Drug happy

[Tim_Carey]

[From Tim Carey (971219.0810)]

Hi all,

I've been fussing with those two posts that David sent and in particular
the personal account by Leo Kay. I'm sure everyone who read it got
different things out of it and perhaps all of us were touched by the
courage and ingenuity of this man. From a PCT perspective, I was intrigued
by the man's account of his working life. Isn't anyone else out there on
CSGnet a little curious at how this man could maintain "a reasonably calm,
in-control appearance" at work and then "discharge" his symptoms when he
got home? If this account is accurate Leo's controlling and discharging of
symptoms went on 8 hours a day, 5 days a week, for _40 YEARS_!!!

Isn't it also curious that "I lasted all those years on the job not because
of staying power, but rather due to an inability to pursue natural
ambitions. Whenever I tried to do so my symptoms would return with a
vengeance"?

Doesn't anyone find these kinds of patterns with symptoms interesting in
terms of processes of control? Am I being too simplistic and missing
something really significant?

It seems to me, that at best, drugs alleviate symptoms ... they do nothing
to directly address how well a person controls perceptions that are
important to them. I'm not even necessarily advocating the total ban of all
drug therapy but I think if we took the focus off the symptoms and looked
at control processes we may become a little smarter about when we drugs and
what we use them for.

If this is a problem with hardware, how is it that the individual is able
to exert so much control in terms of when and where it occurs. Leo at one
point states that he has a "debilitating attention deficit disorder" and at
another point he states that he did "months of personal SSRI research in
medical and public libraries -- not easy for a person with ADD, and of
limited scientific bent" .... not easy indeed, I'm sure Leo is right, in
fact one wonders how he could do it at all if the ADD was debilitating.
Isn't it a little bit curious to anyone besides me that when something
really important came along in Leo's life he wasn't debilitated by ADD
anymore. Or at least the ADD debilitation abated sufficiently to allow Leo
to do the research he wanted to do.

There are other examples throughout this account that illustrate what I'm
talking about but you don't need me to point them out, I think you've got
the idea.

Please don't think I have written this to mock Leo Kay, it sounds like he
has had a dreadful ordeal throughout most of his life and I am genuinely
happy that he has finally found some relief. I just wonder whether relief
might not have come faster and more elegantly if professionals understood
that human beings control perceptions they don't produce actions (or
symtpoms)!!

Cheers,

Tim

[R_Marken]

[From Rick Marken (971218.1500)]

Tim Carey (971219.0810) --

If this [Leo Kay's] is a problem with hardware, how is it that
the individual is able to exert so much control in terms of when
and where it occurs.

An awfully good question!

Isn't it a little bit curious to anyone besides me that when
something really important came along in Leo's life he wasn't
debilitated by ADD anymore.

Curious, indeed!

I just wonder whether relief might not have come faster and
more elegantly if professionals understood that human beings
control perceptions they don't produce actions (or symtpoms)!!

Wonder no more. It would!!

An excellent post, Tim. Thanks.

My own position on drugs (and everything else;-) is as follows:
If it feels good, do it; if it stops feeling good (because you
are doing it too much or because people are putting you in jail
for it) stop doing it.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[DTSDTO]

your right on the mark with your thoughts on Drug therapy--

It is sad to say just how people take psych meds --- I have met many that were
raised on them from childhood to adults. Many others I have interviewed have
a traumatic event in their lives and have been on them from that point on --
and were never able to reorganize (if fact I don't believe they ever will as
long as they are on the Psych meds.

In the evolution of the human mind (approx. 4.5 billion years) it amazes me
how much pathology there is in the world now that drug therapy is available.

Mark

[Martin_Taylor4]

[Martin Taylor 971219 13:10]

Rick Marken (971218.1500)] to Tim Carey (971219.0810) --

If this [Leo Kay's] is a problem with hardware, how is it that
the individual is able to exert so much control in terms of when
and where it occurs.

An awfully good question!

Isn't it a little bit curious to anyone besides me that when
something really important came along in Leo's life he wasn't
debilitated by ADD anymore.

Curious, indeed!

Actually, the answer to both questions is implicit in my message of
[Martin Taylor 971217 1&;44] (Actually 17:44).

+ If I understand David correctly, he is talking about
+a disorder characterized by a locked-up positive feedback loop, for which
+the appropriate technical solution is to reduce the loop gain below unity,
+allowing the elements of the loop to be influenced by signals outside
+the loop. At one point in one of his messages, he even says straight out
+that the therapy "reduces the coupling." That's exactly the right therapy
+if the problem is too tight a coupling.

Lockups in such loops have many forms, even in very simple and small
networks.

To save you from having to read this longish message, the bottom line is
that small changes in the environmental coupling can cause large changes
in the complex behaviour of simple networks, and probably also in locked-up
minds.

Let me tell you about one series of simulations Allan Randall and I did
some years ago. We made a very small network of simple nodes. Each node
was of the same kind--it made a weighted sum of its inputs, and then
did a logistic transform on the sum to produce the output, like this:

         > *********
         > ****
output | **
         > **
         > ****
         >*****_______________________
             weighted sum of the inputs

The nodes were randomly interconnected, and the weights chosen from a
Gaussian random distribution. For most of the experiments, we used a
network of 8 nodes.

The experiment was to look at the effects on the network behaviour of
providing different kinds or inputs and of changing the gain of the nodes
(gain being the slope of the central portion of the squashing function;
gain represents the coupling among the nodes).

Such networks usually show some form of locked-up behaviour. If there
is not enough inhibition in the network weights, all the nodes go to
some stable value and stay there, and inserting pulse or step input may
not change anything (although in some cases it will move the net to another
locked-up stable state).

If there is more inhibition (i.e. either more weights are inhibitory, or
the inhibitory connections have bigger weights), the net can show a wild
variety of behaviours, sometimes locking up in a fixed stable value,
sometimes oscillating with a very short period, sometimes behaving
chaotically. Very small changes in gain in just one of the nodes can
change the behaviour utterly.

More interestingly, if you change the gain of one node slowly so that the
network behaviour moves from, say, a fixed point lockup, through chaotic
behaviour to a steady oscillation, you often find that when you move the
gain slowly back again, the transitions come at different values of gain.
The system as a whole shows hysteresis. This means that there are values
of the gain for which the network can either be locked up or chaotic,
with no difference in the hardware conditions. Different inputs can
switch the network from one state to the other, and back again.

But that's not the end of it. It is very easy to find values of the
_hardware_ parameters for which the network shows as many as five different
behaviours (in our 8-node network), and it can switch from one behaviour
to another if you input, say, a brief low-level impulse into just one of
the nodes at a critical moment. Even when the network is apparently
locked up so that all the nodes take on fixed, extreme values, a small
impulse at one input can shift it into oscillating or chaotic behaviour.

Here, I think, is the answer to the questions posed by Tim and Rick. The
interactions of control systems are obviously more complex than the
interactions of our simple 8-node network, but the principle of operation
is the same--the effects of the output of one Elementary Control Unit (ECU)
feed back to its own perception by a variety of routes, due to the
disturbances each ECU's output provides to the perceptual signals of
other ECUs. Some of these are, in effect, positive feedbacks, some
negative, but they are all differentially delayed, as are the feedback
effects within our simulated 8-node network.

It is very easy to see how such a system of interacting control systems
with coupling that is too tight (either because of high gain in the
individual control loops or because of high mutual disturbance among
the systems) can behave in a multititude of ways depending on the
environmental conditions. When there is a perception that is important
to control, things can change in the potentially locked-up hardware
structure. It could cause a worse lockup, or it could free an already
locked-up loop. When Leo Kay gets home, his environmental perceptions,
and often the perceptions to be controlled at high levels, are usually
much the same, so it is not unreasonable to expect that if a lock-up
exists at one time when he is home, it will recur on another time
when is he home--but not if he has something out of the ordinary to
control.

I don't find it either surprising or curious that an individual with a
hardware problem that is not too severe can find that the problem goes
away when there is something important to do, and can, with effort,
control so as to suppress the lockup -- for a while. To perceive that
a lockup is happening or about to happen may well be a perception for
which control can be learned.

The _fact_ is that simple networks of interacting trivial systems can
show very complex and variable behaviour. The _speculation_ is that this
is what is happening in a locked-up mind. The _fact_ is that in simple
systems, reducing the coupling simplifies the network behaviour, and may
avoid lockup entirely. The _speculation_ is that reducing the coupling
can reduce the problem in a locked-up mind. According to David (and
Leo Kay), the clinical finding is that the speculation accords with the
facts of a mind, at least sometimes. Which suggests that the speculation
might have something to do with what is happening in the mind.

Martin

[Tim_Carey]

[From Tim Carey (971221.1105)]

[Martin Taylor 971219 13:10]

Hi Martin,

Thanks for your post ... I have to admit though that a lot of it went over
my head. I live for the day when I can participate intelligently in these
kinds of discussions!!

