Assuming control; reorganization

[From Bill Powers (940410.0930 MDT)]

David Goldstein (940408) --

Pardon me for horning in on questions addressed to someone else, but
how can you specify that only one person is to answer without
explicitly saying so? And even if you do, who is going to pay
attention to you?

Rick, when a control system is not in the control mode, are you
saying that it is still controlling?

If you say yes( because somehow the selection of mode of
functioning must occur), who or what part of the person does
this? I assume that your answer will somehow involve the
reorganization system properties ...

If something is done, it is done for a purpose -- unless it isn't.
But assuming that this mode-switching does have a purpose, the
purpose can only be that of some higher-level system. The system
suffering the switching doesn't have the ability to do that
switching itself. The general principle is that if you see something
like mode-switching happening inside the hierarchy, it is being done
TO something BY something else, and the purpose of doing it is in
the "something else." So if a control system is switched from normal
control mode to imagination mode, this must be done by some higher
system for its own purposes, and is not decided upon by the system
that finds itself in either mode. Mode-switching of a lower-level
system represents the normal control operation of a higher-level
system.

So a system that is planning some operation might, as a normal part
of the planning, switch lower systems into the imagination mode.
This makes all reference signals that would normally implement the
plan by being send to lower systems turn into perceptions
immediately, with no need for physical action. "First, I'll lift
that car and put it on top of the other car; then I can sweep the
floor where it was."

Obviously, the success of planning depends on how realistically one
can imagine, and that in turn depends on the level at which
imagination takes place. If it takes place at a very high level, it
is easy to imagine making impossible things happen. It's true that
if you lifted one car out of the way, you could then sweep the floor
where it was. Stated as a purely verbal or logical solution, this
would indeed solve the problem. But this abstract solution suffers
the disadvantage that you can't lift cars -- when you switched to
the control mode you would just pull a bunch of muscles loose.

By imagining at a lower level, you would find that you have no
experiences of lifting cars, so when you try to imagine actually
doing that, you would come up blank. The planning system would have
to come up with a different solution. When you imagine at lower
levels, you put constraints on what can happen at higher levels even
in imagination. This is why pure mathematical reasoning and verbal
reasoning tend to be impractical: the means of getting to the goal
makes logical or grammatical sense but can't actually happen in the
"real" world -- that is, the world of experience that includes _all_
levels. That's why, in English, "throw the horse over the fence some
hay" is funny.

All this is just by way of illustrating that you don't need to
invoke the reorganizing system as the first explanation of mode-
switching. The above discussion lacks some explanatory details, but
shows how mode-switching could be worked into the normal operation
of the hierarchy without any change of organization. We just have to
imagine control systems that act on lower systems in ways other than
by adjusting reference signals.

I think we should reserve the reorganization explanation for those
situations in which we really can't think of any systematic
explanation for what's going on.

Rick, don't we have to empirically demonstrate that a person
who engages in action X in situation Y is controlling some
perceptual variable?

I second this motion, and call Bill Leach's attention to it, too. In
thinking of tentative explanations for phenomena, it's one thing to
propose that a control process is responsible, but another to prove
it. If we just decide that everything is control, we'll be doing
exactly what psychologists and others have so irritatingly done:
assume a theory and then without proof assume it's true in
explaining everything. Skinner decided that the environment had to
be the ultimate cause of any behavior, and without proof of that
assertion he then proceeded to reinterpret everything as if it were
true. This is how you found religions, not sciences.

When we say that control is involved in any behavior, we should at
least have some supporting evidence in mind, such as resistance to
disturbance or lack of a relationship between actions and repeating
consequences -- the main two phenomena that make us think of
control. If you just use high-level imagination, you can make any
process look like control even if it isn't. Evidence is low-level
imagination, based on experience, that rules out other explanations
without contradicting the idea that control is involved.

