From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!pindor Thu Feb 20 15:22:07 EST 1992
Article 3866 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!pindor
>From: pindor@gpu.utcs.utoronto.ca (Andrzej Pindor)
Subject: What is it that Searle's Chinese Room understands?
Message-ID: <1992Feb19.162222.18604@gpu.utcs.utoronto.ca>
Sender: pindor@gpu.utcs.utoronto.ca (Andrzej Pindor)
Organization: University of Toronto Computing Services
Date: Wed, 19 Feb 1992 16:22:22 GMT


    In the following I'll assume that the readers of this posting are
familiar with the original article by Searle. However I'd like to repeat 
the story which Searle feeds the ROOM (first it is English Room, then
Chinese):
   "A man went into restaurant and ordered a hamburger. When the hamburger
arrived, it was burned to a crisp and the man stormed out of the restaurant
without paying for the hamburger or leaving a tip."

Then the Room is fed a question "Has the man eaten the hamburger?"  
The problem which Searle tries to deal with is whether correct answer to
this question (together with correct answers to modifications of this story)
can lead to a conclusion that the Room understands the _story_. I'd like
to stress very strongly that the problem is about understanding the story
and not the language it is written in. To explain what I mean by this
distinction (in case it is not clear to some people, 'understanding' being
such a complex term, contrary to what some people like Hanard would like
others to believe) let me propose a following alternative story:
   "A man went into a bawdy house and asked for a mature brunette. He was 
shown into a room where he found a woman who could be his grandmother. He 
stormed out of the house without paying or leaving a tip"

Now I assume that most people would agree that these two stories have 
something in common, that at a certain level they are basically the same. At
this level it is irrelevant whether we are talking about a restaurant or
about a bawdy house. The only role the precise meaning of the words play is
in helping us uncover this deeper (or more general) sense of the story. This
is the sense which CR program uncovers, i.e. understands. Everyone will agree
(I trust) that if a story like this was told to a person in a language which
he/she understands, but the person would be unable to answer the question
correctly, you would conclude that the person 'does not understand' the story.
By the same token, if CR answers the story correctly, it _understands the 
story_. 
     Clearly there are different understandings involved here. There is  
understanding which involves sensual experiences which correlate with words
or situations they describe and understanding which correlates all parts
of the story together. In the second case the direct meaning of the words 
used is only a tool used to code (or uncover) the meaning of the story. This
is the meaning which CR can uncover by syntactical manipulations, without
having to have direct sensual experiences relating to the words. If
someone wants to call this a 'superficial' understanding, its fine with me.
My only purpose here is to point out that there is a meaning of the word
'understanding' which applies to CR. I may however add that some people might
consider that understanding of the above story, which is limited to salivating
at a word 'hamburger' but does not include ability to answer the question
whether the man ate the hamburger, a very superficial understanding.
    The above argument ties of course very strongly with a distinction between 
ontological and epistemological, pointed out on this list very recently.
-- 
Andrzej Pindor
University of Toronto
Computing Services
pindor@gpu.utcs.utoronto.ca


