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 10:48:14 EST 1991
Article 1883 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: Searle, again
Message-ID: <5796@skye.ed.ac.uk>
Date: 5 Dec 91 17:50:20 GMT
References: <2127@ucl-cs.uucp> <91338.113617KELLYDK@QUCDN.QueensU.CA>
Reply-To: jeff@aiai.UUCP (Jeff Dalton)
Organization: AIAI, University of Edinburgh, Scotland
Lines: 54

In article <91338.113617KELLYDK@QUCDN.QueensU.CA> KELLYDK@QUCDN.QueensU.CA (Doug Kelly) writes:
>2)I have a sequence of 1's and 0's (stored in a file, for
>example)  What do they mean?
>
>"Well, gee, heck- could be just about anything", I hear someone say.
>
>Exactly.

This sort of thing was discussed a length a while.  Some people
(McCarthy and O'Rourke?) would, I think, be inclined to argue that
if there's an interpretation of a file as rules that the person in
the room could understand and follow, it must be the only right
interpreation of the file (barring a "monkeys on typewriters"
sort of accident).

More specifically the past discussion was about Searle's claim
that the input the the Chinese Room could be interpreted as almost
anything: stock reports, chess moves, etc.  McCarthy and O'Rourke
denied that it could.  I aggued that it could at least have more
than one interpretation (and, indeed, that it was easy to interpret
it as stock market reports, perhaps for imaginary stocks).  The
counter argument was that if there was also an interpretation
that explained more of the structure, it had to be the right
interpretation.

I still think that's wrong, for various reaons, eg, (1) there
might be more than one interpretation that explained lots of
the structure, (2) an interpretation that explained more might
just be more "paranoid", and (3) it's simply wrong to suppose
that if a message (in some language) appears it must have been
put there intentonally (and hence be the "right" interpretation).
Messages found when playing rock music backwards is an example.

>My reading of Searle is that the point he is trying to make
>(in the Churchlands/Chinese room paper, at least)  is that
>
>"A system can not possess understanding based solely on
>symbolic maniuplation."
>
>most of the arguements in this thread seem to miss this point.
>Arguements about neural nets, robotic or analog systems with
>physical sensors, etc.  fall entirely outside the systems that
>Searle was considering in his arguement.  (If he addresses these
>in subsequent work, my apologies)

Searle considers robots, at least, because of the so-called
"robot reply": if you give the Room some sensors, the ability
to manipulate objects, etc, it will understand.  Searle points
out that the outputs of the sensors, the control instructions
for the manipulators, etc, are just more symbols that have to
-- somehow -- be given meanings.  So it's symbol manipulation
again.

-- jd