To save you from having to read this longish message, the bottom line is
that small changes in the environmental coupling can cause large changes
in the complex behaviour of simple networks, and probably also in

locked-up

minds.

I don't understand this ... I didn't think that from a PCT perspective that
anything in the environment _caused_ any kind of behaviour.

The experiment was to look at the effects on the network behaviour of
providing different kinds or inputs and of changing the gain of the nodes
(gain being the slope of the central portion of the squashing function;
gain represents the coupling among the nodes).

I may have this all wrong, but this paragraph sounds like you are trying
effect the output by varying the input, I thought control systems worked
the other way around ... the varied their output to get the input they
want.

Such networks usually show some form of locked-up behaviour.

My whole point in the first post is that no-one is thinking of examining
what variables Leo might have been controlling for, again, the statement
above seems to be focussed on the output as though that is what the system
is "doing". Was this systm unable to control at this point?

More interestingly, if you change the gain of one node slowly so that the
network behaviour moves from, say, a fixed point lockup, through chaotic
behaviour to a steady oscillation, you often find that when you move the
gain slowly back again, the transitions come at different values of gain.
The system as a whole shows hysteresis. This means that there are values
of the gain for which the network can either be locked up or chaotic,
with no difference in the hardware conditions. Different inputs can
switch the network from one state to the other, and back again.

But that's not the end of it. It is very easy to find values of the
_hardware_ parameters for which the network shows as many as five

different

behaviours (in our 8-node network), and it can switch from one behaviour
to another if you input, say, a brief low-level impulse into just one of
the nodes at a critical moment. Even when the network is apparently
locked up so that all the nodes take on fixed, extreme values, a small
impulse at one input can shift it into oscillating or chaotic behaviour.

Again, the above paragraphs seem to be trying to explore aspects of a
system's output. I was more interested in Leo's input.

other ECUs. Some of these are, in effect, positive feedbacks, some
negative, but they are all differentially delayed, as are the feedback
effects within our simulated 8-node network.

I'm afraid I don't understand positive feedback and it's role in control
systems.

It is very easy to see how such a system of interacting control systems
with coupling that is too tight (either because of high gain in the
individual control loops or because of high mutual disturbance among
the systems) can behave in a multititude of ways depending on the
environmental conditions.

Here it sounds as though the environment is controlling the system's
output. What am I missing?

When there is a perception that is important

to control, things can change in the potentially locked-up hardware
structure.

This was perhaps the whole point of my post. That when there was something
important, Leo could control effectively. I would be interested in research
that explored this ability to control in more detail.

I don't find it either surprising or curious that an individual with a
hardware problem that is not too severe can find that the problem goes
away when there is something important to do, and can, with effort,
control so as to suppress the lockup -- for a while. To perceive that
a lockup is happening or about to happen may well be a perception for
which control can be learned.

I've always thought of epilepsy as a problem with hardware and even though
I've known people with epilepsy, I have never known them to be able to make
a seizure go away if, at the time it occurs, there is something more
important to do. Would diabetes be similar?

Which suggests that the speculation

might have something to do with what is happening in the mind.

My very uninformed opinion is that it has everything to do with what is
happening in the mind. I just don't think current concepts of mental
disorder ackowledge the control that people might have. My understanding is
that mental disorders are currently viewed in the same way as physical
illnesses ... they just "happen to you". This might indeed be the case, I
just don't think we'll ever know until we start looking at these problems
in terms of what input (if any) the person might be controlling.

Thanks for taking the time to reply,

Cheers,

Tim

[R_Marken]

[From Rick Marken (971221.1010)]

Martin Taylor (971219 13:10) --

I think what Tim Carey and I were questioning about Leo Kay's
testimony was the assumption that the apparent success of
drugs indicated that a _hardware problem_ had been solved.
Tim's point was (I think) that you simply can't tell by just
looking at what are seen as "manic-depressive" outputs what a
person is doing (trying to control), whether there is even a
problem, whether all problems classified as "manic-depressive"
are the same (as is implied by placing them all in the same
diagnostic category) or whether the problem (if it exists)
is the result of a hardware or software malfunction.

You said earlier:

If I understand David correctly, he is talking about a disorder
characterized by a locked-up positive feedback loop, for which
the appropriate technical solution is to reduce the loop gain
below unity

Assuming that the problem is what you call a "locked up positive
feedback loop" you still don't know that it's a "hardware"
problem. The positive feedback that occurs briefly in the
"Levels of control" demo (for example) is not the result
of any hardware problems (which I take to be organic failures
in the case of living systems, like neurons that won't fire or
that won't stop firing). Perfectly normal control systems can
oscillate or "lock up" (at maximum output) simply because they
are in conflict (controlling the same variable relative to
different references).

Drugs could appear to "fix" such problems by chemically
knocking out the conflict (while the drugs are in effect).
The drugs could do this by reducing to zero the firing rate
of the reference signals going to the conflicted systems;
there is no longer a conflict because one or both of the
systems no longer controls for the conflicted perceptual
variable. I think strokes (a definite hardware problem)
often have this "helpful" effect on conflicts; many of my
father's conflicts about initimacy with his children seemed
to disappear after his stroke about 15 years ago; suddenly
he's all lovey-dovey -- but paralyzed so he still can't
hug (God, the big comedian in the sky, has such a great sense
of humor; sometimes he's so funny I'm almost tempted to
believe in him;-))

The fact that drugs can help people with control problems
(depression is a control problem) does not mean that the
problem was, indeed, a "hardware problem". It may be,
and I believe that in most cases it is, a "software
problem"; that is, one that can be "fixed" (in principle)
by reorganization (a process that is likely, by the way,
to involve hardware changes).

I have nothing against the "responsible" use of drugs for
therapy; drugs are fine when a person feels that he is in
desperate straights, nothing else seems to work and the person
himself agrees to the drug treatment (after being informed
of all known risks). What I dislike is the self-deception
involved in thinking that, because some drug sometimes does
help people with behavioral/mental problems, the cause of
these problems is definitely hardware (brain chemistry).

Best

Rick

···

--

Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

[Tim_Carey]

[From Tim Carey (971222.0525)]

[From Rick Marken (971221.1010)]

Thanks for the post Rick, you clarified a lot of what I was trying to say.

I have nothing against the "responsible" use of drugs for
therapy; drugs are fine when a person feels that he is in
desperate straights, nothing else seems to work and the person
himself agrees to the drug treatment (after being informed
of all known risks).

Just for the record, I have nothing against the "responsible" use of drugs
either. I just think that until we start investigating the "true" nature of
these problems through sound PCT research, we are using drugs
serendipitously, not responsibly.

Cheers,

Tim

[David1]

Rick,

I believe it was Bill Powers who suggested that OCD might be a hardware
problem which was helped by Prosac. Remember, the idea that after
Prosac, Leo Kay still had problems but could approach them in a normal
way?

Did you see the abstract of the study by Scwartz which I posted. He
reported on changes in the metabolic activity of certain brain areas and
the associations between certain brain areas. These changes were
brought about by effective drug treatment or effective
cognitive/behavioral therapy. Are these hardware or software changes?

In PCT, what part of the theory is "hardware" and what part is
"software" ? Aren't these terms somewhat inappropriate for people and
animals since we know that experiences can modify the structure of the
nervous system?(number of cells, connnections, for example).

I have had discussions on the use of drugs for psychological symptoms
and physical symptoms with Bill Powers and Tom Bourbon. There is a
philosophical opposition to the use of drugs which may be worthwhile to
spell out again. Some of the beliefs I recall are: We don't know how
the drugs work exactly. If we do too much, the system will stop making
an effort. For example, by taking thyroid hormone, the thyroid gland
cuts back even more. If we reduce anxiety by drugs, a person will not
be motiviated to make life changes. Drugs can have terrible side
effects, such as Tardive Dyskinesia and blood changes. The
philosophical opposition is to drugs in general and not just to drugs
which work on the brain. It seems to go with self-regulation approaches
and is not limited to PCT.

···

From: David Goldstein
Subject: Drug happy
Date: 12/21/97

[David_Wolsk2]

I feel there is abundant evidence from all sorts of situations ..... birth
defects, head injuries, strokes, cataract removal, split brain operations
..... and the aftermaths ..... that talking about hardware problems as
givens is ignoring all the evidence of the brains enormous capacity for
reorganization. That reorganization is obviously built into the control
systems. This is one reason I found Bill's approach to evolution so compelling.

My hardware is heading south to Seattle and I may need some drugs after
paying 1.45 Cdn for 1 US dollar.

