From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uunet!psinntp!scylla!daryl Thu Jan  9 10:34:20 EST 1992
Article 2578 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uunet!psinntp!scylla!daryl
>From: daryl@oracorp.com
Subject: Understanding and the Systems Reply
Message-ID: <1992Jan8.223252.18468@oracorp.com>
Organization: ORA Corporation
Date: Wed, 8 Jan 1992 22:32:52 GMT

Pardon me for making yet another thread about Searle's Chinese Room,
but...

In his response to the so-called Systems Reply, Searle makes the
following argument: (paraphrased from memory)

     Imagine that a man memorizes a set of rules for
     manipulating Chinese symbols in such a way that he can pass the
     Turing Test for understanding Chinese (that is, he can converse
     fluently in Chinese). If the rules are purely syntactic, the man
     *still* doesn't understand Chinese. Since the man *is* the system in
     this case, then it follows that the system doesn't understand, either.
     So much for the Systems Reply.

Searle argues that, regardless of whether the man can converse
fluently in Chinese, he still doesn't understand Chinese, at least not
in the same way that he understands English. Therefore fluency is not
sufficient to demonstrate understanding.

I would like to modify Searle's thought-experiment in a way that shows
that his conclusion (that the man doesn't Chinese in the same way he
understands English) does not follow from his argument.

Imagine that instead of starting with a man, we start with a computer
program that can pass the Turing Test for English. Without assuming
that such a program can *really* understand anything, I will use the
word "understand" to mean "appears to understand". We then give to the
computer program a syntactic set of rules for manipulating Chinese
characters that would allow it to pass the Turing Test for Chinese.
Now consider the following conversation with the computer:

    Q. Do you understand Chinese?

    A. No, I don't.

    Q. How can you say that? You are perfectly fluent in Chinese!

    A. Well, it's a fake. I don't understand a word of Chinese. For
       example, I have absolutely no idea what the Chinese word for
       "hamburger" is. If this symbol-manipulation allows me to appear
       to understand Chinese, I certainly don't understand Chinese in
       the same way that I understand English.

The conclusion, that the program doesn't understand Chinese in the
same way it understands English, is the same as in Searle's
thought-experiment. However, in this case, we know for a fact that
the computer understands Chinese in *exactly* the same way that it
understands English, by symbol-manipulation.

However, in this case, we can see where the problem lies. The fact
that the computer claims that it doesn't understand Chinese (when
asked in English) is simply due to its inability to relate concepts in
one language (Chinese) to concepts in another (English). It doesn't
English-understand Chinese because it can't translate Chinese into
English. However, it understands English and Chinese in the same
way; it can't Chinese-understand English, either.

Now, back to Searle's thought experiment. When the man says that he
doesn't understand Chinese, he is simply saying that he doesn't know
how to relate Chinese-concepts with those concepts he is more familiar
with. In his case, these include not only English words, but also
sense-impressions, sounds, smells, tastes, etc. This inability to
relate concepts in different languages is all that is needed to
explain why, subjectively, the man says that he doesn't understand
Chinese. Searle's response to the Systems Reply does *not* show that
there is a significant difference between the way the man understands
English and the way he understands Chinese, only that the man can't
relate the two.

Daryl McCullough
ORA Corp. 
301A Harris B. Dates Dr.
Ithaca, NY 14850-1313


