Levels of analysis

[From Bill Powers (2003.07.07.1100 MDT)]

For those who are worrying about the dwindling list, there are now 128
subscribers, two of them concealed. 96 of them are in the USA. At least one
subscriber is a newsnet list.
I think the most we ever had was about 145. Some listings are so old that I
seriously doubt their validity.

I have finally understood where Marc Abrams is coming from. He has got into
the literature of superreductionists, people who study calcium channels,
intracellular signaling, and that sort of thing. I have no word of
criticism for such work, but I do have cautions for anyone who thinks this
is the "real" picture of human organization. It tells us no more about
human organization than grinding up a radio and doing a chemical. analysis
of it would tell us about amplifiers, automatic gain control, tuned
circuits, or loudspeakers. It might tell us what these things are made of,
but it would not tell us the organization that makes them function as they do.

Nature is organized at many levels, even non-living nature. There are laws
that apply at each level, but the laws of one level do not reveal the laws
of the next level up (or down). You can understand how transistors work,
but to understand how a radio works you have to know how the transistors
are connected to each other and to passive components of the circuit. You
can't get that information from looking at a transistor, or a hundred
transistors. It is the organization of the circuit that makes it into a
radio. The same parts, differently organized, might make a metal detector
or an audio signal generator. Going the other way, the physical and
chemical properties of any one component represent emergent laws that grow
out of the properties of quarks and other fundamental particles. You would
have to know how the quarks are organized in order to find the laws at the
level of chemical reactions.

The level of organization represented by PCT is to cell chemistry as the
operation of a radio is to the holes and electrons in silicon, the atomic
matrix in copper, the laws of electromagnetism in an inductor, At an
intermediate level of analyzing the radio, we have components like
transistors, capacitors, inductors, resistors, and insulators, which can be
further analyzed into the detailed physics and chemistry of matter. At an
intermediate level in PCT, we have neurons, neural signals, muscle fibers,
hormones, and secretions, which can be further analyzed into underlying
biochemical and cell properties.

A neural comparator, for example, can be as simple as a neuron which
receives impulses that excite it to fire at the same time it receives, from
a different input connection, impulses that suppress the firing rate. The
net firing rate of the neuron is proportional to the difference between the
total amounts of excitation and inhibition. The net effect drops just to
zero when excitation equals inhibition. The firing rate thus represents the
difference between the excitatory and inhibitory inputs in terms of the
rates at which impulses reach the neuron. An analysis of an artificial
control system at the same level would say that a silicon comparator is a
transistor (or integrated circuit) with one input where incoming signals
act positively to increase the rate at which electrons pass through the
transistor, and another input where incoming signals act negatively
to decrease that rate. Thus the rate at which electrons are emitted from
the transistor represents the difference between the positive and negative
input effects, the function that a magnitude comparator has to carry out.

A chemical comparator can be made with allosteric enzymes (and other ways).
I'll skip the details, though they are known.

The neural and electronic circuits perform the function of magnitude
comparison, reporting a match as zero ouput and a mismatch as an output
signal of a particular magnitude. In both cases, a pair of such devices
would be needed to report mismatches of both signs.

In both cases, the operation of comparison can be further analyzed into
more detailed phenomena. An electronic comparator might be made of a pair
of field-effect transistors. In a field-effect transistor (there are
several kinds), we have a voltage applied to an insulated electrode,
setting up an electric field that either repels electrons from entering the
gate region, or attracts them into the gate region. A positive voltage on
the drain (collector) sweeps up all electrons that get into the gate
reqion. These processes can be analyzed further into states of excitation
of the electrons at the quantum level, and processes at that level can be
analyzed into phenomena of quantum chromodynamics. Of course as we pass to
each lower and more detailed level of analysis, we lose the ability to
predict phenomena at the higher level, because we lose the organization of
relationships among different elements at the higher level. At the ;eve;
opf transistors, analyzing a field-effect transistor does not tell us that
there are two transistors connected in a common-source configuration. At
the level of quantum effects, there is no such thing as a transistor.

In the nervous system, the comparator can be analyzed in terms of
neurotransmitters that alter the ionic concentrations inside the receiving
neuron, changing the electrical potential (voltage) at the axon hillock.
This basic potential is maintain by continuously operating calcium pumps,
and is modified by the effects of incoming neurotransmitters, both
excitatory and inhibitory. When the potential at the axon hillock exceeds a
specific threshold, there is an electrical breakdown, which triggers a
breakdown in adjoining parts of the axon, and sends a wave of breakdown
traveling along the axon at anywhere from 1 to 100+ meters per second,
depending on the axon diameter. All the breakdowns are quickly restored by
the calcium pumps, but another impulse can't be generated until the
potential has risen enough, which takes a millisecond or so. These
processes can be analyzed further into the properties of organic molecules
which fold and unfold to open and close pores and even act like tiny motors
carrying molecules from one place to another, and from there we can go even
further into quantum mechanics. As we pass downward from level to level, we
lose the laws that operated at each previous level and begin to study new
laws, all of which are basically laws of organization.

Obviously, one could never learn how a radio works by studying flows of
electrons through transistors and other physical components. Even to start
to understand what is going on inside a radio, we have to get some idea of
what each part of the total system _does_. For example we will find
capacitors that store electrical charges, and inductors that generate
magnetic fields when electrons flow through them. In certain places we will
find a capacitor with its leads connected across the ends of an inductor,
so we have electrons being stored in the capacitor but quickly flowing
through the inductor as the capacitor discharges. We will also find that
current flow through the inductor generates a magnetic field which quickly
collapses, and in collapsing drives electrons out of the inductor -- and
into the capacitor. Clearly there is some special relationship here between
these two components that can't be understood by studying either one by
itself. What they form together, as some here will know and some won't, is
a "tank circuit," and this tank circuit can be used to tune in specific
frequencies of incoming electromagnetic radiation. When you turn the tuning
knob of a radio, you are changing the characteristics of a variable
capacitor connected across an inductor, and so changing the natural
frequency of a tank circuit (in a "local oscillator" circuit).

