Re: randomness and free-will (long post)

Tom Riley (TRILEY@VMD.CSO.UIUC.EDU)
Sun, 29 May 1994 05:07:46 CDT

\in reply to Dwight Read on randomness, free will and crossing the
grass!

----------------------------Original message----------------------------
In reading through the posts on free will I get the sense that roughly the
following argument is made:
(1) free will => behavior is unpredictable

(2) random phenomena => unpredictable phenomena

therefore free will can be subsumed under random phenomena.

This seems to run into an objection and a counter argument:
Objection: What we call "random" often is merely a way of saying we don't
understand the causal processes involved (which may be many and complicated)
or that we lack sufficient information to make a prediction (e.g., Graber's
comment about the tossed coin being predictable if we had sufficient
information...).

Counter argument: At a quantam level there may be truly random phenomena.

But the counter argument has a counter argument, as suggested by Yee.
Namely, the linkage betweeen random phenomena at a quantam level and my
decision to write this post is extraordinarily tenuous.

Suppose we take a different tack and try to define what constitutes free
will. On definition might run as follows. If an agent makes decision A at
time t under a given set of conditions, and at anotther time t' makes a
different decision B even though identical conditions hold at time t' as did
at time t, then the agent is exhibiting free will.

This leads to a problem: What process could lead to different outcomes under
identical conditions? This brings back randomness as by definition truly
random phenomena would have this character. But to say that randomenss would
lead to the above property, and that free will would lead to the same
property (that is, lead to different outcomes until indentical conditions)
does not establish that free will = randomness as it is possible to have two
different processes which overlap in at least some of their outcomes.
Another way that free will could come about (which would not be acceptable as
an argument in scientific discourse) is that there is some extrinsic quality
that makes it possible for the properyt to be true (e.g., god gave us the
capacity for free will).

But is this strict definition of free will necessary? Consider
nonlinear systems which are sensitive to initial conditions. WHile such
nonlinear systems are deterministics and would NOT exhibit the equivalent of
different behaviors under identical conditions, identical conditions only
exist in ideal domains such as mathematics. What ever means the brain has
for representation of, e.g., numbers, it surely does so only with a certain
degree of precision, so that even under identical external conditions, the
brain's internal represenation will deviate in some manner from the external
values. Or to put it anothter way, two situations that are PERCEIVED to be
identical from the viewpoint of the brain's representation, may deviate
slightly from each other. If the process used by the brain to go from inputs
to behavior is highly nonlinear, it follows that very different decisions can
ensue (even if they are driven determinisitically) from what are perceived as
identical conditions; i.e., we will have the illusion that we are making
"free will" decisions; further, it would not be possible to demonstrate that
this is merely an illusion unless one can measure with greater accuracy the
inputs that are acted upon by the brain than the accuracy with which they are
perceived by the brain at the time the behavior takes place.

This assumes the brain is acting in some highly non-linear manner--which
certainly cannot be ruled out in advance, as O'Brien comments (to paraphrase
what he said).

Let me trot out again my earlier model for free will as yet another
alternative. If we recognize that there are two levels at which processing
and decision making takes place, namely a language driven, consciousl level
and a pre-language, non-conscious level, then we might consider the conscious
level as evaluating the deision made at the non-conscious level and embedding
that decision into a "meaning system" which allows the decision to be
evaluated against other meanings to arrive at a higher level decision which
can override the lower level decision. For example, often people walk
directly across a lawn rather than following the sidewalk. It is not
unlikely that something like the following takes place. The visual system
feeds information into a part of the brain which calcuates an optimal path in
terms of energy expenditure and conscludes that walking across the lawn using
a straight path is the optimal path. This kind of decision making would be
occurring at a pre-language, non-conscious level. That decision is highly
predictable. However, before that decision is acted upon it is feed into the
conscious level of the brain which evaluates that decision in terms of a
meanings such as: If I walk across the lawn I will be contributing to
creating a path and I consider that lawns with paths are highly undesirable.
Or I might consider that to walk across the lawn will cause others to think I
am a selfish boor. i.e., it is not implausible to consider that the
deterministic decision made at the non-conscious level is evaluated by
criteria unrelated to how the non-conscious mind made the decision, and that
evaluative process may (or may not) lead to an overide of the decision make
by the non-conscious level. Now it may be that the conscious level is also
determninistic (but clearly there is much less certainty as to the
associations that will be made with the "walk across the lawn decision" than
there was to the non-conscious decision to walk across the lawn; that is, the
latter might only involve a few simple, evolutionarily selected strategies
and so be very predictable, whereas teh former invovles making associations
which are not so obviously connected to the posited behavior). Even if that
is the case, we have the possibility that what free will "means" is:
override the lower level decision. THat is, free will is the concept we have
invented to talk about this override process, and like other cultural
constructions, we take it as having reality.

By positing two levels it is possible to accomodate that fact that at one
level many behaviors are very predictable (e.g., people will walk across the
lawn) but doesn't require that all behavior be seen as deterministically
driven in a simple manner; i.e., the evaluation that is made at teh level of
meanings is obviously predicable only with extreme difficulty or may even be
unpredictable in the sense of non-linear systems by virtue of small
differences leading to major shifts in terms of evaluations that are made.

D. Read
READ@ANTHRO.SSCNET.UCLA