From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wupost!uunet!mcsun!uknet!edcastle!aiai!jeff Mon Dec  9 10:48:32 EST 1991
Article 1914 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wupost!uunet!mcsun!uknet!edcastle!aiai!jeff
>From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.ai.philosophy
Subject: Re: A Behaviorist Approach to AI Philosophy
Message-ID: <5816@skye.ed.ac.uk>
Date: 6 Dec 91 19:23:13 GMT
References: <YAMAUCHI.91Nov24030039@magenta.cs.rochester.edu> <5727@skye.ed.ac.uk> <YAMAUCHI.91Nov27203011@magenta.cs.rochester.edu> <5739@skye.ed.ac.uk> <1991Dec6.020944.4967@syacus.acus.oz.au>
Reply-To: jeff@aiai.UUCP (Jeff Dalton)
Organization: AIAI, University of Edinburgh, Scotland
Lines: 85

In article <1991Dec6.020944.4967@syacus.acus.oz.au> william@syacus.acus.oz.au (William Mason) writes:
>Gday,
>      I find much of this interesting.  However I'm not happy to
>find some circular arguments as represented below.
>
>>Again this might depend on exactly how it works.  Moreover,
>>if the materials are sufficiently different, the behavior will
>>also be different, though perhaps not in significant ways.
>
>In mathematics, there is a simple agreement that says we'll all argee
>on the starting conditions before an exploration begins.  Statements
>like the above seem to add nothing.  In my mind the above translates to
>something like ...
>
>	OK, if I ignore/discount operative assumption #2, then I can prove
> 	that all deductions based on it invalid.
>
>Of course!  Is that really the point, though ?
>
>I did not cite a source, because we are all guilty of this at one time
>or another.  [ back to our regular programme ... ]

You should have cited the source.  I find this kind of indirect
attack much more annoying.

I don't try to give a full argument in every message.  It takes too
much time, and the same points keep coming up again and again.  Before
dismissing one of my articles in this way, you might want to consider
that it was posted in the context of a larger discussion in which I
had already explained at greater length.

Let's restore some of the context, at least:

   Case III: A robot behaviorally indistinguishable from human, similar
   in physical capabilities and structure (bipedal, two dextrous
   arms/hands, stereo vision/hearing, etc.), but very different in
   appearance (e.g. kevlar and titanium rather than skin and bones).

To which I replied:

   Again this might depend on exactly how it works.  Moreover,
   if the materials are sufficiently different, the behavior will
   also be different, though perhaps not in significant ways.

Now let me unpack what I said, slightly, wiht some more indiction
of context:

1. There's an argument that only the behavior matters.  However,
that has to be shown, not just asserted.  What is there other
than the bahavior?  Well, it might matter how the behavior is
produced.  (This has been discussed at length, in several posts,
and not just by me.)

2. There is an implicit claim here that a robot built of different
materials than are humans could have behavior indistinguishable from
humans.  I note that this will not be possible, if the materials are
sufficiently different.  (It should be obvious that this might be so.
For example, if the robot is more massive than a human, it would have
greater inertia.)  However, the behavioral differences might not
be such as to be significant for the question of whether or not it
understood.)

That is, this thing you called circular, and a discounting of operative
assumptions, was nothing of the sort.

>So if I say X is indistinguishable from human behaviour, then it follows 
>that X's behaviour would NOT be "different" as I understand the term.
>
>Alternative debating styles that say X is different, belong in the
>arena of rehetoric.

Let me repeat my point, just in case that's necessary, since you
repeat yours.

If you say "X is indistinguishable from human behaviour", this may be
an assumption that cannot be realized.  (Consider "suppose a glass of
water has behavior indistinguishable from human behavior".)

However, nice person that I am, I'll let you have this point
("perhaps not in significant ways") and leave you with the
reference to my earlier arguments that glasses of water might
well be producing the pehavior in a different way, a way that
did not count as understanding.

-- jd