The rest of the radio is similarly composed of components interacting with
each other to perform specific functions: gain control, intermediate
frequency amplification, audio rectification, audio amplification, and
conversion of electrical signals into audible sounds. None of these
functions is visible in any one component; they are carried out by
components acting together according to their individual properties. Each
function generally uses components like those used to carry out other
functions; it is the organization more than the component that determines
what is done.

Now apply this to the neural control system. A network of neurons can
create an output neural signal having a frequency that is proportional to a
weighted sum of a set of input neural signal frequencies. This function can
convert a set of perceptual signals into a new signal representing a
property of the whole set -- a new and more general perception. That signal
can enter a comparator along with a reference signal to produce a signal
indicating the difference in frequencies between perception and reference,
and the resulting difference signal or error signal can be used to operate
actuators, muscles, so as to reduce the difference between perception and
reference. The result of this whole set of processes all being connected
(in the correct way) into a closed causal loop is the phenomenon we call
"control". Control is not a property of any one component in this
organization. It is a property of the whole organization. It can't be found
by examining any part in greater detail. In fact if we drop to the level of
calcium channels and intracellular signaling, we will completely lose sight
of what this whole assemblage of components _does_, which is to control its
own inputs and keep them in a match with the reference signal it is receiving.

In trying to understand human organization. we must understand each level
of organization before we can begin to understand the level below. Of
course as our picture of underlying processes becomes clearer, we can
profitably modify our concept of the higher level, too. If we try to
penetrate too many levels in one jump, we will become lost in
incomprehensible details -- just look at the reports in Nature or Science,
where we are snowed under by blizzards of unconnected facts, or facts with
connections only over a very tiny fraction of the whole picture. That will
basically get us nowhere, and people have been complaining about this as
long as I've been reading the literature.

A last note. People use the word control in different ways, which include
the meanings of "to limit," " to influence or affect," and "to determine."
In PCT it means none of those, so we would never say, for example, that
opening a calcium channel controls the flow of calcium ions. If something
is controlled, under the PCT definition, then any disturbance tending to
alter it will be opposed by a change in the action of the control system
that is affecting it. To do this, a closed-loop system and not a
single-stage causal link is always required. A variable can't be controlled
under this meaning unless something senses its state, compares what it
senses against some reference standard, and acts on the basis of any
discrepancy to change its action on the controlled variable in the right
direction for negative feedback.

There is nothing wrong with studying the details, either of electronic
circuits or boplogical systems. But true understanding is not to be found
at any one level of analysis.

Best,

Bill P.

[From David Goldstein (2003.07.07.2043 EDT)]
[From Bill Powers (2003.07.07.1100 MDT)]

Bill and listmates,
Does the levels of analysis arguments apply when going to higher levels?
For example, the items in the Areq set of items I sent you are the kinds
of perceptions that clinicians utilize all the time when thinking about
people who seek help. These are the "atoms" that they work with.
David

[From Bill Poweer4s (2003.07.07.1926 MDTZ)]

David Goldstein (2003.07.07.2043 EDT)--

Does the levels of analysis arguments apply when going to higher levels?
For example, the items in the Areq set of items I sent you are the kinds
of perceptions that clinicians utilize all the time when thinking about
people who seek help. These are the "atoms" that they work with.

I hope it was clear that these aren't the levels in HPCT; they're probably
more like sublevels at the HPCT logic level.

I can't make much of the list of items -- a lot of them seem to be about
conflict, and others are about how people try to resolve or at least cope
with conflicts. But I don't see how getting people to rate themselves using
these statements gives us any idea of what's wrong. I suppose this could be
called a higher level of analysis, but isn't there such at thing as getting
so general that you aren't talking about anything specific any more? I
don't have anything helpful to say here, I guess.

Best,

Bill P.

from [ Marc Abrams ( 2003.07.08.0030 ) ]

[From Bill Powers (2003.07.07.1100 MDT)]

The level of organization represented by PCT is to cell chemistry as the
operation of a radio is to the holes and electrons in silicon, the atomic
matrix in copper, the laws of electromagnetism in an inductor, At an
intermediate level of analyzing the radio, we have components like
transistors, capacitors, inductors, resistors, and insulators, which can

be

further analyzed into the detailed physics and chemistry of matter. At an
intermediate level in PCT, we have neurons, neural signals, muscle fibers,
hormones, and secretions, which can be further analyzed into underlying
biochemical and cell properties.

We have Brain/ Mind. Now revealed for the first time in this post Bill has
now introduced Mind/Brain/Radio.

This is your response to my questions and statements?

Are you going to write this up and put this in your new B:CP edition? Good
luck your going to need it.

Marc

Marc

[From Rick Marken (2003.07.08.0850)]

Bryan Thalhammer wrote:

I realize it is not the policy to ban abusive posters. I know that because
Bill himself said to us he is averse to that kind of thing...

Well, I have the technology. Any message that contains references to the
offending parties will be sent to the bin....Yet, I CANNOT ABIDE THE
UNPROFESSIONAL BEHAVIOUR OF some individuals. It hurts to
read such vile writing. What the heck?

I understand how you feel. But be careful not to throw out the baby with the bath
water.

