[From Bill Powers (920819.0900)]
Penni Sibun (920818, 920819) --
Here is one of our problems:
when i am driving, snow on the road or rain on the windscreen is
extremely immediate and concrete.
Here is another:
i don't have any particular emotional investment in what people
are made of. i was the one who suggested it's all physics at the
bottom.
A third:
you really think that how much work a process takes is determinable by
introspection? that doesn't sound very scientific to me.
A fourth:
i was using very gross examples, where i thought the work involved
would be obvious, that is, it involves muscles, rather than primarily
neurons.
From my standpoint, these are the same views of behavior that led to
behaviorism. It's assumed, I believe, that the Observer can see real
"immediate and concrete" reality itself, without interpretation -- just the
facts. It's assumed that the world is as it is, and that all you need to
know about people is what happens to them and what they do. How they are
constructed inside in order that they can behave in that environment as
they do is of no interest (and makes no difference). The fact that all
motor behavior and all sensory experience is created by neurons is of no
importance. In principle, we can understand everything about behavior by
watching the interaction of environmental things, events, processes, and
situations with organismic activities -- behaviors. Behavior -- what
organisms DO -- is to be explained in terms of observable interactions
only. It's assumed that the mechanics of behavior will be explained, in the
end, by physics and chemistry; there's no need to ask WHAT physics and WHAT
chemistry. That is the scientific way of dealing with behavior. If these
are really the tenets of interactionism, then interactionism is little
different from behaviorism.
Control theory is based on an approach that is basically different from the
ground up. It's assumed that perception results from neural activity based
on sensory inputs -- that there is no other way to know what is going on
outside the organism (and that applies to the scientist as well as to the
subjects under study). It's assumed that all Observers must see the world
this way, as neural signals standing for a world of which they know nothing
directly -- but experienced, of course, as a real concrete external world
and a body living in it. It's assumed that all observers act by producing
neural signals that activate muscles, and know of their own actions only
through sensing of muscle efforts and sensing of the effects on other
perceptions. If Observers aren't brains, then at least they get all their
experiences and produce all their actions via brains; there's no channel
linking awareness to the outside world that bypasses the neural processes
of perception. There's no way for them to act other than by sending neural
signals to muscles and glands. The entire experienced world, from the most
concrete and simple aspects to the most abstract thoughts about them,
exists as patterns of neural firing in the brain.
That, of course, is a model. It's a model featuring a device called "the
brain," whose internal activities are experienced by a propertyless
Observer. It's consistent with our models of physics and chemistry, applied
either to the internal parts of the model or to the hypothetical reality
outside it. It's consistent with what is known about the physical structure
of the body -- biochemistry and neurology. Control theory brings all these
models together into a single consistent framework, without claiming any
property for a scientist that the subjects don't also have, without
claiming that the scientist has any way of acting or knowing the truth that
others don't also have. And control theory goes further -- it proposes an
internal organization that can account for the way we really observe
behavior to work instead of just how it has been imagined to work.
There are, of course, more questions unanswered than answered by HPCT. Many
people on this net are trying to answer them, trying out various
possibilities, rejecting some and carrying others forward. But behind all
these conjectures is a common understanding of the nature of the problem,
which is very different from yours. Most of the people who are looking for
answers are convinced that just taking appearances for granted and trying
to find the rules is futile. They are trying to find a model for the
organization of the system that is responsible for both experience and
behavior, so that when it is placed in any environment it will behave as
real organisms do -- and experience it as at least human organisms do.
There are a few phenomena that PCT has uncovered which are easy to see and
which no other theory can yet explain. The main one is that what people do
with their muscles is variable, yet the outcomes of the muscle activity are
repeatable and resistant to disturbance. From any existing scientific point
of view, this phenomenon is counterintuitive and inexplicable.
In the kind of explanatory system you're presenting, this fundamental
phenomenon doesn't even appear, because all of your descriptions are cast
in terms of the outcomes produced by motor activities -- moving the car
here and there on the road, for example. In looking for an explanation of
such outcomes, this explanatory system doesn't ask how the behaving system
can produce such repeatable outcomes by such variable means. It looks to
other aspects of the apparent world for explanations -- those outcomes are
just the "easiest" ones that the environment makes possible, elicits,
encourages, or whatever. The fact that those outcomes are continuously
being disturbed, not aided, by the environment is overlooked: if the
outcome is stable, the environment must have made that stability possible.
The fact is that the environment is always working to disrupt that
stability. What you describe as an explanation of behavior seems to me more
like presenting a series of problems calling for an explanation. The
statements you offer as explanations seem to involve more than a bit of
magic, and more than a modicum of arm-waving. The arm-waving isn't evident
from within the framework you're describing, because there is one question
that simply doesn't arise: HOW CAN THAT POSSIBLY WORK? But to anyone who
actually tries to make working models of any proposed explanation, the
question of HOW is the crux of the matter. If the explanation entails a HOW
that is impossible, or that flatly contradicts our other models of reality
such as physics and chemistry, that alone is enough grounds to reject the
explanation. If the explanation doesn't even consider the question of HOW,
then it's not an explanation at all. It's just a description.
You're under a misapprehension about PCT, as Mary pointed out. This model
began as a model of human behavior, based on a study not only of real
people doing real things, but of physiology and neurology. The clever
machine to which the model was first applied was the real thing, a human
being. Predictions of human behavior were the first ones made on the basis
of the model. The Little Man arm program, the crowd program, the beginning
of models of the Beerbug, the interest in Pengi and Sonja, came up only as
a way of making the model more detailed, or illustrating the principles of
the model in other settings in ways that might communicate to people in
other fields.
Whether this communication succeeds or fails depends in large part on
whether the recipient of these arguments considers the basic phenomena of
control worth some attention, and whether he or she is willing to consider
the world as consisting of perceptions rather than an objective world and a
perceiver of that world.
Have you tried out the rubber-band demo? I should think that it would offer
an excellent comparison of the interactionist kinds of explanation and the
control-theoretic kind.
···
----------------------------------------------------------------------
Best,
Bill P.