I won't say we have to go so far as to perform a formal Test in
order to say anything about control, but the closer you can get to
that the more believable what you say about control will be.

ยทยทยท

---------------------------------------------------------------
Bill Leach (940408.2028 EDT) --

I agree that people are always controlling (who wouldn't that
has read anything here).

See above.

Right, wrong or indifferent, I see reorganization as being
random except that it probably functions in small areas of the
brain as a result of any one particular error signal.

This kind of reorganization must surely exist, but I'd prefer to
call it "tuning" because it requires most of a control system to be
present before it can do much good. Also, tuning can be done
systematically, and so doesn't really need the randomness of
reorganization.

The main reason reorganization was introduced way back when was to
explain why organisms learn tasks or skills because they are hungry,
thirsty, horny, in pain, and so forth. Why should it be possible to
train a pigeon to walk in figure-eights just by giving or
withholding pieces of grain? There's no logical or evolutionary
reason, from the pigeon's point of view, why doing this silly act
should have any connection to getting food. How is it that we will
learn just about ANYTHING if it turns out to have an effect on a
physiological state that's important to us?

It seems that there doesn't have to be any systematic or natural
connection between what we learn and why we learn, at least at the
most basic level of learning. If you go hungry unless you find a way
to get 5 from adding 2 and 2, you will find a way. If something bad
happens to you after you walk under a ladder, even some of the time,
you will learn not to walk under ladders. As Skinner found, this
reorganization process is susceptible to forming superstitions; you
might say that most learning is really superstition.

But that's the price we pay for having a system that can solve
survival problems in the total absence of any understanding of what
makes the world work. This is what the theoretical reorganizing
system was designed to do. It was designed as a system that could
build a hierarchy of control without relying on any physical
principles, any logical reasoning, or any cognitive understanding of
anything. The basic requirement is simply that there be a set of
intrinsic variables that represent the state of the organism, and
some inherited set of reference values to go with each one,
specifying the desired state (a specification that would be a
product of evolution). There doesn't have to be any build-in
knowledge of what to DO about any discrepancies. If one variable
stands for the state of the body's energy supply, there is no need
for anything in the organism to understand that correcting the
discrepancy from the reference level requires eating something.
Random reorganization will eventually run across that solution,
making the discrepancy disappear and halting reorganization.

Of course some inherited pre-organization would make random
reorganization much more likely to work -- starting out with things
like arms and legs and eyes, for example, and a nervous system that
contains the raw materials for doing different kinds of
computations. Just how much preorganization exists is unknown, but
it is surely limited. It is probably the most extensive in the
simplest organisms, which can't learn different ways of satisfying
their needs. In the human organism it must be minimal -- else how
could we learn to keep warm and fed by holding onto two Jacks and
drawing three new cards?

David Goldstein asked

Why do we need this test if we can automatically assume that
the person is controlling.

And you replied

It is not that we "can automatically assume", there are no >choices

there, the person IS controlling. The question is

"what" are they controlling (perception naturally, but what
particular perception).

See my comment to David, above. Asking WHAT a person is controlling
is in fact asking IF a person is controlling something. If you can't
find any variable you can prove is under control, you're in a pretty
weak position, explanation-wise. I don't care how logically
satisfying the control hypothesis is, if you can't pull up some
evidence that will discriminate the control explanation from any
other, you're just working on faith and deserve to be treated like a
True Believer. Since this is one of my main objections to other
theories, including things like S-R theory, flat-earth theory,
international conspiracy theory, and economic theory, it is only
fair that I consider myself subject to the same requirements.

For PCT research purpose, you must know the controlled
perception for the research to have any meaning at all. That
indeed is precisely why "hard" PCT research is so damnedably
difficult. Detecting the "controlled perception" is close to
impossible to do with certainty. Even in the "tracking
experiments" if the subject changed references it would be
close to (?) impossible to determine what the new reference is
and the subject may not be a reliable source for that
information.