Think about it this way. At least Marc cares enough about PCT to keep posting and
prodding us about it. This prodding can lead to gems like Bill's post on "Levels
of Analysis". I have never read a clearer explanation of the relationship between
different levels of scientific analysis. And we really have Marc to thank for it.
The radio analogy was brilliant: you can no more determine the functional
organization of human behavior by learning about the functional organization of
the neurons and calcium pumps from which that organization is built than you can
learn about the functional organization or a radio by learning about the
functional organization of the transistors and capacitors from which the radio is
built. An obvious point but one that is apparently not so obvious to some
"superreductionist" neurobiologists and biochemists.

It would be nice if more people would post informed questions, suggestions or
criticisms of PCT in a civil manner. But the fact is that people rarely post
unless they are "fired up" about something. And some of the best stuff posted on
the net has been fueled by passionate (and sometimes not very civil) debates with
these fired up individuals. ("Fired up" is a technical, PCT term for a larger than
nominal error signal in a high level control system).

Anyway, it's nice to hear from you, Bryan. I'm sorry you can't make it to the
meeting. Maybe next year we have it in your neighborhood.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

Dear Members of the CSGnet.

I realize it is not the policy to ban abusive posters. I know that because
Bill himself said to us he is averse to that kind of thing. As Fred
mentioned, that is what makes him one of the two "two most tolerant
individuals I have ever seen in action." For Rick's part, I believe that he
endures far too much abuse by certain individuals.

Well, I have the technology. Any message that contains references to the
offending parties will be sent to the bin. I can "ban" them unilaterally,
but I know how my filters are designed, not M@ar[, so I will take care of
this in my own way. My only other alternative for changing my perception of
CSGnet is to unsubscribe. But I am not desirous of doing that, because I
have invested so much in the theory, conversations, and research. Yet, I
CANNOT ABIDE THE UNPROFESSIONAL BEHAVIOUR OF some individuals. It hurts to
read such vile writing. What the heck?

Love you all,

--Bryan Thalhammer
Chicago, IL

···

-----Original Message-----
From: Control Systems Group Network (CSGnet)
[mailto:CSGNET@LISTSERV.UIUC.EDU]On Behalf Of Marc Abrams
Sent: Tuesday, July 08, 2003 12:50 AM
To: CSGNET@LISTSERV.UIUC.EDU
Subject: Re: Levels of analysis

>From [ Marc Abrams ( 2003.07.08.0030 ) ]
> > ....
We have Brain/ Mind. Now revealed for the first time in this post Bill has
now introduced Mind/Brain/Radio.

This is your response to my questions and statements?

Are you going to write this up and put this in your new B:CP edition? Good
luck your going to need it.

Marc

Marc

from [ Marc Abrams (2003.07.08.1216) ]

[From Rick Marken (2003.07.08.0850)]

I understand how you feel. But be careful not to throw out the baby with

the bath

water.

Think about it this way. At least Marc cares enough about PCT to keep

posting and

prodding us about it.

I appreciate the sentiments and glad you understand that I do care. But I'm
not trying to 'prod' or 'poke' or 'instigate' anything. If you check out
the small little list I put together on Yahoo you'll see where I'm trying to
go. Yahoo | Mail, Weather, Search, Politics, News, Finance, Sports & Videos . I have no ill will
toward anyone on this list. Not Bill, not Rick, not Bryan. I believe I have
some very legitimate questions about HPCT, none of which include
_destroying_ it.

This prodding can lead to gems like Bill's post on "Levels of Analysis". I

have never read a clearer > explanation of the relationship between
different levels of scientific analysis. And we really have > Marc to
thank for it.

A gem? A nice metaphor, but hardly a gem. Given my understanding of Bill's
position, this radio metaphor is quite cosistent with his Electric circuit/
computer/brain/mind position.

When you guys actually come up with the specific correlates to our CNS let
me know. I am _real_ interested and I am not discounting the fact that they
very well might exist. I just haven't seen them.

The radio analogy was brilliant: you can no more determine the functional
organization of human behavior by learning about the functional

organization of

the neurons and calcium pumps from which that organization is built than

you can

learn about the functional organization or a radio by learning about the
functional organization of the transistors and capacitors from which the

radio is

built. An obvious point but one that is apparently not so obvious to some
"superreductionist" neurobiologists and biochemists.

Yep, This is why it makes for only a nice metaphor and nothing else. You may
not be able to do it with electronics, but you can do it with organic
materials. Ever here of a science called chemistry? Here they take molecules
and make Airplanes out of them. Do you think they might make the mistake of
using soft plastic molecules instead of titanium molecules for some of its
parts. Please tell me molecules don't matter in chemistry? Please tell me
the laws of physics are not important to understanding space flight? Give me
a break

You seem to think, you're the only two people in the world who have a clear
picture on how things come together.

It would be nice if more people would post informed questions, suggestions

or

criticisms of PCT in a civil manner.

Where and how have I been _uncivil_? By answering your snide remarks with
ones of my own? Give me a break. and, Ah yes, Bryan has posted more in the
last 2 days then he has all year and the last time he posted, it was about
the same thing, except I was not involved. One day Bryan when you grow up
you'll be able to deal with all kinds of people and all kinds of language.
This net doesn't need a moralist and someone sitting in judgement of anyone
else. What have you contributed to the science besides Net etiquette. Grow
up and get over it. Get a life. I thought I was bad at taking to many
things personally. I don't come close to approaching you in that regard.

But the fact is that people rarely post unless they are "fired up" about

something. And some of the > best stuff posted on the net has been fueled
by passionate (and sometimes not very civil) debates > with these fired up
individuals. ("Fired up" is a technical, PCT term for a larger than

nominal error signal in a high level control system).

I think you should take a good long hard lok in the mirror on this one and
see exactly _who_ the 'fired-up' people are. You and Bill take a backseat to
no one.

Anyway, it's nice to hear from you, Bryan. I'm sorry you can't make it to

the

meeting. Maybe next year we have it in your neighborhood.

