Drug happy

[Martin_Taylor4]

[Martin Taylor 971226 17:30]

Tim Carey (972612.0720)

Your posting puzzles me a bit. I tried to answer [Martin Taylor 971224 2215]
some questions in your message of (972412.0800) (the date is a bit puzzling,
too:-), that indicated you did not have the technical background to
understand how control works, for example:

What do you mean by a network?

Are we speaking about control systems within one person or between people?

What's a "coupling"?

How do you get a positive feedback loop? Can a negative feedback
loop turn into a positive feedback loop?

What's an amplifier?

What's a flip-flop?

What's an oscillator?

And several more of the same general nature. In addition, you made a lot
of comments that your _could not_ have made if you had a reasonably
substantial understanding of how control works.

I attempted to answer these questions and comments as best I could,
thinking it might help you to develop an understanding of control and the
kinds of thing that can happen when control systems interact. You reply
by answering:

I don't know what impression you got from my last
post but I'm sorry if I've not communicated my position clearly enough. I
believe I _am_ learning about control.

If that is so, what was your reason for asking the questions and making the
commentsthat led me to believe you lacked the necessary background?

Was it purely an altruistic gesture, to get me to take the time to write a
tutorial for those who are less well informed? I don't mind doing that, but
I wrote it with the apparently false idea I would be helping _you_.

I have copies of demo 1 and demo 2 on
my computer which I go over from time to time, I have visited Bill's web
page, and I regularly do the demos on Rick's web page.

Well, that's good. But then why ask "what's an amplifier" and " how do
you get a positive feedback loop," and why make all those other comments
that suggested to me that you would benefit from studying the demos?

My point was that I
am _only_ interested in learning about control and I am having trouble
deciding what _is_ and _isn't_ "control-talk" in your posts.

_Everything_ in my messages has to do either with elementary control
systems or the interactions among elementary control systems. For many
of the phenomena that occur when control systems interact, it simply
doesn't matter that the interacting entities are control systems.

Interacting control systems have some special properties that interacting
non-linear amplifiers don't have, but in the context of a discussion of
David Goldstein's obsessive-compulsive issue, those special properties
don't come into play much. They _do_ come into play, as I tried to show,
when we get to issues such as why Leo Kay said he could control if something
important was to be done, or why his problem showed up more at home than
at work.

A network is a number of entities that have some kind of connection
between them. If you prefer mathematical terminalogy instead, it's
a "graph." One kind of graph or network is a road map, in which the
entities are cities and the connections are the roads.

Here's a good example Martin. I'm not interested in road maps, and I'm only
interested in "a number of entities" if these entities are control systems.

If you learn the basics, you can apply them where you want. If you don't,
all you will get is epicycles. I tried to use an analogy with which I
assume you are familiar, to answer a _direct_ question you asked. You
don't want the answer, fine. Don't ask the question. Networks of control
systems is what we are interested in. If you don't know what a network
is, how on earth can you begin to understand how networks of control
systems behave?

I find this a hard question to answer, without the necessary foundation.
But I'll try. Quick answer: the dynamics is the way something changes
over time when left to itself. Less quick answer: Think of a child's
swing, hanging from a tree branch.

Martin, I'm not interested in child's swings either.

I get the impression that you want only a description of the way the
movements of the planets look from central Brisbane, and wouldn't be
interested in how they might look from Adelaide, let alone that from
another viewpoint their motions might look like ellipses ("what's an
ellipse?") or that they might be succinctly described by three simple
laws and some measurements of their mass ("what's a mass?").

You wanted to know what I meant by dynamics...or did you? You said you did.
Why was that, if you did not?

Networks of control systems follow very complex dynamics, under some
fairly general coupling conditions, and the tighter the coupling, the more
likely it is that the control systems will not control well, if at all.

Change these couplings, and you can change the network dynamics--apparently
often to the benefit of the person concerned. There. I've used in one
sentence three terms you asked about, three critical terms. In my
answer to you, I used them in a context with which I assumed you might
be familiar. Having understood the terms in that context, I hoped you
would understand them in the context you are interested in. But no:
"I'm not interested in child's swings." Oh, well. Sorry to waste your
time.

