Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!unmvax!pprg.unm.edu!hc!lll-winken!uunet!portal!cup.portal.com!dan-hankins
From: dan-hankins@cup.portal.com (Daniel B Hankins)
Newsgroups: comp.ai
Subject: Re: Making fires and making minds - the laws of physics prevail
Message-ID: <18020@cup.portal.com>
Date: 6 May 89 07:08:25 GMT
References: <3019@tank.uchicago.edu>
Organization: The Portal System (TM)
Lines: 185

In article <3018@tank.uchicago.edu> cs_bob@gsbacd.uchicago.edu writes:

>>>[... why do people believe in free will? ...]
>> 
>>     It's a useful hack that serves until something better comes along,
>>just like Newton's laws of motion, geocentric astronomy, the ether theory
>>of space, alchemy, and other notions too numerous to mention.
>
>I tend to agree that the belief in freewill exists because it is useful, 
>but you should realize that almost any biological trait we possess can be 
>characterized in the same way, so I would hesitate to call it a 'hack'.

     There are lots of biological traits that can be described as hacks. 
Of course, they aren't hacks in the sense that they were invented by some
being, but they are hacks in the sense that they are crude but useful
features.  They serve a purpose which could be achieved better by some
other feature.
     A good example of this is the tails of many simians.  The tail is a
hack in that it is a crude way of maintaining balance when compared to an
advanced nervous system.  Monkeys of various varieties have them, while
more intelligent simians such as chimpanzees, baboons and great apes do
not.


>Your comparison with Newton's laws, etc., is somewhat strained. That is, I
>see an individual's belief in his own freedom of will as something quite
>different from his belief that the world is flat.  Perhaps I'm wrong, but
>the former seems to me to be somehow related to the individual's cognitive
>functioning, while the latter is not.

     So what?  Both are models of the behavior of the world or parts of it.
Remember that like many other so-called cognitive concepts, we first learn
what free will is by having behavior consistent with it pointed out to us. 
We then generalize from that behavior to a concept of free will.  And it is
always pointed out by examples of _human_ behavior, never animal or
natural.
     But once that generalization is made, if one is willing to open up
one's mind to the possibility that other things can have 'free will', then
one can see behavior consistent with it everywhere.  The kind of behavior
predicted by the notion of free will turns out to be identical to the kind
of behavior predicted by the notion of non-linear dynamic systems (i.e.
chaos).
     'Man is the measure of all things' is a recent idea, stemming largely
from Judeo-Christian notions of the place of Man in the Universe.  Some
cultures still do not share it.  No wonder our distant ancestors believed
that free will can be ascribed in equal measure to all active and
capricious things, from animals to weather.


>>More recently, there have grown other ways of looking at these
>>issues that are as or more effective in practice than those based on free
>>will.
>I assume that you're talking about behavioral modification [...]
>[...] Very, very few therapists still cling to the Skinnerian dogma.
>Analysis  is important, if only to arrive at the proper behavioral
>therapy. Simply treating the symptom generally does not make the problem
>go away.

     True, but:  this does not invalidate the central thesis of
behaviorism, which in its purest form is simply radical empiricism.

     Absence of evidence is not the same as evidence of absence.

     If all one is doing in behavioral therapy is treating the symptom,
then one is oversimplifying behaviorism.  I do not believe that behavioral
science is at a point where one can have a full-fledged behavioral
treatment program.  Some aspects of human behavior are understood, but by
no means all or even enough.  A great deal of data collection and
scientific method remains to be performed.

     Trying to treat someone behaviorally on the basis of a few days or
weeks of observation is roughly equivalent to trying to alter the behavior
of an unsupervised neural network program without knowing the full history
of its input.

     Note that even traditional analysis is in some sense behavior
modification;  there are a lot of feedback systems in human behavior (in
Skinnerian terminology, one might call them 'hidden' or 'internal'
behaviors), and by introducing additional information into the system or
modifying that system's feedback through analysis one is performing a kind
of behavior modification.