Yes, the next time you hear from Bryan it will probably be to scold someone
else for a perceived mis-use of the language and etiquette, that's what he
seems to respond to on this net.

Marc

[From Bill Powers (2003.07.08.1119 MDT)]

Marc Abrams (2003.07.08.1216)--

Yes, the next time you hear from Bryan it will probably be to scold someone
else for a perceived mis-use of the language and etiquette, that's what he
seems to respond to on this net.

You can truly be a boor, Marc. Do you ever read what you post before
sending it? There have been squabbles on CSGnet, but nobody has come even
close to the ugliness of your posts. What title are you trying for, "The
Guy Nobody Wants To Talk To"?

Bill P.

[From Rick Marken (2003.07.08.1430)]

Marc Abrams (2003.07.08.1216) --

> Rick Marken (2003.07.08.0850)]

> This prodding can lead to gems like Bill's post on "Levels of Analysis".

A gem? A nice metaphor, but hardly a gem. Given my understanding of Bill's
position, this radio metaphor is quite cosistent with his Electric circuit/
computer/brain/mind position.

It's neither a metaphor nor a position. It is a description of the _fact_ that you
can't tell how a system (such as a radio or human being) functions learning only
how the components of the system function.

When you guys actually come up with the specific correlates to our CNS let
me know.

  See Bill Powers (2003.07.08.1024 MDT). Some of the correlates are in the CNS,
some are in the PNS. I would particularly like to hear your answer to the
following:

Many feedback loops similar to the one in Fig 7.3 have been
found and physically traced in human and animal brains. So
what's your problem?

What _is_ your problem?

> The radio analogy was brilliant: you can no more determine the functional
> organization of human behavior by learning about the functional organization
> of the neurons and calcium pumps from which that organization is built than
> you can learn about the functional organization or a radio by learning about
> the functional organization of the transistors and capacitors from which the
> radio is built. An obvious point but one that is apparently not so obvious to
> some"superreductionist" neurobiologists and biochemists.

Yep, This is why it makes for only a nice metaphor and nothing else. You may
not be able to do it with electronics, but you can do it with organic materials.

How? How can you tell from the biochemical properties of a neuron that the neuron
is part of a control system organization rather than an S-R organization, for
example? I think you can no more tell from the biochemistry of a neuron whether
it is part of an open or closed loop system than you can tell from the electrical
properties of a transistor whether it is part of a TV or radio.

Ever here of a science called chemistry? Here they take molecules
and make Airplanes out of them. Do you think they might make the mistake of
using soft plastic molecules instead of titanium molecules for some of its
parts.

A plane is a good example of a functional organization that can be implemented
using many different chemical substances for its components. Wings, for example,
have been built of paper cloth, wood and metal. You could nit tell, by looking at
the substance, whether it was part of a plane or a house. It's how the components
are organized that matters. You do need to know chemistry in order to be able to
intelligently select the materials out of which to build the plane. But you have
to know aeronautical engineering in order to be able to put those materials
together in a way that results in a plane rather than a brick.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

from [ Marc Abrams (2003.07.08.2239) ]

[From Rick Marken (2003.07.08.1430)]

It's neither a metaphor nor a position. It is a description of the _fact_

that you

can't tell how a system (such as a radio or human being) functions

learning only

how the components of the system function.

Rick, Rick, Rick. No, you cannot know how a sytem functions by knowing how a
component of that system functions. You know by knowing how the components
_ACTUALLY_ _interact_ with one another and how those processes interact with
higher level processes, etc. etc. _NOTHING_ happens unless a cell
functions., NOTHING. I have read and heard that there are anywhere from 6 to
10 'levels' of organization in biological systems. Let me present a
different one that makes just as much sense as Bill's. I would even venture
to say that maybe they are the same.

"From a single-cell physiologist interested in neuronal integration and
synaptic transmission. Such a position is privileged because it lies between
the realms of the molecular and the systemic, as they relate to brain
function.

Single large neurons have physical demensions observable at low optimal
magnification, that of a tenth of a millimeter. That is big enough to be
dissected by hand with pins, using a good magnifying glass (Deiters 1856).
Moving just two orders of magnitude further down to the micrometer level,
which requires a good microscope, one is at the scale of synaptic
transmission. One may observe synapses at the union between nerve and
muscle, for example. Two orders of magnitude further down, at tens of
nanometers, with the aid of electron microscopy, we find the realm of of
single ion channels and of signal transduction and molecular biology,

If on the other hand we wish to roam orders of magnitude above the
physiology of single cells, we find at two orders of magnitude above, and in
the centimeter realm, the world of systems that is the scale of pennies,
buttons, and fingernails. At a further two orders of magnitude up, we come
to meters and to the world of motricity and cognition that characterizes
human beings. That is, we arrive at the realm of chairs and telephones and
other objects that one can hold in one's hand or under one's arm.
Most neuroscientists feel that two orders of magnitude above and below one's
central focus is "horizon enough", and that anyone attempting four orders
above and below reckless. However, there are some who attempt such a
dangerous dynamic range. They probably know that the risk of failure is the
price of synthesis, without which there are only fields of dismembered
parts."

Dr. Rodolfo Llinas 2000

Where do you think PCT lies? How about HPCT? using this scale.

  See Bill Powers (2003.07.08.1024 MDT). Some of the correlates are in the

CNS,

some are in the PNS. I would particularly like to hear your answer to the
following:

Boy, you really got me now. How about an answer to the sensory-binding
issue?

> Many feedback loops similar to the one in Fig 7.3 have been
> found and physically traced in human and animal brains. So
> what's your problem?

What _is_ your problem?

What 'causes' the feedback and how it's 'implemented'.

