[From Bill Powers (920522.0900)]
Interchange between Greg Williams and Gary Cziko:
Greg:
My central claim is that the moment-to-moment operation of an individual's
control structure is NOT MIRACULOUS, but rather a function of its past
history (perhaps in a probabilistic way -- I have no desire to get into
questions about "absolute" determinism).
Gary:
Who could argue with this? How could an individual's control structure
be anything but a "function of its past history?"
Greg:
Anyone who buys into state-determined dynamical models won't argue. >Some
who don't, will. Some of the latter postulate "miraculous" (non->history-
determined) alterations in organismic physiology due to a >"self" and/or
"God," among other (non-physical?) things. I was just >trying to make sure
that everyone realizes that such postulations are >ADJUNCTS to PCT.
Point 1:
If a miracle occurs, it becomes part of the history of the affected system.
To call something a "miracle" is only to say "I don't understand how that
could have happened." To a believer in lineal cause and effect, the
unlikely event of the knot between two rubber bands staying nearly
stationary while two people wrestle the ends of the rubber bands around is
scientifically unexplainable: it is a miracle. The fact that the
environment reinforces organisms for doing what is good for themselves is a
miracle.
In fact, gravity is a miracle, radioactive decay is a miracle, the
constancy of the velocity of light is a miracle, quarks are miracles, the
Big Bang is a miracle. There are no physical explanations for such
phenomena: they happen or exist, and that is all we can say about them. We
can hypothesize causes for such phenomena such as luck or divine
intervention, but we still don't understand how the phenomena came to be.
Miracles are things we can't explain. To call them miracles is normal human
egotism; it is a subtle claim that nobody will ever understand more than we
do. Every generation since the dawn of history has thought it had finally
reached the ultimate of human achievement and understanding. This was true,
of course. Only it was just a position on a slope.
Point 2:
The concept of a "state-determined system" comes, I believe, from LaGrange.
Give me a description of the universe at a given instant, and knowledge of
all natural laws, and I will tell you the state of the universe from then
on. This is what physical determinism is about, isn't it?
Quantum mechanics put the first nail in the coffin; Chaos screwed the lid
down. Control theory threw the first handful of dirt into the grave of
LaGrange.
The problem, as we now know, is that the LaGrangian concept works only
backward. Given the present state of the universe and knowledge of all
natural laws and intervening events, I could tell you pretty much what the
state of the (local) universe was a week ago. But given the state a week
ago, I could not have predicted the local state today. That's because of
hypersensitivity to initial conditions, which occurs in enough phenomena to
make sure that no measurement of initial conditions can be precise enough
to distinguish the actual trajectories from the possible trajectories of
all variables indefinitely into the future. Looking backward, we can see
how the confluence of various forces led to the present state of affairs;
the initial conditions must have been within some very narrow range. But
picking sets of initial conditions at random from within that narrow range,
we would often predict very different outcomes from each set, even though
the sets differ only infinitesimally -- or not at all. In the limit, we
find that infinitesimal changes of initial conditions can often lead to
finitely different outcomes: the famous "bifurcations" of chaos theory, the
butterfly effect. If this can happen for some phenomena, there is a finite
horizon past which no prediction is even in principle possible. For complex
phenomena, this horizon can be measured in days, or even seconds, depending
on the details.
Control theory introduces a type of physical organization that can
eliminate all but a narrow range of the outcomes of chaotic and randomly
perturbed processes. What determines the outcome is then not "history", but
the present organization of the controlling living system. Once the fact of
control became established in one part of the physical universe, that part
became able to preserve the kind of organization that controls, and passed
along from generation to generation this ability -- the ability to
predetermine outcomes independently of the laws governing the rest of the
universe, the ability to manipulate physical variables so that their lawful
effects are those intended by the controlling system.
This phenomenon was not imagined by LaGrange. LaGrange imagined blind
natural forces that could produce only the outcomes they produced, so that
only one outcome was ever possible. He knew nothing about hypersensitivity
to initial conditions (although he could have figured it out by
contemplating his mathematics a little further). He certainly did not
understand purposive behavior -- in his day, that was strictly a religious
or mystical concept in the realm of miracles. To LaGrange, as I understand
his ideas, physical determinism was absolute and fixed forever. This idea
can no longer be supported.
Control theory provides the final justification for "free will." Free will
is not a phenomenon of causation, but a phenomenon of organization. It
isn't a very good term for this phenomenon, but it's the best people could
do by way of describing their relationship to the nonliving world.
Nonliving objects are clearly constrained to behave if and only if, and
exactly as, present-time external influences cause them to behave. Within
the horizon of prediction, their behavior is explainable by the LaGrangian
hypothesis. In fact, history has no influence on them, because one does not
need to know the history of the universe to predict the effects of current
causes. All that is needed is a description of unchanging natural law, and
of the current state of affairs. How that state of affairs came to be is
irrelevant.
But moving among these objects and interacting with them are objects of a
different kind, which carry purposes within themselves and are thus capable
of diverting the flow of cause and effect in the nonliving world to produce
consistent and repeatable consequences in a highly "unnatural" way. No
longer is there hypersensitivity to initial conditions; in fact, even
rather large differences in initial conditions make no significant
difference in the outcome, just the opposite of processes in the nonliving
world. The effects of disturbances, systematic or random, are
systematically counteracted to keep the outcome from straying far from the
intended one. Certainly, the effects are sufficient to keep the outcomes
far more consistent than they could have been without the intervention of
living systems. As a result, the horizon of predictability is extended far
beyond the limits of a purely Lagrangian universe. Prediction becomes a
matter of understanding intentions. In my case, an intention formed nearly
40 years ago is still determining outcomes, despite four decades of random
perturbations and tendencies of natural processes to branch and branch
again as they pass natural bifurcation (or multifurcation) points.
The term "free will" does not designate something that exists in an
objective universe (or not) and which we then must prove (or disprove) to
be something real. It's an attempt to describe an experience. Control
theory, by showing how purpose works, suggests a clearer view of this kind
of experience. It supports the idea that this phenomenon is not shared by
all physical structures in the same way physical properties like mass and
energy are shared. It is something unique to living systems, giving them
hegemony, for better or worse, over all nonliving objects, and even over
the course of natural -- but nonsentient -- events.
···
----------------------------------------------------
Best,
Bill P.