>Wait a minute. there is a huge difference between the causal determinism
>you've been suggesting and 'an entire wavefront of possibilities'.
>Physicists treat quantum events as though they are truly random events
>that follow a known probability distribution. The so-called 'collapse' of
>the wave function is the actual position at the time of measurement. It
>cannot be predicted, only observed.

     Since you wrote this, I've done some deeper reading into quantum
mechanics, and now understand the issues much better than I did before.

     I was wrong, and you were right.

     Quantum physics doesn't allow us to predict anything except in a
probabilistic way.  The wave function only collapses at the moment of
actual interaction.  Not human observation;  that is not necessary.  I can
describe an experiment which demonstrates this

     But this does not give one free will that a computer cannot have;  QM
applies equally to human brains and computer circuits.  If you wish to
argue that it applies better to brains than circuits, then you have to show
that human neurons amplify quantum effects in a way that electronic circuits
do not.  Even if you could, the electronic circuits could be modified to do
the same kind of amplification.

     If humans can have free will, then computers can too, unless you
define free will as something only humans have.  In which case you have
assumed what you wish to prove.

     In any case, QM does not give anything free will, unless you are
willing to define free will simply as a matter of an imprecise modelling,
an inability to predict the next state.  A simple thought experiment will
show this.

     Suppose that a man gives up his 'free will', and agrees to be told
what to do.  From this point onward, he is a puppet.  His instructions come
from a computer screen in front of him.  He can be told to press either the
button marked _yin_, or the one marked _yang_.  The computer is connected
to a measuring device like the one in Schroedinger's box.  That is, there
is a 50% probability (from the wave function of the isotope) that the
computer will instruct him to press yin or yang.
      So the computer opens the shutter, waits the specified time, and
looks at the scintillation counter.  No emission happened;  the wave
function collapsed and the isotope did not decay.  So the computer
instructs the man to press yang, and he does.
      Is this free will?  Or just indeterminancy?


>>That which we define as Mind is an observable information behavioral
>>phenomenon.  That is to say, there is some set of behaviors sharing
>>certain characteristics that we label as Mind.
>               
>That is not what I define as Mind. Who are 'we'? As far as I can tell, the
>whole problem that Science has with Mind is that it insists on reducing it
>to something it can observe.

     All things must be experienced to be known.  That which is not
observed is not known.  Don't tell me atoms are not observed;  they are
indeed observed, though indirectly.
     After all, how did you originally learn about Mind?  It was either by
having certain phenomena pointed out to you and labelled with the term
(from which your brain generalized the concept), or by being defined as a
grammatical collection of other words, each of which in turn either has a
basis in observation of phenomena or is defined with other words and so on.
At some point all the words must be associated with observed phenomena or
you don't know what they mean.


>>     In order to have knowledge of something, one must observe it or
>>infer it from observable phenomena. 
>
>Or, we could _experience_ it.

     Experience is observation.


>I, for one, have no reason to believe that the human mind transcends
>physical law. For those of you who feel that this implies that there is no
>thing as free will, I suppose that we can leave it at that. My gripe here
>is with a particular defense of the 'the will is not free' position.

     Well, like I said, I will back off on the computability portion of the
argument.

     But I warned you that I was a radical empiricist, and on that position
I will stand.

     I am also convinced that it is reasonable to say that 'free will' and
'incomplete modelling' are equally potent ways of talking about certain
aspects of human behavior.  The notion of incomplete modelling is more
powerful in several ways.

     1. It is more general.  It accounts for the behavior of not only
humans but a whole multitude of other physical systems.

     2. It is more testable.  You can come up with ways to falsify it. 
You can't falsify 'because I felt like it'.

     3. It is more repeatable.  You can try the same thing a lot of times
on a lot of different people and get a nice probability curve, even when
some of the parts of the model remain unknown.


Dan Hankins