program is running, which program should be
running, and if there is an error downloads a replacement program to
a lower-level system.

That is actually a normal situation. If a processor crashes, you often need to
reload the software. But I meant something more akin to learning. You
effectively replace the old control node (or HPCS) with a new one that is
better for handling the current situation, or is more efficient (say a
different F2 or F4). In addition, since (due to bad design) the H values are
often hard coded into the software, to change it you need to download a new
version.

Later
Paul

[papanicolaou]

From Tom Bourbon [920922 -- 16:45]

Re: *really* naive questions; (from: eric harnden 920922.0940)

mr marken, in his letter to estes, reasserts the essential
argument that study should be directed not toward behavioral
outputs but toward controlled inputs. further, he indicates that
the reaction to this and other assertions is 'yeah, so what...
that's obvious', and goes on to say that it doesn't seem like its
obvious, as demonstrated by the experimental modes of mainstream
behavioral researchers.

      The reaction Rick described occurs very often. We have seen
it numerous times in the rejections of manuscripts in which we
report the results of modeling control behavior with PCT. The most
common form it takes is, "But we already know that." Next most
common is, "But *everyone* knows that." I guess psychologists who
use the second form are being generous and suggesting that even
non-psychologists understand what is going on. Truth is, most
"common folks" DO understand this -- but hardly any psychologists
do. The research methods used in behavioral science make that fact
perfectly clear.

now, i was under the impression that behavioral studies often
involved the presentation of stimulus to an organism, and an
observation of response.

      Your impression is correct. There is your proof that they
don't understand the phenomenon of control. They believe there are
"stimuli," acting as discrete independent antecedent variables,
followed by discrete dependent consequences called "responses." Do
you know of places in the behavioral literature where scientists
clearly, consistently, and unambiguously state that what most
people call "stimuli" are really disturbances (by any name they
care to call them) that affect variables controlled by the
individual who is the object of study? And do those same
scientists clearly, consistently, and unambiguously state that what
most people call "responses" are really actions that nearly
perfectly eliminate the effects of disturbances to controlled
variables? My questions are not meant to be trivial, nor are they
"put downs" of your post. I do not know of statements like those
I mentioned -- except in the PCT literature -- and I am willing to
admit that I might have missed something.

i was further under the impression that these stimuli were
presented under a not entirely unfounded set of assumptions of
what, in the local terminology, the organism is controlling for.
please explain to me why it is invalid for a mainstream
behaviorist to say 'yes, well of course the organism is going to
avoid pain, and prefer food. in this supposedly new jargon, i am
being told what ialready know: that it is controlling for lack of
pain and hunger.

      Here, your impression is far from what I read in the
behavioristic literature. I know of no place where traditional
behavioral scientists, especially those who call themselves
behaviorists, who speak of their animals "controlling for"
anything. The entire language of the field is dominated by claims
that behavior IS CONTROLLED BY environmental stimuli. Many people
who read the behavioral literature ASSUME, as you seem to have done
and as I often want to do, that scientists MUST think of their
animals as controlling -- as seeking and avoiding. But those terms
are explicitly proscribed in the more radical forms of behaviorism.
Skinner, to the night before he died, rejected the use of "states"
like pain or hunger as explanations for behavior. YOU already know
organisms control for their experiences -- don't be too generous in
attributing the same level of understanding to "real" behavioral
scientists. Make them prove it.
      My experiences (bad experiences at the hands of reviewers and
editors) have taught me that unless and until a person demonstrates
that her or his theory or model of behavior explicitly addresses
the issues I just mentioned, I should not assume the person "knows"
about the phenomenon of control and about control systems.

what interests me is the characterization of its outputs, since it
is the nature of these that ultimately affects itself, its
environment, and other organisms.' now, i understand that this
demonstrates, on my part, a basic ignorance of both classical
psychology, and (so far at least) pct. but i am hoping that, by
framing the question so baldly, i can elicit a response strong
enough to clear a few of my cobwebs.

      There is nothing wrong with characterizing outputs. It is
just that when you discover they are part of the phenomenon of
control, you realize the way they are characterized in most of the
behavioral sciences is wrong. W - R - O - N - G. If you look at
them as caused by antecedents, and ignore the fact that they
control inputs, you miss the whole point of what most actions are
about. Of course hardly anyone involved in traditional behavioral
science recognizes that fact, or else they wouldn't stay involved.
When you find outputs that control perceived inputs, you discover
that none of the traditional "laws" apply. That claim does not go
down well with reviewers and editors, even when you back it up with
solid data.
      Believe me, when one manuscript goes through repeated
rejections, all accompanied by claims that, "We already know that,"
but all revealing ignorance of the phenomenon of control, it is
easy -- very easy -- to think, "Why bother?"

Tom Bourbon

MEG Laboratory PAPANICOLAOU@UTMBEACH.BITNET
1528 Postoffice Street PAPANICOLAOU@BEACH.UTMB.EDU
Galveston, TX 77550 PHONE (409) 763-6325
USA FAX (409) 762-9961

[Peter_J_Burke]

[Peter Burke 261095.0940]

[From Bill Powers (951025.0820 MDT)]
What's come over you, Peter? Was it reading Kelso? Or is it that you
never did understand the Test correctly? If you want the full
description, you can look it up in Runkel's _Casting Nets and Testing
Specimens_

I guess what has come over me is that I am playing more the role of
scientist than the role of PCT theorist, and I do seem to have stirred up a
hornets nest. I have not seen such defensiveness in a long time. Yes,
Kelso's book is directly behind some of my questions, but only because it
reminded me of a number of questions that have had while working with PCT
theory over the past ten or twelve years. Bill, I had not seen Runkel's
book, and I thank you for the reference. I ordered it yesterday from an
on-line service called Amazon.com Books (http://www.amazon.com/) an amazing
place which claims over a million titles. I also thank you for carefully
laying out all the steps in a more elaborated version of "the test" than I
was familiar with.

There are a number of questions that have always puzzelled me, for which I
have not seen adequate answers (it is quite possible that they exist and I
am not familiar with them). For example, how is it that the perceptions side
of an ECU exactly corresponds to the standard in dimensionality and units?
That is, why do we perceive the very thing that needs to be controlled
rather than something else? Another example is what is the nature of a
standard that (in a sense) remains intact when we substitute one (lower
level) ECU for another, as when I shut the door with my hand or my foot,
with my eyes open or with my eyes closed (i.e., varying both perceptions and
behavior)? These are the kinds of theoretical or meta-theoretical questions
I am trying to raise. And I don't expect condescention or put-downs or
defensiveness. I expect professional, intellectual and honest answers or
disucssion. I am certainly not attacking PCT or anyone associated with it.
Perhaps that has happened so often that occasionally honest questions are
seen as attacks. I am attacking only in the sense of a scientist who is not
afraid to ask potentially embarrassing questions even of himself. I don't
know that my questions are embarrassing, mayby only to me of my own
ignorance, but I am willing to learn.
Peter

···

------------------------------------------------------------------------
Peter J. Burke Phone: 509/335-3249
Sociology Fax: 509/335-6419
Washington State University
Pullman, WA 99164-4020 E-mail: burkep@unicorn.it.wsu.edu
------------------------------------------------------------------------