David W.
Victoria BC Canada

···

at 21:00 PST 971221 David Wolsk wrote:

At 10:15 21/12/97 -0800, you wrote:

[From Rick Marken (971221.1010)]>
Martin Taylor (971219 13:10) -->
I think what Tim Carey and I were questioning about Leo Kay's
testimony was the assumption that the apparent success of
drugs indicated that a _hardware problem_ had been solved.

[Bill_Powers1]

[From Bill Powers (971222.0535 MST)]

From: David Goldstein
Subject: Drug happy
Date: 12/21/97

In PCT, what part of the theory is "hardware" and what part is
"software" ? Aren't these terms somewhat inappropriate for people and
animals since we know that experiences can modify the structure of the
nervous system?(number of cells, connnections, for example).

I agree that there's an indistinct boundary between hardware and software
in the brain, owing to the brain's ability to reorganize. There are some
people who deny that there's any hardware component at all, and some who
seem to believe that there's nothing more to the brain than physics and
chemistry.

My view, such as it is, has been strongly influenced by my experience with
electro-mechanical systems and computers (both analog and digital), which
has run in parallel with my psychological interests all through my adult
life. I think that there is a core organization in the body which must be
in working order if the brain is to be able to function at all. For
example, there must be neurons that can stay alive and transmit signals
from one place in the brain to another with minimal degradation. There must
be a "power supply," such that when brain activities or motor actions draw
on sources of metabolic energy, the energy is that is used is replaced and
maintained within some fairly narrow range (of course the basic raw
materials must be supplied from outside the organism through motor
behavior). Bones and muscles must be maintained. Dead cells must be removed
and healthy cells must multiply to replace them. There is a whole list of
processes which seem to function as a life support system, and work in ways
that are fairly standard from one person to another and even one species to
another. These are the things I think of as "hardware."

I think of the brain, or the nervous system as a whole, as the user and
controller of this hardware. The brain has sensors distributed throughout
the body as well as on its surfaces. It can produce behavior by energizing
muscles and glands, and it can alter the biochemical state of the body
through efferent connections to every organ system, and via the
neurohypophesis to all the hormone systems.

For the brain to operate properly, its own metabolism and energy supply
must be very closely regulated (the so-called blood-brain barrier). The
energy used in transmitting signals must be promptly replaced. The
insulation of the nerve signal pathways must be maintained, because if
signals stray from one path into another, the functional organization of
the brain will degenerate into chaos. I think that a multiplicity of
neurotransmitters exists primarily as a way of keeping functional units,
packed densely together, independent of one another, in the same way that
the hormone systems communicate through a common bloodstream by substances
that uniquely connect sources to targets. Neurotransmitters diffuse away
from the synaptic gaps, and while the surplus is partly cleaned up by
metabolism, there is still a potential for the signals within one
functional unit to influence signals in unrelated functional units. By
using different substances for synaptic transmission, different parts of
the brain (once organized) keep their functions separated.

All these considerations are completely independent of what a particular
brain _does_ in terms of its internal activities. They simply make it
possible for those activities to occur, just as the transistors and wiring
in a computer make it possible for programs to run without determining what
those programs will be. So I think of these things as the hardware aspects
of the brain.

The software of the brain consists of _what_ information is handled by the
brain. This is completely independent of the physiology and biochemistry of
the brain, because the physiology and biochemistry work the same way no
matter what the brain learns to do. The main effects of biochemistry occur
when something alters the optimal regulation of the underlying brain
physics and chemistry. That's what I call a hardware problem.

This separation of physiology and function is made more complicated by the
fact that the brain senses the state of the body, and learns to interpret
it. If epinephine is injected into the bloodstream, the body reacts in a
host of ways; its biochemical status shifts balance. The brain detects this
change, and concludes that something is happening like what has been
happening before when these changes occurred. It labels the result as
"fear" or "anxiety" or "anger," even though there's nothing else happening
that would indicate the need to flee or seek safety or attack. If a muscle
relaxant is administered, the brain senses the lowering of muscle tension
and concludes that some problem must have gone away; it feels more
peaceful. If a substance enters the bloodstream and creates a feeling of
pleasure, the brain interprets this to mean that something good and
pleasant has happened, as if some error had been suddenly reduced, as if
something important had been accomplished. A substance that generates a
sudden sense of being thrilled or astonished is interpreted by the brain as
if it has had some great insight.

The key phrase in all this is "as if." Each of these feelings, basically
bodily feelings, occurs normally under appropriate circumstances, as part
of normal brain-body behavior. Actually correcting some intrinsic error
feels good; actions that increase intrinsic error feel bad. Eliminating a
conflict results in a lowering of activity, both mental and physical, and
is felt as relaxation. The muscles may actually relax if they have been
held in conflict by opposing systems. Solving any problem lessens the
confusion of continual reorganization.

The injection of active substances into the bloodstream thus has effects
that mimic the consequences of normal behavior. The brain, if naive, has no
way to distinguish these effects from the normal ones, and it can easily be
fooled into acting as though the effects signal their usual internal
causes. If it feels angry, it may look around for something to attack. If
it feels relaxed, it may reduce its efforts to solve a problem. If it gets
the sensations that usually come when some great insight has occurred, it
may take whatever experiences were going on just then -- whatever thoughts
-- as being great insights.

As I understand them, psychoactive substances do not act by altering the
body's biochemical state, but by directly altering activities in the brain.
These effects, even for the educated brain, would be much harder to
understand correctly. The very systems that do the understanding are being
altered. Since the sensations coming from the body are not altered
(initially), there is no way to distance oneself from the chemical effects,
as when one realizes that the injection-induced rush of adrenaline is
really only a bodily sensation and is unrelated to anything one was
thinking or doing. If some substance lowers the capacity to detect the
signals that are normally interpreted as anxiety, one can only conclude
that those signals no longer exist, and no longer call for whatever action
would normally be taken to control them. One says, "My anxiety is cured!"
without realizing that none of the reasons for the anxiety has been
removed. The lump in the breast is still there; the only difference is that
one no longer feels anxious about it.

The main problem with psychoactive drugs is that no drug can carry
meaningful information into the brain, information that has anything to do
with the brain's appreciation of the world or itself. All they can do is
move the brain's biochemical hardware away from the state it had been in.
While the brain itself might be able to induce similar changes, when the
brain does it those changes are related to what is going on in the brain;
they are consequences of its operation and thus carry relevant information.
When the same changes are made by an extraneous substance, they have no
relation to what the brain is doing; they are imposed from outside,
arbitrarily.

The most pernicious of the "feel-good" drugs are those that supply us with
the experiences we normally get from learning, accomplishment, and good
human relations with others. A heroin user is not interested in love or
even sex; every feeling one could possibly get from such things is turned
on in a rush by the heroin; no action other than pressing the plunger is
necessary. The cocain user is filled with marvelous ideas; all difficulties
disappear and every thought is an insight. The shy and self-conscious
fourteen-year-old boy finds that he can make friends with anyone, even
girls, after just a couple of tokes of marijuana. The magical mystery tour!

These drugs are more than "recreational." They are destroyers of
organization. And I think the same is probably true of every psychoactive
drug, no matter how benificent the short-term effects may appear to be.

Sorry, I just can't stay away from that subject, wherever we start talking
about it. I have seen too much misery and too many ruined lives to leave it
alone.

Best,

Bill P.

[Bruce_Gregory9]

[From Bruce Gregory (971222.1100 EST)]

Bill Powers (971222.0535 MST)

These drugs are more than "recreational." They are destroyers of
organization. And I think the same is probably true of every psychoactive
drug, no matter how benificent the short-term effects may appear to be.

You may be allowing your use of the word "drug" to tar very
different kinds of chemistry with the same brush. The SSRI's
work in very different ways from traditional drugs and have
quite different consequences. I do not expect you to take my
word for it, but there are serious discussions of the
differences in the literature. If you really want to understand
the issues, I recommend Peter Kramer's _Listening to Prozac_.
Kramer is very much aware of the issues you raise and discusses
his experiences with people using Prozac. You are assuming that
the SSRIs work against reorganization but you really have little
evidence to support your claims. A better case can be made
that they facilitate reorganization.

Bruce

[R_Marken]

[From Rick Marken (971222.0830)]

David Goldstein (12/21/97) --

I believe it was Bill Powers who suggested that OCD might be
a hardware problem which was helped by Prosac. Remember, the
idea that after Prosac, Leo Kay still had problems but could
approach them in a normal way?

OCD _might_ be a hardware problem, but, then again, it might _not_.
My beef is with the assumption that, because a drug helps, the
problem is chemical (hardware). The fact that Kay still had problems
that could now be approached in a normal way does nothing to remove
my doubts about the problem being "hardware". I found that copious
amounts of alcohol eliminated my shyness problem so that I could
deal with other problems (meeting people, making conversation, etc)
in the normal way. But I don't believe my shyness was the result of
an alcohol deficit (the proof; I am no longer particularly shy and
I no longer consume alcohol either).