Now you're talking my language. We don't need certainty, but we need
pretty damned solid observations and evidence, and spending a lot of
effort on guessing where PCT will go next is just a waste of time --
if we want a science.

David:

If all our efforts to apply the Test fail to identify the
perceptual variable being controlled, do we still want to >assume

that the person is controlling a variable?

Bill:

Yes! Even if there is neural damage or disease. That a >control

system has failed does not change what it is... a

control system.

Dangerous territory there. If you know it WAS ONCE a control system,
sure. But assuming without proof is a good way to become a fanatic.
I think PCTers want the reputation of always having a way to back up
what they say.

Minor stuff:

Dag said

The principles of a new "people engineering" science help us ...

And you said

(That singular form "help" seems to be awkward to me)

Unfortunately, it's the grammatically right form: the number of the
verb (help) has to agree with the number of the subject
(principles). Just eliminate the intervening prepositional phrase,
"of a new people engineering science," and you have
"principles ... help us." It wouldn't sound right to say that the
principles helps us, would it?
----------------------------------------------------------------
Avery Andrews (940410) --

Nice try on sci.cog. You'll probably get remarks about cognitive
tadpoles.
---------------------------------------------------------------
Best to all,

Bill P.

<[Bill Leach 940411.21:15 EST(EDT)]

[Bill Powers (940410.0930 MDT)]

Looks like I got a taste of what Tom has been experiencing. I have not
had any internet access for over 24 hours.

I like to think that I do care about being right and will reasonably
accept critical comment.

I will try to accept that I don't KNOW that humans (and all other living
things) have been proven to be control systems. However, though I must
admit that my experience with a really scientific study of human
behaviour is limited (at best), I must say that I find the "supporting"
evidence overwhelming.

I admit that the jury may still be out on the "common" meaning of the
term behaviour. However, when I say that humans are control systems now,
I mean that they are even if sleeping or unconscious. It seem to me that
the evidence for certain control systems is beyond conjecture.

I second this motion, and call Bill Leach's attention to it, too. In
thinking of tentative explanations for phenomena, it's one thing to
propose that a control process is responsible, but another to prove
it. If we just decide that everything is control, we'll be doing

I will fully agree that religious assumptions are not to our advantage on
any "side." However, once you begin to get the idea that control does
explain many things, particular those that see to defie other
explaination, it seem difficult to think that there can be any other
explaination(s).

more reorganization

The idea that I get about reorganization is that it is always
non-volitional and may definately "make things worse".

Systemic changes would, I think, tend to react comparitively quickly,
reverse very fast in the event of an erroneous change and likely not make
a major control system error... all things that I should think that
reorganization would have trouble with.

Now my whole conception here could well be "out of line" but I see
reorganization as a "violent" "forced" change with no direct "logic"
involved (that is logic as in, for example, planning). Now maybe in an
adult, things "are not so bad" in that even with reorganization induced
changes, there will still be some measure of "cognitive over-ride" and
maybe even some amount ot "restoration of the previous control system"
via programmatic operations.

But that's the price we pay for having a system that can solve
survival problems in the total absence of any understanding of what
makes the world work. This is what the theoretical reorganizing
system was designed to do. It was designed as a system that ...

Reads like a pretty strong belief that humans ARE control systems to me.

See my comment to David, above. Asking WHAT a person is controlling
is in fact asking IF a person is controlling something. If you can't

I agree here certainly. I was sloppy. I recognize that a person might
not be controlling anything in their environment at times. Most of the
time however, A person that is doing something IS controlling but again,
the controlled perception may not be about anything external to the
person at all.

     //////////////////////////////////////////
     / /
     / -bill /
     / Bill Leach, W.R. Leach Co. /
     / bleach@bix.com 71330.2621@cis.com /
     / ARS KB7LX@KB7LX.ampr.org 44.74.1.74 /
     / 919-362-7427 /
     //////////////////////////////////////////