How? How can you tell from the biochemical properties of a neuron that the

neuron

is part of a control system organization rather than an S-R organization,

for

example? I think you can no more tell from the biochemistry of a neuron

whether

it is part of an open or closed loop system than you can tell from the

electrical

properties of a transistor whether it is part of a TV or radio.

See my statement above. You can only tell by the _interactions_ of those
parts and how they intern form bigger processes.

A plane is a good example of a functional organization that can be

implemented

using many different chemical substances for its components.

And what are those 'components' made of?

Wings, for example, have been built of paper cloth, wood and metal. You

could nit tell, by looking at

the substance, whether it was part of a plane or a house. It's how the

components

are organized that matters.

Not entirely true but I agree with the notion of utility you present.

You do need to know chemistry in order to be able to
intelligently select the materials out of which to build the plane.

No, not to _build_ the plane, just to select the material that you will
_use_ to make the parts. I believe it's called 'materials science'

But you have
to know aeronautical engineering in order to be able to put those

materials

together in a way that results in a plane rather than a brick.

Absolutely.

Unfortunately you seem to think you can reverse-engineer the CNS. I don't
think you can

Marc

[From Bill Powers (2003.07.09.0616 MDT)]

Marc Abrams (2003.07.08.2239)--

> [From Rick Marken (2003.07.08.1430)]

> It's neither a metaphor nor a position. It is a description of the
_fact_ that you can't tell how a system (such as a radio or human being)
functions learning only how the components of the system function.

Rick, Rick, Rick. No, you cannot know how a sytem functions by knowing how
a component of that system functions. You know by knowing how the
components _ACTUALLY_ _interact_ with one another and how those processes
interact with higher level processes, etc. etc.

Marc, Marc, Marc. Since the components can be assembled in many different
ways, the manner of assembly makes just as much difference as the nature of
the components that are assembled. Every cell in the body has the same DNA;
what makes one cell into a liver cell and another into a neuron is which
genes become activated and in which combinations as the organism grows in
the placenta.

_NOTHING_ happens unless a cell functions., NOTHING.

This is true, too -- just as a radio will do nothing until its transistors
function. But you still can't discover what the whole system does by
studying nothing but its parts. There are levels of organization. not just
size, that introduce laws which simply don't exist at the component level.
Moreover, those laws can be exemplified in more than one kind of physical
system. This is the reason we can build computers to carry out operations
formerly done only by neurons in a human brain -- not all of them, but
enough to show that the principle holds true.

I have read and heard that there are anywhere from 6 to 10 'levels' of
organization in biological systems. Let me present a different one that
makes just as much sense as Bill's. I would even venture to say that maybe
they are the same.

No, they are totally different. The one you cite is based on size; mine is
based on function and on the relationships involved in hierarchical
control, where higher systems are anatomically separate from lower ones and
operate by using the lower ones.

Where do you think PCT lies? How about HPCT? using this scale.

PCT is organized along a different dimension, not according to the
magnification with which you examine the organism. Brain cells that deal in
system concepts, at the top of the heap, are no bigger than the spinal
motor cells that control intensity of effort. They are also different
cells; we are not just drawing bigger boundaries around the same cells.
Size is irrelevant.

How about an answer to the sensory-binding issue?

That issue is handled in PCT by saying that the kinds of perceptual
computations that exist at a given level apply to all modalities of
perception (they are all carried as perceptual signals). Thus at the third
level, configurations, we have configurations of taste, smell, sound,
vision, kinesthesia, pressure, and so forth, depending on the source of the
perceptual signals. The same is true at every level, through as we get
farther from the intensity level, the different modalities merge into a
common world: logic, for example, applies to all experiences, but doesn't
change with modality.

What 'causes' the feedback and how it's 'implemented'.

Feedback exists any time there is a closed loop of causation in operation
-- that is, when the right causal connections exist. Your question makes no
sense: feedback isn't an event that happens now and then or that requires
special "implementation" to happen. The perceptual world is always there;
it's simply the world you experience, and is present during every waking
moment. There are no times "between perceptions." Your actions are
continually affecting the world and your relationship to it, which means
they are continually affecting your perceptions without any pauses or
time-outs. So feedback is continuously in operation and your actions are
continuously influencing your perceptions -- those subject to control, that is.

See my statement above. You can only tell by the _interactions_ of those
parts and how they intern [in turn? wtp] form bigger processes.

That is the conventional view of hierarchical levels of organization --
that we're just drawing larger and larger boundaries, creating larger and
larger groups of the same components. But this is not how the analysis in
PCT is organized. And it is not how the body is organized, either.

If I argued the way you do, I would just make that flat statement and leave
it there, saying it's up to you to understand it. But I will explain.

The body's cells are specialized. They are grouped not by size, but by
carrying out common functions. This is what creates organs; we are not
solid inside like a potato, but differentiated into modules made of cells
with similarly expressed genes, modules that do different things and
interact with each other as units. Inside the modules there are submodules
that create the inner architecture of the organ.

Consider the thyroxine control loop. The pituitary generates a hormone
called TSH, thyroid-stimulating hormone, which goes into the bloodstream
and is taken up by the thyroid. TSH causes the thyroid to secrete
thyroxine, at a rate that depends on the concentration of TSH. The
thyroxine circulates to all the organs, including the pituitary itself. In
the pituitary, the thyroxine suppresses the manufacture of TSH in
proportion to the concentration of thyroxine. There is also a chemical
substance (that I can't bring to mind right now) that enters the pituitary
from the hypothalamus in the brain, which, in proportion to its
concentration, raises the production of TSH.