Did you see the abstract of the study by Scwartz which I posted.
He reported on changes in the metabolic activity of certain brain
areas and the associations between certain brain areas. These
changes were brought about by effective drug treatment or effective
cognitive/behavioral therapy. Are these hardware or software changes?

Good question. If they are controlled results of the operation
of reorganization (which can, in theory, produce new hardware
configurations such as new perceptual and output functions) then
I would call them software changes; if they are simply side effects
of control (like the bigness of muscles the result when one regularly
controls for lifting heavy objects) then I would call them hardware
changes. It is impossible to tell, from the Scwartz report, whether
these changes are hardware of software changes.

In PCT, what part of the theory is "hardware" and what part is
"software" ?

This is a very good question. Bill Powers (971222.0535 MST) made
a nice stab at this. It's not an easy distinction in PCT since
the PCT model (at least, the reorganization part of the model)
can control things (like neural structures) that we might like
to call hardware. And as David Wolsk (971221) notes, even what
seem like obvious hardware-based behavioral problems -- birth
defects, head injuries, strokes, cataract removal, etc -- can
apparently be delt with, to some extent, by the brain's ability
to reorganize -- which I consider a "software" capability.

I agree with Tim Carey's (971222.0525) position on drug therapy
for behavior problms;

I have nothing against the "responsible" use of drugs either. I
just think that until we start investigating the "true" nature
of these problems through sound PCT research, we are using
drugs serendipitously, not responsibly.

If the drug therapists would just admit their drugs work
"serendipitously" (like fixing a TV by kicking it) then I
would have no problem. The problem with drug therapy occurs when
the people who use it actually think they know what they are
doing. _That's_ the problem. There is nothing wrong with
trying to kick a TV to fix it if you don't know anything about
how a TV works. But there is something wrong with thinking that,
because you have been able to keep a couple of TVs working by
kicking them occasionally, you are now a qualified TV repair
person;-)

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[Bruce_Abbott]

[From Bruce Abbott (971222.1135 EST)]

Bill Powers (971222.0535 MST) --

I agree that there's an indistinct boundary between hardware and software
in the brain, owing to the brain's ability to reorganize. There are some
people who deny that there's any hardware component at all, and some who
seem to believe that there's nothing more to the brain than physics and
chemistry.

Put me in the latter category, if by "nothing more and physics and
chemistry," you do not mean to exclude the nervous system's organization.

I think of the brain, or the nervous system as a whole, as the user and
controller of this hardware. The brain has sensors distributed throughout
the body as well as on its surfaces. It can produce behavior by energizing
muscles and glands, and it can alter the biochemical state of the body
through efferent connections to every organ system, and via the
neurohypophesis to all the hormone systems.

Yes, but it is well to remember that this "user and controller" of this
hardware is composed of that hardware. I think of a digital computer in the
same way: as a machine running a program, it (the machine-running-program)
"uses and controls" the hardware (e.g., determining what digitally-coded
values will be sent to what registers, at what times, what operations will
be performed on them, and so on), but at the same time it _is_ the hardware,
and even part of the program (ROM BIOS) is a permanent part of its
structure. The distinction seems to be one of the hardware itself (the
physical arrangement of matter) and the dynamically changing signals and
states flowing through the hardware as determined through time by its
organization and its inputs.

The software of the brain consists of _what_ information is handled by the
brain. This is completely independent of the physiology and biochemistry of
the brain, because the physiology and biochemistry work the same way no
matter what the brain learns to do. The main effects of biochemistry occur
when something alters the optimal regulation of the underlying brain
physics and chemistry. That's what I call a hardware problem.

There may also be damage to the "wiring" of the brain which may create
biochemical imbalances even though the individual undamaged components are
regulating properly. A nice example is Parkinson's disease: Neurons in the
brainstem that carry signals to the basal ganglia (motor centers in the
brain) begin to die (for reasons not yet known). These neurons release the
neurotransmitter dopamine at synapses in the basal ganglia, where the
dopamine serves an inhibitory function. When the loss of neurons crosses a
threshold, the motor tremors and muscular rigidity that are characteristic
of Parkinson's disease begin to develop, because the loss of inhibition in
the basal ganglia effectively raises the gain in the motor system to the
point of oscillation. Treatment with L-DOPA, a precursor of dopamine,
increases the synthesis of dopamine in the remaining neurons by supplying
more raw material (circumventing the neuron's normal regulation of dopamine
synthesis rate); thus larger than normal amounts of dopamine are released
into the basal ganglia with each impulse, increasing the inhibition there
and bringing about more normal motor function. (Unfortunately, as neurons
continue to die, dosages must be increased, and eventually a point is
reached at which there are no longer sufficient neurons left to do the job.)

This separation of physiology and function is made more complicated by the
fact that the brain senses the state of the body, and learns to interpret
it. If epinephine is injected into the bloodstream, the body reacts in a
host of ways; its biochemical status shifts balance. The brain detects this
change, and concludes that something is happening like what has been
happening before when these changes occurred. It labels the result as
"fear" or "anxiety" or "anger," even though there's nothing else happening
that would indicate the need to flee or seek safety or attack.

Bill Powers, meet Stanley Schachter.

As I understand them, psychoactive substances do not act by altering the
body's biochemical state, but by directly altering activities in the brain.

I don't follow you here. Don't they alter the activities in the brain by
altering the biochemical state of certain parts of the brain?

These drugs are more than "recreational." They are destroyers of
organization. And I think the same is probably true of every psychoactive
drug, no matter how benificent the short-term effects may appear to be.

I agree whole-heartedly with your message that "recreational" drugs can and
do serve to short-circuit normal modes of functioning, allowing the user to
substitute taking the drug for other actions that would normally be required
to achieve similar phenomenal effects. As a result the person may fail to
learn do deal effectively with conditions leading to anxiety, substitute
drugs for family, friendship, and love, and so on. I also agree that simply
prescribing a drug as the avenue of first resort (the quick fix) is to be
avoided unless there is some immediate crisis for which the drug provides an
immediate, though temporary solution (e.g., a person living with
overwhelming dread may be so distracted that he or she simply cannot attend
to therapeutic interventions such as the method of levels). But I also
understand that some conditons can be dealt with effectively (at present) in
no other way, and in such cases the benefit may greatly outweigh the risks.
I don't think anyone here is recommending the cavalier prescription of
psychoactive drugs as the "quick fix" for every problem, but it must also be
recognized that this pharmacological arsenal does have its appropriate
applications for some clients as part of a carefully conducted and monitored
program of intervention.

Sorry, I just can't stay away from that subject, wherever we start talking
about it. I have seen too much misery and too many ruined lives to leave it
alone.

There has also been a tremendous number of lives saved or bettered by it;
one must not forget that either. Otherwise one begins to sound like those
who advocate the banning of innoculation because some have died or suffered
serious side effects from it, ignoring the millions who were saved by such
injections from serious illness and death. As always, one must carefully
weigh the potential costs and benefits in each individual case, and then
hope that the decision made was the correct one.

That's nice. Now, can we get back to talking about PCT? (:->

Regards,

Bruce

[R_Marken]

[From Rick Marken (971222.0930)]

Bruce Abbott (971222.1135 EST) --

Since you're still here (your persistence in the face of our
incorrigibility is truly inspirational) perhaps you could answer
the questions I asked you [Rick Marken (971219.0830)] a couple
days ago:

what do you think are the important discoveries about mind and
behavior that have been made through documented empirical PCT
research over the past 50 years?... perhaps you might also add
some examples of the important discoveries about mind and behavior
that have been made through documented empirical _non-PCT_ research
over the past 50 years.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[Bill_Powers1]

[From Bill Powers (971222.1249 MST)]

Bruce Abbott (971222.1135 EST)--

Put me in the latter category, if by "nothing more and physics and
chemistry," you do not mean to exclude the nervous system's organization.

I do mean to exclude it. Organization, I think, has to be considered
separately from physics and chemistry. It doesn't obey the same laws.

...it is well to remember that this "user and controller" of this
hardware is composed of that hardware. ...
The distinction seems to be one of the hardware itself (the
physical arrangement of matter) and the dynamically changing signals and
states flowing through the hardware as determined through time by its
organization and its inputs.

That will do. Also, before Rick reminds you, in a closed-loop system it's
somewhat difficult to say what "determines" what, since the inputs are also
partly outputs. And the organization has its roots in billions of years of
evolution; it's somewhat futile to speak of "determination" when almost all
of the critical factors are unknown. I don't see why we have to subscribe
to any article of faith on this matter. Why not just say we don't know?

