From newshub.ccs.yorku.ca!torn!utcsri!rpi!zaphod.mps.ohio-state.edu!darwin.sura.net!paladin.american.edu!news.univie.ac.at!hp4at!mcsun!uknet!edcastle!aiai!jeff Mon Oct 19 16:59:27 EDT 1992
Article 7294 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!torn!utcsri!rpi!zaphod.mps.ohio-state.edu!darwin.sura.net!paladin.american.edu!news.univie.ac.at!hp4at!mcsun!uknet!edcastle!aiai!jeff
>From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.ai.philosophy
Subject: Re: Freewill, chaos and digital systems
Message-ID: <7724@skye.ed.ac.uk>
Date: 15 Oct 92 17:38:52 GMT
References: <1992Sep29.204929.421@mp.cs.niu.edu> <7614@skye.ed.ac.uk> <1992Sep30.200231.2428@mp.cs.niu.edu>
Sender: news@aiai.ed.ac.uk
Organization: AIAI, University of Edinburgh, Scotland
Lines: 128

nr = rickert@mp.cs.niu.edu (Neil Rickert)
jd = jeff@aiai.ed.ac.uk (Jeff Dalton)

jd>                                 A person's thoughts are given
jd> to them (in effect) by unconscious processes.  If they're sane,
jd> their thoughts will make sense, seem like their thoughts, connect
jd> when appropriate to previous thoughts, etc.

nr> You don't have to be insane to have thoughts which don't make sense.

It depends on the kind of non-sense.  The point is that you don't
decide to, say, have reasonably coherent thoughts rather than have
them sliced up and jumbled.  

jd> It's often said that it feels like, or seems like, we have free will.
jd> That's certainly true, in a sense.  But I think it takes only a small
jd> shift of perspective before it seems a lot less like we have free will.
jd> Thoughts occur.  We don't decide what thoughts to have like we might
jd> decide whether or not to go shopping.
 
nr> I don't agree with that analogy.  I do agree that our thoughts are
nr> constrained, just as when we are shopping, our choices of what to buy
nr> are constrained by what is on the shelf.

And so, what?  We get a shelf of thoughts to pick from?  A menu
as in Terminator-1?  What does the picking?

The simple fact is that we don't have thoughts about thoughts about
thoughts and so on back forever, and consequently the decision about
what thoughts to have can't always (if ever) be made in thoughts.

On the other hand, we can obviously have thoughts that have at least
some effect on other thoughts.  For instance, you might be thinking
about something unpleasant, decide to think about something else,
come up with another topic, and succeed in switching over.  Clearly,
some of this is happening in thoughts.  But what happens is that you
have thoughts like "I need to think about something else" without
any deliberation in earlier thoughts such as "what shall I think?".  
(Or if you did happen to have an earlier thought like that one,
you didn't have similar thoughts one step back and so on back forever.)
 
jd>                                        We don't really understand how
jd> thoughts are produced or what factors influence them.  Perhaps if we
jd> looked further into the processes involved it would seem that they
jd> were "free", and perhaps it wouldn't.

nr> I agree we don't know this, and can't until much more is known about
nr> the brain.  I'm inclined to believe that thoughts are free, though, in
nr> the sense that there is no outside force rigidly controlling them.

I can agree with that.  But what if there's an inside force (so to
speak) controlling them?  Because it isn't clear that if we looked at
what's going on inside that "free" would seem the right word for it.

nr> It looks more to me to be somewhat like the behavior of a system with
nr> positive feedback at just below the level where oscillations will occur.
nr> That means it is chaotic in the sense that very small input variations
nr> can cause dramatic changes in output.

Could be.

jd> I can agree that thoughts happen in the (physical) brain.  But I
jd> don't agree that the decision event happens in thoughts.  Instead,
jd> the decision is _given to_ your consciousness in thoughts.  The actual
jd> decision is determined by some unconscious process.

nr> I disagree, but perhaps this is a quibble about words.  I would say that
nr> the decision is made in your thoughts, but the unconscious processes
nr> have constrained the choices available in that decision.

If the decision is made in thoughts, it looks to me like you
must be taking "thoughts" as including at least some of the
unconscious processes I've been talking about.  And so my point
is that it's far from clear that all of this, taken together,
amounts to freedom.  Moreover, with a fairly small shift in
perspective in introspection it can seem that decisions are
given to us (presumably by unconscious processes we know little
about and not by space aliens) rather than made.

That is, even though our decisions often do feel free, if we look
at things a little differently it no longer feels quite so free.

jd>                                            However, here is a thought
jd> experiment.  Robots have been developed that can pass the Turing Test.
jd> Indeed, we've decided that they can be considered persons and we've
jd> even given them voting rights.  The True Blue Robot Company starts
jd> making robots.  Now suppose it turns out that these robots tend to
jd> vote for the Conservative Party (which has blue as its colour).

nr> Hmm.  These robots are sounding very much like my neighbors, except we
nr> call it the Republican Party here.  If the robots uniformly voted
nr> Conservative, we would be very suspicious. 

True.

nr>          But if there was a variation
nr> of voting, with only a preponderance of Conservative votes, and if the
nr> extent to which they voted Conservative were to change from election to
nr> election, you might not be seeing anything much different from the
nr> influences of educational background, religion, parental influence, etc,
nr> on human voters.

Well, we have to assume that the conspirators at the True Blue Robot
Company would take care not to be _too_ obvious.

It at least seems possible that they could produce robots that they
_knew_ would behave in a certain way (either individually or statistically, 
depending on whether they want to determine the decision or merely
make it more probable), without it being immediately obvious from
the robots' behavior (or to the robots themselves) that this was the case.

Now, these robots would presumably think they had free will just like
we do.  Their decisions would feel free to them, they might say the
same things we've been saying about how the things that seem to influence 
our decisions are what really does influence our decisions, and so forth.
Nonetheless, it seems to me that there is still a real question about
whether their voting choices are free. 

In this case, we might be able to look at the records of the TBRC,
interrogate the engineers and designers, check the program listings,
and come up with some solid evidence.  We can't do literally the same
for ourselves.  But can we even do anything like it?  Can we look
at our programming, so to speak?  Are we fundamentally different from
the robots?  If so, how?  If not, can we really say we have free will?
And if so, what happens to our feeling that there was a real question
about free will in the case of the robots?

-- jd