We thus have all the relationships required for a control system, and a
control system it is. If anything disturbs the thyroxine concentration in
the bloodstream, such as being used metabolically or having extra thyroxine
injected, the control loop raises or lowers the production of thyroxine by
the thyroid gland to keep the concentration nearly constant. The signal
from the brain, of course, sets the reference level that determines what
level of thyroxine in the bloodstream will be maintained, but the control
relative to that level happens autonomously, in the biochemical systems.

These are relationships among different modules which are distinguished by
function, not size.

Unfortunately you seem to think you can reverse-engineer the CNS. I don't
think you can.

I don't think YOU can. I can.

Best,

Bill P.

from [ Marc Abrams (2003.07.09.1032) ]

···

-

[From Bill Powers (2003.07.09.0616 MDT)]

Marc, Marc, Marc. Since the components can be assembled in many different
ways, the manner of assembly makes just as much difference as the nature

of

the components that are assembled.

Who said anything different? That was my point. It has been found that
neurons, in groups, oscillate at various frequencies, as a group, and that
being only one dimension of neuronal activity, Axon lenghth, coherence,
chemical make-up and the fact that we have over 50 different kinds of
neurons make for a very elaborate and at this point in time largely unknown
mixture. Some things being better understood then others at this point.

Every cell in the body has the same DNA;
what makes one cell into a liver cell and another into a neuron is which
genes become activated and in which combinations as the organism grows in
the placenta.

Never said anything different here.

> _NOTHING_ happens unless a cell functions., NOTHING.

This is true, too -- just as a radio will do nothing until its transistors
function. But you still can't discover what the whole system does by
studying nothing but its parts.

You keep on harping on this and I keep on telling you it's not the parts but
the interactions of the parts
that count and the interstions of those parts. Each level has different
emergent properties.

There are levels of organization. not just size, that introduce laws which

simply don't exist at the >component level.

That might be true, but laws at the component level apply to _all_
components at any higher level. Molecules that are dissolvable in water will
always be dissolvable in water not matter what they ultimately become. here
is the wonder of chemistry. Mix the right molecules and you have certain
properties. Hopefully useful in some way.

Moreover, those laws can be exemplified in more than one kind of physical
system. This is the reason we can build computers to carry out operations
formerly done only by neurons in a human brain -- not all of them, but
enough to show that the principle holds true.

Nonsense. A computer _cannot_ carry out all neuronal functions. A computer
is like an idiot savant. It does certain things exceptionally well and is
absolutely useless for others.. Neurons _do not_ compute like a computer,
neither analog or digital.

No, they are totally different. The one you cite is based on size; mine is
based on function and on the relationships involved in hierarchical
control, where higher systems are anatomically separate from lower ones

and

operate by using the lower ones.

No, If you would have read the final few sentences carefully you would have
seen this:

"Most neuroscientists feel that two orders of magnitude above and below
one's
central focus is "horizon enough", and that anyone attempting four orders
above and below reckless. However, there are some who attempt such a
dangerous dynamic range. They probably know that the risk of failure is the
price of synthesis, without which there are only fields of dismembered
parts."

Llinas _was_ talking about _functionality_. From nuclear to cognitive. What
does 'cognitive' have to do with 'size'?

PCT is organized along a different dimension, not according to the
magnification with which you examine the organism. Brain cells that deal

in

system concepts, at the top of the heap, are no bigger than the spinal
motor cells that control intensity of effort. They are also different
cells; we are not just drawing bigger boundaries around the same cells.
Size is irrelevant.

Wow, you are just so insiteful. That statement is not true of course.
Neurons come in all sizes, but size does not determine function and you're
the only one making that strawman claim about someone else.

>How about an answer to the sensory-binding issue?

That issue is handled in PCT by saying that the kinds of perceptual
computations that exist at a given level apply to all modalities of
perception (they are all carried as perceptual signals). Thus at the third
level, configurations, we have configurations of taste, smell, sound,
vision, kinesthesia, pressure, and so forth, depending on the source of

the

perceptual signals. The same is true at every level, through as we get
farther from the intensity level, the different modalities merge into a
common world: logic, for example, applies to all experiences, but doesn't
change with modality.

Congradulations on solving that problem. You should win the Nobel Prize for
that if your right. No one else has it licked. Of course I think your wrong,
but the Nobel committee has not asked me.

Feedback exists any time there is a closed loop of causation in operation
-- that is, when the right causal connections exist.

Another great insite. So what else is new. I was asking for _SPECIFICS_,
What entites are involved and what's the process?

Your question makes no sense:

Sure it does.

feedback isn't an event that happens now and then or that requires
special "implementation" to happen.

Yes, I know

The perceptual world is always there;

_Discontinuously_, and much differently then the way your hierarchy explains
it. _Much_ differently.

it's simply the world you experience, and is present during every waking
moment. There are no times "between perceptions."

Yes there are.

Your actions are continually affecting the world and your relationship to

it,

Nope. It's _discontinous_. This of course is just one mans opinion, there
are others, would you like me to give you all the cites? Think Llinas does
not "know" about feedback?

Some organizing principles for the control of movement based on
olivocerebellar physiology.

Welsh JP, Llinas R.

Department of Physiology and Neuroscience, New York University Medical
Center, NY 10016, USA.