There may also be damage to the "wiring" of the brain which may create
biochemical imbalances even though the individual undamaged components are
regulating properly. A nice example is Parkinson's disease: Neurons in the
brainstem that carry signals to the basal ganglia (motor centers in the
brain) begin to die (for reasons not yet known). These neurons release the
neurotransmitter dopamine at synapses in the basal ganglia, where the
dopamine serves an inhibitory function. When the loss of neurons crosses a
threshold, the motor tremors and muscular rigidity that are characteristic
of Parkinson's disease begin to develop, because the loss of inhibition in
the basal ganglia effectively raises the gain in the motor system to the
point of oscillation.

I'm not convinced that inhibition is anything special -- it's just a
negative sign on a signal, isn't it? I've heard the above description
before, and while it sounds plausible, I don't see any reason to believe it.

Treatment with L-DOPA, a precursor of dopamine,
increases the synthesis of dopamine in the remaining neurons by supplying
more raw material (circumventing the neuron's normal regulation of dopamine
synthesis rate); thus larger than normal amounts of dopamine are released
into the basal ganglia with each impulse, increasing the inhibition there
and bringing about more normal motor function. (Unfortunately, as neurons
continue to die, dosages must be increased, and eventually a point is
reached at which there are no longer sufficient neurons left to do the job.)

Again, very plausible, but does anyone know that this is what is actually
happening?

As I understand them, psychoactive substances do not act by altering the
body's biochemical state, but by directly altering activities in the brain.

I don't follow you here. Don't they alter the activities in the brain by
altering the biochemical state of certain parts of the brain?

I was thinking of the body as that which is not brain. The brain's
biochemistry is rather well isolated from that of the body.

These drugs are more than "recreational." They are destroyers of
organization. And I think the same is probably true of every psychoactive
drug, no matter how benificent the short-term effects may appear to be.

I agree whole-heartedly with your message that "recreational" drugs can and
do serve to short-circuit normal modes of functioning, allowing the user to
substitute taking the drug for other actions that would normally be required
to achieve similar phenomenal effects. As a result the person may fail to
learn do deal effectively with conditions leading to anxiety, substitute
drugs for family, friendship, and love, and so on.

Good. A big agreement.

I also agree that simply
prescribing a drug as the avenue of first resort (the quick fix) is to be
avoided unless there is some immediate crisis for which the drug provides an
immediate, though temporary solution (e.g., a person living with
overwhelming dread may be so distracted that he or she simply cannot attend
to therapeutic interventions such as the method of levels). But I also
understand that some conditons can be dealt with effectively (at present) in
no other way, and in such cases the benefit may greatly outweigh the risks.
I don't think anyone here is recommending the cavalier prescription of
psychoactive drugs as the "quick fix" for every problem, but it must also be
recognized that this pharmacological arsenal does have its appropriate
applications for some clients as part of a carefully conducted and monitored
program of intervention.

This is fine for those responsible individuals who use therapeutic drugs
carefully and with reluctance. It has not been my experience that this is
the majority. The size of the drug industry and the volume of dispensations
rather suggests that drugs are the first resort in most cases, with very
little time spent on evaluating the individual appropriateness of drug
therapy. The most objectionble assumption is that if a drug is useful for
certain extreme cases, it must be useful for all cases in which similar
symptoms are seen, or might be thought likely.

Sorry, I just can't stay away from that subject, wherever we start talking
about it. I have seen too much misery and too many ruined lives to leave it
alone.

There has also been a tremendous number of lives saved or bettered by it;
one must not forget that either. Otherwise one begins to sound like those
who advocate the banning of innoculation because some have died or suffered
serious side effects from it, ignoring the millions who were saved by such
injections from serious illness and death. As always, one must carefully
weigh the potential costs and benefits in each individual case, and then
hope that the decision made was the correct one.

I don't know if the number is "tremendous." I'll agree that it is nonzero.

That's nice. Now, can we get back to talking about PCT? (:->

What a great suggestion!

Best,

Bill P.

[Martin_Taylor4]

[Martin Taylor 971221 16:00]

Tim Carey (971221.1105)]

[Martin Taylor 971219 13:10]

... the bottom line is
that small changes in the environmental coupling can cause large changes
in the complex behaviour of simple networks, and probably also in

locked-up

minds.

I don't understand this ... I didn't think that from a PCT perspective that
anything in the environment _caused_ any kind of behaviour.

The word "behaviour" is the sticking point here. I used it in the sense
that the behaviour of the Earth is that it follows an elliptical orbit, not
in the sense of purposed action. Also, remember my posting on the use of the
word "cause." You can't ever look into that word's use too closely.

What I tried to say was that if you have even a fairly small, simple
network with both positive and negative feedback possibilities, then
the entire dynamics followed by the network (or by any node in it) may
change dramatically when small momentary changes are made to the inputs
of the network (which is all that the network knows of its environment).
Colloquially, small changes in the environment can cause large changes
in the behaviour of the network.

The experiment was to look at the effects on the network behaviour of
providing different kinds or inputs and of changing the gain of the nodes
(gain being the slope of the central portion of the squashing function;
gain represents the coupling among the nodes).

I may have this all wrong, but this paragraph sounds like you are trying
effect the output by varying the input, I thought control systems worked
the other way around ... the varied their output to get the input they
want.

We aren't talking specifically about control systems here. The experiment
was on simple sum-and-squash nodes. But the implications are valid for
networks in which the elements are control systems that interact because
one's actions disturb another's perceptions, "causing" the other to
resist those disturbances by actions that might disturb the original
control system's perceptions. All of the actions of any control system
have the potential of disturbing the perceptions of another control system
that shares the environment. The control systems form a network, in which
there may be both positive and negative feedback loops. If the couplings
are too tight, which can happen if the gains are too high, the network
can lock up, go into oscillation, or go chaotic, regardless of whether
the individual control systems are trying to act so as to stabilize their
own perceptions. Their actions to control may or may not be successful,
but their success is more likely if the cross-coupling is looser than
if it is tighter.

Such networks usually show some form of locked-up behaviour.

My whole point in the first post is that no-one is thinking of examining
what variables Leo might have been controlling for, again, the statement
above seems to be focussed on the output as though that is what the system
is "doing". Was this systm unable to control at this point?

Apparently. His description sounds very much like it.

I used the analogy of the feedback squeal you get when a loudspeaker is
too close to the microphone that feeds it. It doesn't matter what is said
into the microphone--the same squeal comes out of the loudspeaker. And it
doesn't matter what frequency the squeal is--the solution is to reduce the
coupling between microphone and speaker. I propose that the situation is
analogous.

But that's not the end of it. It is very easy to find values of the
_hardware_ parameters for which the network shows as many as five

different

behaviours (in our 8-node network), and it can switch from one behaviour
to another if you input, say, a brief low-level impulse into just one of
the nodes at a critical moment. Even when the network is apparently
locked up so that all the nodes take on fixed, extreme values, a small
impulse at one input can shift it into oscillating or chaotic behaviour.

Again, the above paragraphs seem to be trying to explore aspects of a
system's output. I was more interested in Leo's input.

If my speculations are anywhere near correct, Leo's input does matter,
but only insofar as some inputs allow him to alter the feedback dynamics,
and may allow him to regain control. But then, unless there is a hardware
alteration, some other input could equally easily put him back into the
locked-up condition in which control is impossible. Under those conditions,
it's probably better to fix the hardware than simply to discover which
inputs can kick him out of the lock-up.

Now, I grant that _when he is controlling_ control failures in part of the
hierarchy can lead to reorganization, and that reorganization might be
enough to eliminate the lock-up possibility. But how do you ensure that
the reorganization doesn't carry the hardware into a condition where
lockup is more, rather than less, likely? I'd say that you do that by
enhancing the likelihood that the system is not locked up, by reducing
the coupling overall, and then allowing reorganization to work. _On the
whole_ reorganization moves in the direction of better control, but in
the short term, it might very well move in the other direction. And when the
hardware has a high likelihood of going into lock-up anyway, that's
not safe.

I'm afraid I don't understand positive feedback and it's role in control
systems.

Positive feedback is nasty in control systems. It's quite simple if you
understand how a control system works. In a control system, or more
generally in a negative feedback loop, the effects of an infinitesimal
change in the value of the signal anywhere in the loop vanish over time.
The loop returns to its stable state. In a positive feedback loop, the
effect of an infinitesimal change in the value of any signal will
snowball over time, becoming as large as the physics of the loop will
allow--at least if the feedback gain is greater than unity. When that
happens, the externally visible behaviour depends only on the loop itself,
not on the inputs to the loop (whether they be reference signals to a
control system, or disturbances from outside).

Usually what happens in a positive feedback situation is that some parameter
of the loop changes when the signal gets large. For example, the output
of some amplifier (and there must be at least one in the loop) cannot
go higher than some limit, so the loop stabilizes when the output of that
amplifier eaches the limit. More dramatically, the signal values may
increase until some component of the loop literally explodes (Bang! Smoke!).
Either way, the positive loop gain goes down at least to unity (to zero
in the case of explosion:-).

