[From Bill Powers (2003.03.04.0332 MST)]
Marc Abrams (2003.03.02.1852)--
I have found the source code (in Delphi) for the program I wrote for David
Goldstein, and will work up some instructions for how to use it with Vensim
to do an analysis of control parameters. If I could figure out how to do
the "Powell Optimization" algorithm, Vensim wouldn't be needed, but it will
be useful to show modeling methods as well as to perform the analysis.
The hypothesis we will be testing immediately is that stress leads to
reorganization which is detectable as _dis_organization of a previously
skilled performance. Tracking experiments, or any other experiment
involving control of a visual variable using a mouse, are handy because we
can use a computer to measure the performance accurately and rapidly. All
we really need is a running measure of root-mean-square tracking error,
isn't it?
I had originally thought of using level of difficulty in the tracking task
to govern the amount of error, but that would leave no way of measuring
skill independently. I think now that the tracking task should be used
simply as a measure of skilled performance, at a constant (fairly high)
level of difficulty set by the rapidity with which disturbances vary. This
implies finding some _other_ way of creating errors in control systems
other than those used in the tracking task. The article that was discussed
a few days ago used five methods of inducing stress, including pain -- you
might not like using that one on yourself! Sitting with one foot in cold
but not freezing water might serve instead. Check with your MD, of course,
and your heart specialist before you start stressing yourself.
These are all just preliminary thoughts.
The mention of neurotransmitters a few days ago brought to mind some
half-baked thoughts that I think we need to consider now and then. I have
no conclusions to offer yet, just some things to consider.
The basic question is, what is a chemical signal and what is a structurally
important building block of the body? An example of a structurally
important building block is calcium, a constituent of bone for which there
is no substitute. It is the physical and chemical nature of calcium which
makes it important as a component of load-bearing bones. In contrast,
consider the role of calcium in the functioning of neurons. Calcium
channels open and close, and calcium ions move into and out of the
nerve-cell as it fires and recovers. But the fact that calcium is involved
is unimportant, because what matters about nerve cells are the signals they
send from one place to another, and the way the signals interact in the
cell body. If a neuron were replaced by a suitable transistor circuit, so
that incoming neural signals were transformed into outgoing signals in
exactly the same way as before, there would be no difference in functioning
of the nervous system. Would there?
I'm not sure if the "substitution" idea is adequate for this argument --
bones could be replaced by steel rods, I suppose (of course steel rods
could not recover from a break). The idea I'm trying to get at is the
difference between a component of the body that plays a role in the gross
physical structure of the body, and a component that serves only to carry
information about a physical variable from one place to another. In the
latter case, the properties of the signal itself have no bearing on the
information that is carried, just as transistors work just as well whether
their charge carriers are negative electrons or positive holes.
In the nervous system, signals pass from sensors or computing networks to
other locations, where they present the information that such-and-such a
distant variable has such-and-such a value. The signals are prevented from
reaching the wrong destinations by the fact that they are carried in
discrete, and usually insulated, fibers. They can't go anywhere else.
At the destinations, however, the signals have to jump a gap, a synapse,
and the information that the signal (and the distant variable) has a
certain value is carried by neurotransmitters. While the transmitters are
in the gap they can leak out, even though the gap is extremely small, and
this can constitute a problem in the brain where neurons are densely
packed. If all synapses used the same neurotransmitter to carry the signal
from the incoming fiber into the cell body, the leakage could very well
have the effect of sending a copy of the signal to the wrong cell bodies as
well as the right one. The evolutionary solution to this problem is to use
different neurotransmitters which are recognized only by the correct cell
bodies.
What I'm working up to here is the proposition that the _chemical type_ of
neurotransmitter employed in different volumes of the brain has no
functional significance other than the fact that receptors exist which
respond exclusively to just one type, so that signals in adjacent volumes
do not interfere with each other. The system would work just the same if
the types of neurotransmitters were somehow interchanged.
If this proposition is true, then it is futile to ask what there is about
different neurotransmitters _as chemicals_ that produces various conditions
such as schizophrenia. It is not the chemistry of the neurotransmitter that
is the problem; the problem likes in the reasons for the lack or excess of
that transmitter, which trace back to a lack or excess of activity in the
neurons that produce that type of transmitter. In other words, the parts of
the brain where the signals originate are creating the problem, or at least
passing along the effects of the real cause. That would be what I call a
functional problem, a problem in how the brain thinks, rather than a
structural problem.
It's possible, of course, that there can be genetic defects that alter
brain chemistry in a global way that can't be corrected, as it were, by
right thinking. But from what reading I do in this field (Nature and
Science, primarily), the evidence is not yet nearly good enough to
distinguish most structural defects from functional ones. The tendency in
medicine is to assume that all brain problems are structural and therefore
should be treated by manipulating chemistry, but that conclusion is
unjustifiable; it is simply a predudice. It leads to looking at local
chemistry changes without asking whether the incoming signals have changed
because of some problem in a totally different region of the brain -- or
even in the external world. Depression can certainly be linked to changes
in neurotransmitter concentrations, but I'm sure that in come cases such
changes are easily be traced to changes in external circumstances, such as
the death of a loved one and a subsequent need for some drastic functional
reorganizations.
It would be astonishing to me to find that any change in mental
organization was not accomnpanied by significant chemical changes in the
brain. But we must be wary of about jumping to conclusions about cause and
effect. At one time in my life I could be plunged into gloom by the sight
of a few numbers in my newspaper's listings of stock prices. I'm quite
prepared to believe that a lot of neurotransmitters changed their
concentrations as a result. But I'm equally sure that taking Prozac to make
me feel better would not have done anything about the basic problem, which
would have been just as needful of a solution when the Prozac was
withdrawn. The only possible excuse for taking Prozac in that situation
would be that I was incapacitated by the overwhelming size of the error
signals and needed a temporary crutch even to start thinking about a
solution. I will admit that much, grudgingly.
So, anyway, that is the status of my thinking on that subject -- still
inconclusive, but the basic problems seem real.
Best,
Bill P.
Measure the “anxiety” you can measure the error and stress. ( Not really, but you can measure some kind of response to error ). So I have basically come up with the hypothesis that CV’s are prioritized by the level of anxiety or error is present in the control loop. Pass a certain threshold and Reorganization kicks in. There is never any reason for positive feedback to occur when the goal of a loop is to either to stop (die) or slow down.