I think we have well and truly established that I am unable, at this stage,
to participate intelligently in a conversation that is "more general" I
think it would be ultimately more rewarding for both of us if we just stuck
to discussions about control systems.

Yes, it would be more rewarding if we just stuck to a discussion of the
retrograde motion of Mars than if we talked about the more general case
that massive objects attract each other. Yes, indeed:-(

systems. The organisation of _a_ control system is simple: one perceptual
input function, one comparator with a reference input, one output to the
environment. If the system is actually to _control_, that output must be
hitched to the environment in some way that allows it to affect the
perceptual input function in such a way that the error is reduced

(negative

feedback).

Isn't the "hitching" of the output through the environment accounted for by
the feedback function?

I mean the couplings between the output function and the physical variables
that are arguments to the CEV function. That's part of the feedback
function.

The "hitching" couplings and the perceptual input function together
account for the feedback function, _not_ the other way around. To say
that the feedback function accounts for the couplings between the output
function and the physical world is rather like saying that the retrograde
motion of Mars accounts for gravity.

disturbance signal source. Best to use "disturbance signal value," if
I understand your question properly, and the answer is "NO, the reference
signal is irrelevant to the disturbance signal value."

No, Martin I wasn't talking about how the reference is related to the
disturbance. I believe in your statement you were talking about the
disturbance causing actions. I was suggesting that the actions are caused
_not only_ by the disturbance but also by the reference perception that is
specified at that particular time.

Of course...and as I also said. If you missed my discussion of "cause"
a few days ago, I'll send it to you privately. If you want to pursue the
"cause" of the actions, don't forget the nature of the perceptual function,
the nature of the output function, and the organization of all the
supporting control systems in the hierarchy. And lots of other surrounding
conditions, too.

I don't know how beneficial any of this discussion is for either of us at
the moment.

This particular message probably isn't, since it is based on my total
confusion as to what you want to achieve, and the fact that you say you
understand how control systems function, but ask questions that suggest
just the opposite. I really don't know how to answer helpfully.

I really appreciate you taking the time to explain your
position and various ideas that are important to you, I just think we are
in very different places at the moment.

I'm not clear even what you mean by my "position." I have mentioned certain
facts, and speculated (and labelled as speculation) that those facts might
be relevant to a condition in real people. You enquired as to what I meant
by some concepts that are fundamentally necessary if you are to understand
the facts, and I tried to explain them. What "different places" do you mean?
I'm trying to deal with issues you raise, and provide some of the background
you appear to need in order to understand what I am saying--as you requested.

Maybe you could be clearer, and then I could perhaps be more helpful.

Martin

[Bill_Powers1]

[From Bill Powers (971227.0520 MST)]

[Martin Taylor 971226 17:30]

Pardon me for butting in on your conversation with Tim Carey, Martin, but
you seem to be insisting that there's no way to understand PCT without
starting from the most general network principles and then deducing control
systems as a special case. This just isn't true. If it were, then I
wouldn't understand PCT either.

Control is a phenomenon, not a theory. An adequate model of control can be
constructed to explain the phenomenon without bringing in all the possible
organizations that are NOT control systems. If we observe that human
behavior demonstrates the properties of control systems, then to explain
what we observe requires only that we explain control. All the other
possibilities become irrelevant until such time as we find some other
identifiable mode of operation.

It's a disservice to PCT as a subject of study to suggest even that one
must understand concepts like amplification to grasp the principles of
control in a practical way (not that I think Tim Carey would be incapable
of understanding it, if it were explained appropriately and he wanted to
understand it). If the only way to understand mathematics were to start
with Godel's Theorem and work down to arithmetic, there would be no way to
get there from here -- there wouldn't even be any mathematicians.

I don't like this game of "Anything you can generalize, I can generalize
further." That's the game academic snobs play with each other, the point
being to establish who is smarter than whom, not to reach any useful
conclusions and particularly not to teach anyone anything. In fact teaching
is the exact opposite of the point of this game. The real point is to
convey the message, "I'm so much smarter than you that there's no point in
your even trying to grasp what I'm talking about -- just listen to my
conclusions and stop pretending that you're in the same class with me."