More commonly, and less dramatically, what happens is that the sensitivity
of some amplifier to changes in its input is reduced at high signal values,
so that whereas a 1 mv signal change at 1v may cause a change of 10 mv
in the output, a 1 mv signal change at 10v may cause an output change
of only 0.1mv. We call such an amplifier a saturating amplifier, in which
the dynamic gain reduces as the signal value increases. As the signal
value gets larger, the contribution of the amplifier to the loop gain
gets less and less, until eventually the dynamic loop gain comes down
to unity, and the signal stops getting bigger. Such a positive feedback
loop may well have two stable states, one in which the signal at point X
has a high positive value, and one where the signal at that point has
a large negative value.

A flip-flop is a prime example of the kind of positive feedback loop
that has these two stable states. The basic flip-flop has two amplifiers
connected so that a positive change in the output of one induces a
negative change in the output of the other (which induces a further
positive change in the output of the first, and so forth). The two
amplifiers are both of the saturating kind, so at some point the
output of each stabilizes, one at a large negative value, the other
at a large positive value.

In a flip-flop in an electronic circuit, each amplifier has an independent
input from outside, as well as its input from the other amplifier. If the
input from outside to amplifier A is positive while the input coming from
the other amplifier is negative, the output of amplifier A can get less
positive (remember that inputs are inverted at the output), which tends
to make the output of amplifier B less negative, making the output of
amplifier A less positive, making... The states of the two amplifiers
swap, so that amplifier A goes negative and amplifier B goes positive.

Oscillators also work by positive feedback. In an oscillator, the critical
point is that the feedback involves some delay.

Suppose you feed a very tiny sinusoid into the input of an amplifier, and
the output of the amplifier is fed back to the input with some delay deltat.
At time t0, let's say that the sinusoid you input is at a positive peak
value. Now, if at time t0+deltat, the output comes back positive and
amplified, while the input is again at a positive peak, the effect is
the same as if you had input a bigger sinusoid. And again at t0+2*deltat,
and so forth. At t0+0.5*deltat, the input sinusoid is at a negative
peak, and so it is at t0+1.5deltat, when the amplified version of that
peak comes around again, making the sum even more negative. And so it
goes, the sinusoid getting bigger without limit, until the amplifier
saturates (or explodes). The input sinusoid isn't even needed to keep the
oscillation going. The tiniest noise at the input can start the oscillation
going, and it keeps itself going after that.

With more complex circuits of amplifiers interconnected in different ways,
with some positive and some negative interconnections, and with delays in
the effect of one on itself through the others, very complicated behaviours
can occur.

It is very easy to see how such a system of interacting control systems
with coupling that is too tight (either because of high gain in the
individual control loops or because of high mutual disturbance among
the systems) can behave in a multititude of ways depending on the
environmental conditions.

Here it sounds as though the environment is controlling the system's
output. What am I missing?

Does the above help? The environmental conditions are the coupling weights
and whatever inputs may disturb the network.

The experiments were done with simple amplifying nodes. The analogy with
control systems is that a control system's actions disturb other control
systems, and those disturbances "cause" countering actions in the other
control systems. Those actions also disturb yet other control systems,
including the one originally disturbed (with, of course, some delay). It
depends on the gains of all the control systems and their efficiency of
control whether there are any positive feedback loops with gains greater
than unity in the whole structure.

The ideal situation is when reorganization has built a structure in which
actions are effective in control, with minimal cross-disturbance of other
control systems. That allows the loop gains of the individual control systems
to be high (good control) without generating high-gain positive feedback
loops. But with a given structure, it is possible to avoid positive
feedback loops by reducing the gains overall. I think that's what David
was talking about when he said the drug therapy reduced the coupling,
which is why I intervened originally.

When there is a perception that is important
to control, things can change in the potentially locked-up hardware
structure.

This was perhaps the whole point of my post. That when there was something
important, Leo could control effectively. I would be interested in research
that explored this ability to control in more detail.

It can be simulated, perhaps. But I'm not at all sure that it's a general
phenomenon. It may be. I can well imagine that there are points in the
structure that participate in a positive feedback loop when they vary,
but would break the loop if they were externally stabilized. In an
electronic oscillator, it would be like clamping the voltage of some
point on the delayed feedback path.

I don't find it either surprising or curious that an individual with a
hardware problem that is not too severe can find that the problem goes
away when there is something important to do, and can, with effort,
control so as to suppress the lockup -- for a while. To perceive that
a lockup is happening or about to happen may well be a perception for
which control can be learned.

I've always thought of epilepsy as a problem with hardware and even though
I've known people with epilepsy, I have never known them to be able to make
a seizure go away if, at the time it occurs, there is something more
important to do. Would diabetes be similar?

I rather had the impression that the diabetes analogy would be the removal
of the output amplifier from the control loop--the insulin cannot be
produced even though the signals for producing it are there. Maybe I'm
wrong--I know nothing about it medically.

The epilepsy clamping seems to me to be more analogous to a flip-flop
in which the loop gain is so high that there isn't an input available
that can reverse the state of the flip-flop, or, to change the analogy,
an oscillator driving so hard that no available signal can clamp the
feedback loop. However, in the regard, one might note that certain
pulsing visual inputs can start the epileptic loop working in its
locked-up mode. It is conceivable that there might be a way to phase some
input that would get it out of that mode. I don't know.

Actually, "I don't know" is a pretty good tag line to all of this:-)

Martin

[Martin_Taylor4]

[Martin Taylor 971223 12:45]

Rick Marken (971221.1010)]

Martin Taylor (971219 13:10) --

I think what Tim Carey and I were questioning about Leo Kay's
testimony was the assumption that the apparent success of
drugs indicated that a _hardware problem_ had been solved.

One can't, of course, tell. And it really depends on what you call
"hardware." I call hardware the system of connections, with their
coupling weights, gains, and all the rest of the parameters from
which the behaviour of a complex system might be simulated. It's
not like a Von Neumann computer, in which program (software) and data
(software?) can be distinguished from silicon and copper (hardware).
In a neural network (real or simulated), the structure of the network
takes on the function of program and silicon, both, leaving only
data (values on signal lines) to be something else (software?).

Tim's point was (I think) that you simply can't tell by just
looking at what are seen as "manic-depressive" outputs what a
person is doing (trying to control), whether there is even a
problem, whether all problems classified as "manic-depressive"
are the same (as is implied by placing them all in the same
diagnostic category) or whether the problem (if it exists)
is the result of a hardware or software malfunction.

I'd agree with that, but I didn't think I even mentioned or contemplated
"manic-depressive" syndrome (a syndrome which, in myself, on the advice
of a biochemist friend, I seem to have successfully treated with occasional
potassium chloride ingestion). I was dealing with obsessive-compulsive
behaviour, in which the externally observable symptom is repetitive
action that has no apparent relation to what the patient seems to
want to control--just as the squeal of a loudspeaker has no apparent
relation to the words spoken into the microphone.

But these systems are more complex, and I pointed out that even trivial
systems can behave in wildly varying manners with no change in the
hardware. Only the data change. But the hardware is what permits these
different behaviours. Different hardware, even by simply reducing one
or two gains, gives you vastly different repertoires of behaviour.

Assuming that the problem is what you call a "locked up positive
feedback loop" you still don't know that it's a "hardware"
problem. The positive feedback that occurs briefly in the
"Levels of control" demo (for example) is not the result
of any hardware problems (which I take to be organic failures
in the case of living systems, like neurons that won't fire or
that won't stop firing). Perfectly normal control systems can
oscillate or "lock up" (at maximum output) simply because they
are in conflict (controlling the same variable relative to
different references).

Exactly what I call a hardware problem! It's a problem to which coupled
systems are prone, by virtue of their hardware construction. And small
changes in the data (either reference or disturbance inputs, for example)
have the potential of making great changes in the behaviour--including
possibly the ability to control.

The fact that drugs can help people with control problems
(depression is a control problem) does not mean that the
problem was, indeed, a "hardware problem". It may be,
and I believe that in most cases it is, a "software
problem"; that is, one that can be "fixed" (in principle)
by reorganization (a process that is likely, by the way,
to involve hardware changes).

I think I tried to get across the idea that if the problem is a hardware
lockup (or if you prefer, a software lockup in the hardware),
then the effect of Prozac ought to be to permit reorganization to occur
safely (that is, if David's description of the effect of Prozac is
correct).