Motor control is defined as the process of restricting the output of the
motor nervous system so that meaningful and coordinated behavior ensues. The
high dimensionality of the computation underlying motor control is presented
and a simplifying framework is outlined. Evidence that movements are
performed non-continuously is reviewed as is the construct of the 'motor
synergy' as a fundamental unit of control. It is proposed that the pulsatile
nature of movement and the tendency of muscle collectives to be activated as
synergies reflect processes that the nervous system has evolved to reduce
the dimensionality of motor control. We propose that the inferior olive
simplifies the computation underlying motor control by biasing the
activities of spinal and cranial motor systems so that discrete collectives
of muscles are predisposed to contract at specific times during movement.
The well-characterized oscillatory activity of olivary neurons is postulated
to provide a pacemaking signal and to restrict the control process to
particular moments in time while the process of electrotonic coupling and
uncoupling of assemblies of olivary neurons is proposed to underlie the
spatial distribution of synergic muscle activations. It is proposed that the
olivocerebellar contribution to the control process is to allow movements to
be executed rapidly in a feedforward manner, so that the need for sensory
guidance and feedback is minimized.

which means
they are continually affecting your perceptions without any pauses or

time-outs. So feedback is > continuously in operation and your actions
are

continuously influencing your perceptions -- those subject to control,

that is.

Believe what you will. It only _appears_ that way. Just like temperture
'feels' warm and cold and not the _actual_ molecular movement. But molecular
movement _IS_ what you feel

That is the conventional view of hierarchical levels of organization --
that we're just drawing larger and larger boundaries,

Wrong again. It's not simply larger and larger boundries. Each level has new
and emergent properties based on the interactions. They are called
transformations.

creating larger and larger groups of the same components.

Nope.

But this is not how the analysis in PCT is organized. And it is not how

the body is organized, either.

Who said it was?

The body's cells are specialized. They are grouped not by size, ...

These are relationships among different modules which are distinguished by
function, not size...

When you get on a roll you stick with it. Strawmen every where.

>Unfortunately you seem to think you can reverse-engineer the CNS. I don't
>think you can.

I don't think YOU can. I can.

Right.

Marc

[From Bill Powers (2003.07.09.1119 MDT)]

Marc Abrams (2003.07.09.1032)--

>The perceptual world is always there;

_Discontinuously_, and much differently then the way your hierarchy explains
it. _Much_ differently.

> it's simply the world you experience, and is present during every waking
> moment. There are no times "between perceptions."

Yes there are.

That's extremely interesting. Do you mean that your perceptual world just
goes black in between perceptions, with no perceptions at all happening?
For example, as I sit here, I am experiencing a room that's lit by light
from the windows. In fact, I've been perceiving light for at least six
hours (since I woke up), and my vision never blacked out for an instant.
Also, I've known where I am in my environment continuously since then, and
have always been in the place I intend to be right up to now, so that
control process has been continuous for six hours.

I suppose you're referring to the fact that sometimes we are doing one
thing and sometimes another. But each of these processes is also
continuous, going smoothly from a beginning through a middle to an end, not
in a series of instantaneous jumps and jerks.

Perhaps the problem is the way you understand the term "continuous." Its
basic meaning is that the variables involved go from one state to another
state by going through all the intervening states. The opposite of
"continuous" is "discrete," in this mathematical usage. A discrete variable
can jump from one value to another value without ever having any of the
intermediate values. An example of a continuous variable is the angle at
your elbow joint. You can move your arm from one angle to another, slowly
or quickly, but during the change (which takes a finite amount of time) the
angle passes through every possible angle between the limits. An example of
a discrete variable would be the truth value of a logical proposition like
"Today is Wednesday." At midnight tonight, the truth value of that
statement will change from "true" to "false," with no states in between and
no time required for the change.

Or are you really talking about stimulus and response, saying that there is
always an external stimulus to start any action?

>Your actions are continually affecting the world and your relationship to
it,

Nope. It's _discontinous_. This of course is just one mans opinion, there
are others, would you like me to give you all the cites? Think Llinas does
not "know" about feedback?

Who cares about the citations? Just give me the reasons, if you understand
them. Ten guys can stand up and shout "it's discontinuous" without ever
saying anything worth listening to. I want to know why _you_ say that. If
you're saying it only because one or two or twenty famous people said it,
but don't understand why they said it, then of course all you can do is
refer me to them. But if you understand _why_ it was said, it no longer
matters who said it, and you can support your case all by yourself. I don't
think you can make a valid case for behavior being discontinuous. Put up or
shut up.

Some organizing principles for the control of movement based on
olivocerebellar physiology.

Welsh JP, Llinas R.

Department of Physiology and Neuroscience, New York University Medical
Center, NY 10016, USA.

Motor control is defined as the process of restricting the output of the
motor nervous system so that meaningful and coordinated behavior ensues. The
high dimensionality of the computation underlying motor control is presented
and a simplifying framework is outlined. Evidence that movements are
performed non-continuously is reviewed as is the construct of the 'motor
synergy' as a fundamental unit of control. It is proposed that the pulsatile
nature of movement and the tendency of muscle collectives to be activated as
synergies reflect processes that the nervous system has evolved to reduce
the dimensionality of motor control. We propose that the inferior olive
simplifies the computation underlying motor control by biasing the
activities of spinal and cranial motor systems so that discrete collectives
of muscles are predisposed to contract at specific times during movement.
The well-characterized oscillatory activity of olivary neurons is postulated
to provide a pacemaking signal and to restrict the control process to
particular moments in time while the process of electrotonic coupling and
uncoupling of assemblies of olivary neurons is proposed to underlie the
spatial distribution of synergic muscle activations. It is proposed that the
olivocerebellar contribution to the control process is to allow movements to
be executed rapidly in a feedforward manner, so that the need for sensory
guidance and feedback is minimized.

This is sheer gobbldygook. In the first place he's assuming that control is
carried out by computing the neural signals required to produce a desired
end-result, and then issuing the commands. All the complications he talks
about exist ONLY if this is the right model, which it is not. As soon as
you switch to the correct model there is no need for the "simplifying
framework." These guys are throwing big words around as if they know what
they are talking about, which they absolutely do not. It's a huge bluff.

Well, sorry for that outburst. But not very sorry.

Bill P.