What's odd about this is that when you decide to teach PCT, you are
perfectly capable of doing so in a lucid way, accessible even to beginners.
It's only when you get into these tangled webs of complexity that you lose
touch with the common man, and I must say, with reality.

Best,

Bill P.

[Martin_Taylor4]

[Martin Taylor 971228 13:30]

Bill Powers (971227.0520 MST)]

[Martin Taylor 971226 17:30]

Pardon me for butting in on your conversation with Tim Carey, Martin, but
you seem to be insisting that there's no way to understand PCT without
starting from the most general network principles and then deducing control
systems as a special case. This just isn't true. If it were, then I
wouldn't understand PCT either.

Butt away. If I'm acting ineffectively to control my perceptions, much
better to have additional disturbances that might lead to effective
reorganization. Anyway, I imagine that if you didn't have so much else
to be concerned with, you would have tried to help Tim Carey yourself.

However, from my side, I was most certainly _not_ starting from network
principles and deducing control theory as a special case. I started by
mentioning to CSGnet in general that when you have a network of control
systems (whatever they may be), the behaviour of the network can be
quite complex, and the repertoire of behaviours of any particular
network depends strongly on the couplings among the control systems
of which the network is composed.

Tim asked a series of question about this, that I tried to answer quite
carefully. Some of his questions were about items so basic to control
that it seemed to me that he could not have had any understanding of
how control works. How can you ask what positive feedback is, and how
it comes about, while at the same time understanding what negative
feedback is, and its place in control? How can you understand the
effects that occur when feedback is delayed in a control loop with high
gain, if you don't know what an oscillator is? I deduced from Tim's
questions that he did not have the background necessary to understand
control, and that he might well start by looking at your demos.

Without understanding the fundamentals of control, I think it must be
very hard to understand how to apply to networks of control systems the
lessons learned from networks of simple non-linear summing amplifiers
(and when not to try to apply those lessons).

It must be even harder if you don't know what an amplifer is, or what
a network is.

Tim then came back with a whole lot of stuff about how my attempts to
explain the basic concepts were of no interest to him, and that he was
very familiar with your demos, as well as Rick's. That seemed, and
seems, to be in direct contradiction to the notion that he wanted to
understand the concepts I introduced in speculating about obsessive-
compulsive disorder. I remain puzzled, but willing to attempt to
continue if I could determine where the problem lies.

Control is a phenomenon, not a theory. An adequate model of control can be
constructed to explain the phenomenon without bringing in all the possible
organizations that are NOT control systems.

Can it be done without understanding the meaning of feedback, and the
relation between positive and negative feedback?

It's a disservice to PCT as a subject of study to suggest even that one
must understand concepts like amplification to grasp the principles of
control in a practical way

OK. Perhaps you could demonstrate how to explain control without the
concept of amplification. I can't. What would a term such as 1/(1+G)
even mean in such an explanation? Why should anyone begin to take your
word that a "magical" system somehow always manages to stabilize its
perception of something in the real world when influences in the
real world are buffeting the thing perceived. If they know what an
amplifier is, they don't have to take your word. Moreover, they can
then see to what degree this stabilization happens, and under what
circumstances it doesn't happen.

I don't like this game of "Anything you can generalize, I can generalize
further."

Who's doing that? My game is "This complicated idea can be understood in
this simple context. It's the same idea when it appears in the complicated
context you are trying to understand."

In fact teaching
is the exact opposite of the point of this game.

That's your opinion, I guess. I find it usually helps to explain things in
a simple context first, before trying to apply the ideas in complicated
situations.

What's odd about this is that when you decide to teach PCT, you are
perfectly capable of doing so in a lucid way, accessible even to beginners.

When the person claims to be both a pre-beginner and an advanced student,
that's when I get into trouble.

It's only when you get into these tangled webs of complexity that you lose
touch with the common man, and I must say, with reality.

I'd change that "complexity" to "simplicity", but I deny the loss of touch
with reality. What we have been talking about is a simplified approach to
a real experiment. No more than that. If you want rarified abstraction
and complexity, look elsewhere.

Martin