In other words, I think I agree with what you are saying, but not with
the language in which you cast it. Does that make any sense?-)

Martin

[Tim_Carey]

[From Tim Carey (972412.0800)]

[Martin Taylor 971221 16:00]

Hi Martin,

Thanks for your post ... I'm still trying to understand the concepts you
talk about. I must admit that I've been so concerned with learning the
basics of PCT (and I'm still not sure if I have ... this is such a slippery
thing) that often I have skipped the more technical conversations. The
discussions a while ago on flip-flops is a good example. I didn't feel I
was ready to tackle such technical stuff. Even now it seems way over my
head but since you've addressed some of my issues personally I figure I'd
better take the plunge sometime ..... consider this a plunge. At this stage
though, I feel qualified in most instances to only ask questions a lot of
what you talk about I'm not able to discuss because I don't understand it
yet. I hope you don't find this post too tedious ....

What I tried to say was that if you have even a fairly small, simple
network with both positive and negative feedback possibilities

What do you mean by a network? Is this a group of individual systems? Are
they within one person or do you mean control systems in different people?
Are there levels in this network as in the HPCT model? What do you mean by
positive feedback possibilities ... I know you explain what positive
feedback _is_ a little later, but I thought the functional unit of a
control system was a negative feedback loop.

the entire dynamics followed by the network

What do you mean by this statement? How does a network "follow" dynamics?
What are the dynamics that the network follows?

Colloquially, small changes in the environment can cause large changes
in the behaviour of the network.

Assuming the changes in the environment cause anything at all, I would
presume that the changes would be on a continuum from "no change" to "lots
of change"?

We aren't talking specifically about control systems here.

Actually Martin, control systems are all I'm interested in. When I've got a
good enough handle on how they operate I might be interested in considering
other things but if I understand the model of PCT at all, the organisation
of a control system explains the behaviour of _all living things_ .... this
is enough for me to deal with at the moment.

But the implications are valid for
networks in which the elements are control systems that interact because
one's actions disturb another's perceptions, "causing" the other to
resist those disturbances by actions that might disturb the original
control system's perceptions.

Are we speaking about control systems within one person or between people?
Aren't one's actions caused _jointly_ by environmental disturbances and the
reference of the control system under consideration? So is it fair to say
that a disturbance is always defined (at least in part) by the reference of
the control system being disturbed?

If the couplings

What's a "coupling"?

>My whole point in the first post is that no-one is thinking of examining
>what variables Leo might have been controlling for, again, the statement
>above seems to be focussed on the output as though that is what the

system

>is "doing". Was this systm unable to control at this point?

Apparently. His description sounds very much like it.

But how do we know? My point is that there is currently no research being
done that I know of to investigate this phenomenon?

I used the analogy of the feedback squeal you get when a loudspeaker is
too close to the microphone that feeds it. It doesn't matter what is said
into the microphone--the same squeal comes out of the loudspeaker. And it
doesn't matter what frequency the squeal is--the solution is to reduce

the

coupling between microphone and speaker. I propose that the situation is
analogous.

Would you describe the feedback squeal as output? If so it sounds like you
are manipulating what goes in, to get the output that you want (reduced
squeal). Is this consistent with PCT? I thought in PCT we manipulated our
output to get the input we wanted?

If my speculations are anywhere near correct, Leo's input does matter,
but only insofar as some inputs allow him to alter the feedback dynamics,
and may allow him to regain control. But then, unless there is a hardware
alteration, some other input could equally easily put him back into the
locked-up condition in which control is impossible. Under those

conditions,

it's probably better to fix the hardware than simply to discover which
inputs can kick him out of the lock-up.

I have a lot of questions about this paragraph. What are feedback dynamics?
What do you consider "hardware"? Isn't getting the "right" input what a
control system "does"? I don't understand what you mean by "some inputs
allow him to alter the feedback dynamics" ... how do inputs "allow" us to
do anything? Doesn't a control system compare inputs to references?

I'd say that you do that by
enhancing the likelihood that the system is not locked up, by reducing
the coupling overall, and then allowing reorganization to work.

How would you do this?

And when the
hardware has a high likelihood of going into lock-up anyway, that's
not safe.

Again, what do you mean by hardware?

Positive feedback is nasty in control systems.

OK, I understand how positive feedback works, but as I said earlier I
thought a negative feedback loop was the functional unit of a control
system. How do you get a positive feedback loop? Can a negative feedback
loop turn into a positive feedback loop? Can they change back?

More commonly, and less dramatically, what happens is that the

sensitivity

of some amplifier to changes in its input is reduced at high signal

values,

What's an amplifier?

A flip-flop is a prime example of the kind of positive feedback loop
that has these two stable states.

What's a flip-flop? Is it a special kind of control system or two control
systems interacting or is it something completely different? If it's
something completely different, I may be interested in learning about it at
a later stage but at the moment I just want to talk PCT.

Oscillators also work by positive feedback. In an oscillator, the

critical

point is that the feedback involves some delay.

What's an oscillator? I guess all the above questions about flip-flops
apply here as well?

The input sinusoid isn't even needed to keep the
oscillation going. The tiniest noise at the input can start the

oscillation

going, and it keeps itself going after that.

Again Martin, in this example it sounds like getting the right output is
the focus. Is that what PCT is about? I don't understand how you "feed"
input into a system. I thought that a control system was a closed causal
loop and the two independent variables were the reference and the
disturbance. Do you mean that you apply disturbances? Again, wouldn't you
be applying disturbance to see how actions varied to keep a paticular input
(that you suspect might be controlled) constant?

The experiments were done with simple amplifying nodes. The analogy with
control systems is that a control system's actions disturb other control
systems, and those disturbances "cause" countering actions in the other
control systems.

As I said earlier, to speak of actions being "caused" as well as
considering the disturbances to the system don't you also need to consider
the reference of the system?

Those actions also disturb yet other control systems,

including the one originally disturbed (with, of course, some delay). It
depends on the gains of all the control systems and their efficiency of
control whether there are any positive feedback loops with gains greater
than unity in the whole structure.

Doesn't it also depend on the references of the control systems?

I think that's what David
was talking about when he said the drug therapy reduced the coupling,
which is why I intervened originally.

My point is ... how do we know? I'm very wary of applying PCT jargon to
case studies and research that has been conducted from within the paradigms
of conventional psychology.

Actually, "I don't know" is a pretty good tag line to all of this:-)

I agree. Will we ever "know" until phenomena are explored from a PCT
perspective?

Cheers,

Tim

[Martin_Taylor4]

[Martin Taylor 971224 2215]

Tim Carey (972412.0800)

[Martin Taylor 971221 16:00]

Hi Martin,

Thanks for your post ... I'm still trying to understand the concepts you
talk about. I must admit that I've been so concerned with learning the
basics of PCT (and I'm still not sure if I have ... this is such a slippery
thing) that often I have skipped the more technical conversations.

Many of your questions do leave that impression. But if you want to learn
PCT, you _have to_ learn about control. One of the best ways to do that
is to try out Bill Powers' suite of tutorial demonstrations (if you have
a PC-compatible). The demos lead by easy stages through learning what
control is. In deference to the idea that you will attempt Bill's
demos, I will answer only a few of the more basic questions you ask. If
you still have more questions after having gone through Bill's suite, I'll
try answering them then. Go to http://animas.frontier.net/~powers_w/
or ftp://burkep.libarts.wsu.edu/csg/ if you can't do ftp from your
browser. Take the sets of files in "demo1" and "demo2" in "billdemos".
The other demos are for later, after you understand control a bit
better.

What I tried to say was that if you have even a fairly small, simple
network with both positive and negative feedback possibilities

What do you mean by a network?

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. In a road map,
the important properties might be the labels for the cities and the
driving time between them--or something else. In my experiment, the
entities were simple sum-and-squash nodes and the connections were
the links between them. The important properties (the "couplings")
were the weights that determined how much of the output of one node
appeared at the input to another. Like the road map, in which the
coupling between two cities might be the inverse of the driving time
between them (closer cities are more tightly coupled).

Is this a group of individual systems?

Yes, including the degree to which each affects another.

Are
they within one person or do you mean control systems in different people?

Yes. Either.

Are there levels in this network as in the HPCT model?

No. And it's not a variant of the HPCT model. I have been using it to
suggest what may happen when several organisms, each a hierarchic
perceptual control system, interact. I have also been using it to
suggest what may happen when several elementary control units within
one hierarchic control system interact. The first is "social psychology,"
the second is an aspect of individual psychology.

What do you mean by
positive feedback possibilities ... I know you explain what positive
feedback _is_ a little later, but I thought the functional unit of a
control system was a negative feedback loop.

It is. But when two control systems interact, the interaction may well
bring a postive feedback loop into being. That's what happens, for instance,
in a direct conflict. If the two control systems have output functions that
are integrating amplifiers, the output of each just keeps growing without
limit, as each output is a disturbance to the other's CEV, and each
requires that the other increase its output if the error is not to
increase.

the entire dynamics followed by the network

What do you mean by this statement? How does a network "follow" dynamics?
What are the dynamics that the network follows?