[From Rick Marken (2003.07.11.0830)]

I would like to make one more comment regarding this "Levels of Analysis" thread.
If analysis at the level of neurophysiology really did give us some privileged
understanding of phenomena at the level of behavior then neurophysiologists would
surely have discovered by now that behavior is organized around the control of
perceptions (however those perceptions are implemented neurophysiologically). But,
to my knowledge, the fact that behavior is the control of perception is as unknown
in neurophysiology as it is in psychology.

I do think control theorists should have some basic understanding of
neurophysiology. I would list neurophysiology (and, for that matter, physiology in
general) as one of the basic prerequisites to the study of PCT, along with math,
at least to the level calculus 1, computer programming, basic physics and
chemistry and English composition. I think it's important to develop models at the
behavioral level, to the extent possible, within the knows constraints imposed by
the next lower level, the level at which this these behavioral functions are
implemented; the level of neurophysiology.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

Richard,

I am curious about the type of work that you do for the Rand
Corporation?

Thanks

···

On Friday, Jul 11, 2003, at 05:35 US/Pacific, Richard Marken wrote:

[From Rick Marken (2003.07.11.0830)]

I would like to make one more comment regarding this "Levels of
Analysis" thread.
If analysis at the level of neurophysiology really did give us some
privileged
understanding of phenomena at the level of behavior then
neurophysiologists would
surely have discovered by now that behavior is organized around the
control of
perceptions (however those perceptions are implemented
neurophysiologically). But,
to my knowledge, the fact that behavior is the control of perception
is as unknown
in neurophysiology as it is in psychology.

I do think control theorists should have some basic understanding of
neurophysiology. I would list neurophysiology (and, for that matter,
physiology in
general) as one of the basic prerequisites to the study of PCT, along
with math,
at least to the level calculus 1, computer programming, basic physics
and
chemistry and English composition. I think it's important to develop
models at the
behavioral level, to the extent possible, within the knows constraints
imposed by
the next lower level, the level at which this these behavioral
functions are
implemented; the level of neurophysiology.

Best regards

Rick
--
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

regards,

christopher

[From Rick Marken (2003.07.11.2330)]

christopher geddes wrote:

I am curious about the type of work that you do for the Rand Corporation?

Hi Christopher

You're new to CSGNet, right? Or am I having a senior moment?

Anyway, I'm happy to tell you what I do at Rand but I'd appreciate it if
you'd quid pro quo by introducing yourself (at least to me) and tell us
what you do, and what your interest is in PCT.

Regarding my work, I am a behavioral scientist at Rand which means that I
do behavioral research on topics related to health care, education and
other policy issues. My main research interest at the moment is related to
prioritizing health care safety practices in terms of their likely
effectiveness. As part of this research program I have developed a control
theory-based model of prescribing that provides a framework for studying
factors that affect the rate at which medication order errors occur. The
model (to the extent that it is correct, and it still needs much more
testing against medication error data) should make it possible to predict
which safety interventions are most likely to produce the greatest
reductions in medical error rates (and, thus, the greatest improvements in
medical safety). The model is described in a paper that will be appearing
soon in _Ergonomics_ ( see the reference at the CSG main page,
http://www.ed.uiuc.edu/csg/). I plan to discuss this work at the CSG
meeting later this month.

Best regards

Rick

···

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

Hello Rick,

I have actually been on the list for over a year, essentially
participating as an observer.

I am interested in finding or developing a model which enables me to
more accurately produced certain results in my environment, hence my
interest in PCT.

Thanks

···

On Friday, Jul 11, 2003, at 23:28 US/Pacific, Richard Marken wrote:

[From Rick Marken (2003.07.11.2330)]

christopher geddes wrote:

I am curious about the type of work that you do for the Rand
Corporation?

Hi Christopher

You're new to CSGNet, right? Or am I having a senior moment?

Anyway, I'm happy to tell you what I do at Rand but I'd appreciate it
if
you'd quid pro quo by introducing yourself (at least to me) and tell us
what you do, and what your interest is in PCT.

Regarding my work, I am a behavioral scientist at Rand which means
that I
do behavioral research on topics related to health care, education and
other policy issues. My main research interest at the moment is
related to
prioritizing health care safety practices in terms of their likely
effectiveness. As part of this research program I have developed a
control
theory-based model of prescribing that provides a framework for
studying
factors that affect the rate at which medication order errors occur.
The
model (to the extent that it is correct, and it still needs much more
testing against medication error data) should make it possible to
predict
which safety interventions are most likely to produce the greatest
reductions in medical error rates (and, thus, the greatest
improvements in
medical safety). The model is described in a paper that will be
appearing
soon in _Ergonomics_ ( see the reference at the CSG main page,
http://www.ed.uiuc.edu/csg/). I plan to discuss this work at the CSG
meeting later this month.

Best regards

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

regards,

christopher

[From Rick Marken (2003.07.15.0830)]

christopher geddes wrote:

I have actually been on the list for over a year, essentially
participating as an observer.

Thanks, then, for participating as a questioner.

What, by the way, is your field of interest? Psychology? Engineering? Something
else?

I am interested in finding or developing a model which enables me to
more accurately produced certain results in my environment, hence my
interest in PCT.

I think the model you want, then, would be one that pertains to the results you
want to produce more accurately. For example, if you want to produce more accurate
spaces between lines in optical gratings then optics is the model for you. If you
want to produce more accurate time stamps then perhaps nuclear physics is the
model you want.

PCT can't really help you produce intended results more accurately. PCT is a model
that says _that_ living systems are organized the production of intended results
and that explains _how_ systems produce these results (by acting to controlling
perceptual representations of these results). As my model of error shows, even
very highly skilled living control systems do not always produce what observers
would agree is an accurate result.

Best regards

Rick

···

---
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org