From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!midway!spssig!markrose Tue Nov 19 11:09:26 EST 1991
Article 1243 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!midway!spssig!markrose
>From: markrose@spss.com (Mark Rosenfelder)
Newsgroups: comp.ai.philosophy
Subject: Re: Chinese Room Variant
Keywords: ai philosophy searle expert system
Message-ID: <1991Nov08.171238.37090@spss.com>
Date: 8 Nov 91 17:12:38 GMT
References: <1991Nov7.151439.3353@osceola.cs.ucf.edu>
Organization: SPSS, Inc.
Lines: 38
Nntp-Posting-Host: spssrs7.spss.com

In article <1991Nov7.151439.3353@osceola.cs.ucf.edu> clarke@next1 (Thomas Clarke) writes:
>[...] The child is also taught the rules for forming @ and & of longer strings 
>in terms of the tables for singleton strings.
>
>After much tutelage by an expert system, the child is able to perform & and @  
>for arbitrary pairs of strings with nearly flawless accuracy.
>
>My question is:  Does the child understand the concept of number? arithmetic?  

1. These arguments often depend for their plausibility on extraneous factors--
in this case, on renaming the integers with arbitrary signs.  There would
be no difference in the child's experience or abilities (whatever they mean)
if you used the integers instead of letters, + and * instead of @ and &.
You've just taught the poor kid a different set of symbols.  This renaming
contributes nothing to your arguments, but makes the child's resulting
abilities seem more divorced from "real arithmetic."

2. Your child, then, can add and multiply... only.  For this reason he's a
bit of a straw man, because you ask if he knows arithmetic.  Of course he
doesn't-- he knows only a few of its operations.  (Remember that the
computer, which is the ultimate target of your story, is better at arithmetic
than you.)

3. The meat of the problem, of course, is that the child has not been taught
any correspondence between the rules he is learning and any actual use of
them.  Confronted by a problem such as finding the total cost of four Ninja
Turtles comic books at 75 cents each, he would have no idea that the skills
taught him by the strange man in the white coat have anything to do with it.
So no, he doesn't know arithmetic.  

If you really want to see the results of such an experiment, look at most
schools, worldwide, which mostly teach operational techniques, and rarely
relate them to their uses... Richard Feynman gives some appalling examples
in _"Surely you're joking, Mr. Feynman!"_  But I digress.

Relative to computers, this point is again a bit of a straw man, since 
the use of arithmetic in a computer is rarely just for fun-- it's always
serving some practical purpose.  But it's true that the CPU doesn't know that.