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. The swing has two dynamic possibilities.
In one, it moves back and forth under the branch; in the other, it swings
over the top and round and round without changing direction. Which
dynamic it follows depends on the energy in its motion when it passes
the bottom of its arc. Now imagine that there's another similar swing
humg from the same branch. These swings are _coupled_. Any motion of one
swing moves the branch and through the branch moves the other swing.
If the first swing is moving back and forth, and the other is initially
stationary, the second swing will start moving back and forth in
synchrony with the first. The _system_ can follow a dynamic in which the
two swings are moving synchoronously (remember, these are _similar_
swings--it won't work if they are much different).

The network follows much more complex dynamics. And under some hardware
conditions (i.e. particular values for the gains and coupling weights
in the network) it can follow a variety of different dynamics, even
of different classes, switching from one to another after slight
disturbances.

Technically, there are three classes of dynamic. The simplest is called
a "fixed point." In a fixed point dynamic, if the system is left to
itself, it comes to rest in some fixed state eventually (as will
the swing because of the air friction I neglected in my description).
The second is called an "oscillator dynamic." The system never comes
to rest, but repeats a series of states over and over again, if it is
left to itself. Often some point in the system is used as an output,
to provide a useful waveform. You can buy boxes that produce sinusoids,
square waves, triangle waves, and the like. Inside the box there is
an oscillator. The third class of dynamic is "chaotic." In a chaotic
dynamic, the system never exactly repeats a state it has been in before,
although most chaotic systems almost repeat themselves. so that they
can look for a long time as if they are oscillators, if you don't look
closely and accurately. One characteristic of chaotic dynamics is that
if you wait long enough, the most infinitesimal difference in initial
conditions will show up eventually as the largest possible difference
in conditions.

Colloquially, small changes in the environment can cause large changes
in the behaviour of the network.

Assuming the changes in the environment cause anything at all, I would
presume that the changes would be on a continuum from "no change" to "lots
of change"?

No. That's the important point about all this. What you say is often true,
but there are important exceptions, sometimes called "bifurcation points".
Think of the swing, but imagine that instead of a rope it has a rigid
suspension rod. Now push the swing a little. It moves back and forth.
Push it a little harder. It moves back and forth over a slightly wider
arc. Push it quite a bit harder. It swings up to nearly vertical, and
back again to nearly vertical on the other side, but it still swings
back and forth. Now push it ever so slightly more than that. This time,
it doesn't come back, but swings right over the top and round and round
for ever (in the absence of friction). That's a bifurcation point, and
the real world in which control systems operate is full of them.

We aren't talking specifically about control systems here.

Actually Martin, control systems are all I'm interested in.

Notice I said "specifically." Control systems are included, but the
discussion is more general.

When I've got a
good enough handle on how they operate I might be interested in considering
other things but if I understand the model of PCT at all, the organisation
of a control system explains the behaviour of _all living things_ .... this
is enough for me to deal with at the moment.

But it is wrong. You said "the organisation of a control system..." But
PCT deals with the organization of a hierarchy of interacting control
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). If the output is hitched up oppositely, the system will have
positive feedback and won't control. If it is hitched up too strongly,
the system will oscillate, and won't control. And that's only with one
control system. When you have two, there are other problems, and when
you have a whole hierarchy, things get _very_ interesting. That's the
area on which I'm trying to get a handle.

But the implications are valid for
networks in which the elements are control systems that interact because
one's actions disturb another's perceptions, "causing" the other to
resist those disturbances by actions that might disturb the original
control system's perceptions.

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

Yes.

Aren't one's actions caused _jointly_ by environmental disturbances and the
reference of the control system under consideration? So is it fair to say
that a disturbance is always defined (at least in part) by the reference of
the control system being disturbed?

That's a misreading of the idea of PCT. The word "disturbance" has caused
problems in the past, and may do so again in future. A control loop has
two places where input comes from outside. One is the reference signal
input to the comparator, the other is the disturbance signal influence
on the Complex Environmental Variable defined by the perceptual input
function. The source of the disturbance signal is independent of anything
in the control loop, and is unknowable to anything in the control loop.
Sometimes the word "disturbance" has been taken to mean the disturbance
signal "wire", sometimes the disturbance signal value, and sometimes the
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."

I used the analogy of the feedback squeal you get when a loudspeaker is
too close to the microphone that feeds it. It doesn't matter what is said
into the microphone--the same squeal comes out of the loudspeaker. And it
doesn't matter what frequency the squeal is--the solution is to reduce

the

coupling between microphone and speaker. I propose that the situation is
analogous.

Would you describe the feedback squeal as output? If so it sounds like you
are manipulating what goes in, to get the output that you want (reduced
squeal).

No. The point is that manipulating the input has almost no effect on the
squeal. It is determined by the properties of the loop.

If my speculations are anywhere near correct, Leo's input does matter,
but only insofar as some inputs allow him to alter the feedback dynamics,
and may allow him to regain control. But then, unless there is a hardware
alteration, some other input could equally easily put him back into the
locked-up condition in which control is impossible. Under those

conditions,

it's probably better to fix the hardware than simply to discover which
inputs can kick him out of the lock-up.

I have a lot of questions about this paragraph.

Yes, and I don't think I can answer any of them in a way that will make
much sense to you unless you understand all the foregoing. If you do
understand the foregoing, I think you will not want to ask these
questions. If you understand, and still want to ask them, ask again.

What's an amplifier?

You really do have to know that sort of thing if you are going to try
to understand PCT. An amplifier is something that has an input and an
output, for which the output is larget than the input, but related to
it. A simple amplifier has an output o =, G*i where i is the input. The
factor G is called the gain of the amplifier. There are other types.

The input sinusoid isn't even needed to keep the
oscillation going. The tiniest noise at the input can start the

oscillation

going, and it keeps itself going after that.

Again Martin, in this example it sounds like getting the right output is
the focus.

There's no "right" or "wrong" output. I merely describe what happens
with certain connections. You surely wouldn't want your control systems
connected so that they become an oscillator, would you:-) If you connect
them in some ways, they would, and then they would presumably be
providing the "wrong" output--at least by somebody's criterion.

The experiments were done with simple amplifying nodes. The analogy with
control systems is that a control system's actions disturb other control
systems, and those disturbances "cause" countering actions in the other
control systems.

As I said earlier, to speak of actions being "caused" as well as
considering the disturbances to the system don't you also need to consider
the reference of the system?

Yes, but in this case all you need to assume is that there exists some
reference, which is the case if a control system is actually controlling.
If there is a reference value that changes reasonably slowly, the output
of a control system, as experienced at the CEV, will correlate highly
and negatively with the disturbance value. That's "cause" in the
colloquial sense--and as I said, you can't ever look too closely at the
meaning of that word.

Those actions also disturb yet other control systems,

including the one originally disturbed (with, of course, some delay). It
depends on the gains of all the control systems and their efficiency of
control whether there are any positive feedback loops with gains greater
than unity in the whole structure.

Doesn't it also depend on the references of the control systems?

No. Or rather "no, if the system is linear." But actually the nonlinearities
are important here, and it could easily happen that with some settings
of the reference values in some of the interacting control systems, the
lockup could disappear--at any rather the network dynamics in a simulation
could well change.

I think that's what David
was talking about when he said the drug therapy reduced the coupling,
which is why I intervened originally.

My point is ... how do we know? I'm very wary of applying PCT jargon to
case studies and research that has been conducted from within the paradigms
of conventional psychology.

I'm not talking jargon. I'm talking the results of very simple simulations
of simple networks that show behaviour that looks a bit like what David
described, and which show the same kind of escape from lock-up when the
coupling in the simulation is reduced that David described as happening
"when the coupling is reduced" by Prozac.

I have no idea whether what happens in the simulations is actually what
happens in the biological control structure that is in trouble. I suggest
it as a possibility, though obviously the interaction of the myriads of
control systems in the person is much more complicated.

As an afterthought, you might understand all this a lot better if you were
to look up Stuart Kauffman's "At Home in the Universe" as well as doing
Bill Powers' demos at http://animas.frontier.net/~powers_w/.

Martin

[Tim_Carey]

[From Tim Carey (972612.0720)]

[Martin Taylor 971224 2215]

Many of your questions do leave that impression. But if you want to learn
PCT, you _have to_ learn about control.

Hi Martin,

Thanks for your reply ... 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. I have been communicating privately
with Tom Bourbon for about 18 months, 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. 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. Perhaps you
just use different terminology to the other guys but you mention a lot of
stuff that I've never talked about with Tom and so I'm trying to work out
what I need to learn and what I don't.

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.

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. Can you use examples
that are organised like a control system? I believe a thermostat is
organised that way.

Notice I said "specifically." Control systems are included, but the
discussion is more general.

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.

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?

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.

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

Regards,

Tim