Path: utzoo!utgpu!utstat!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!hplabs!hpl-opus!hplchm!curry
From: curry@hplchm.HP.COM (Bo Curry)
Newsgroups: comp.ai
Subject: Re: Question on Chinese Room Argument
Message-ID: <280001@hplchm.HP.COM>
Date: 24 Feb 89 00:26:22 GMT
References: <4298@pt.cs.cmu.edu>
Organization: HP Labs Chemical Systems Dept, Palo Alto
Lines: 59

Steven Harnad writes:
>
>There are two senses of "understand," a subjective and an objective
>one. The first (1) is what I mean when I say "I understand English" and
>the second (2) is what I mean when I say "He understands English." The
>first is primary. What I say and mean by "I understand" is based on
>direct, incorrigible, first-person evidence. When I say "HE
>understands," I am merely INFERRING that what's true of him is the same
>thing that's true of me when I understand. I can be WRONG (very wrong)
>about (2) but not about (1). It is (1) that is at issue in Searle's
>Argument, though people keep conflating it with (2).
>
>That's all there is to it. It's not a matter for linguists (any more
>than what I mean by "I am in pain" vs. "He is in pain" is). The only
>ones who worry about mechanisms here are cognitive modelers.
>And I am not a philosopher.
>-- 
I'll have to disagree that a speaker claiming "I understand X" is incorrigible
in the same way as a speaker claiming "I am in pain".  Dennet has written
extensively, and compellingly, on this issue.  When I was a graduate
student (not in philosophy :-) I often encountered students who claimed,
with perfect sincerity, that they understood thermodynamics.  I was
in a much better position than they to judge the truth of their claims.
When I studied Latin, I often thought I understood a poem or passage
(i.e. I had puzzled some meaniing out of it, which I believed corresponded
to the author's intent), and was later (embarassingly) proved wrong.
If the phrase "I thought I understood X" has any meaning, then it
clearly must be possible to be wrong about one's own understanding.
Compare "I thought I understood the menu (but was proved wrong when the
waiter brought my order)" to "I thought the needle hurt, but I was wrong".
The first sentence is perfectly sensible, whereas the latter sounds surreal.
Pain is a much more elusive beast than understanding.
It is also possible to come up with instances when one claims *not* to
understand, yet is mistaken in that claim.  This is a bit rarer, but
seems to occur if the understander expects something subtler or deeper
than is really there.  For example, I may hear a joke, and not find it
funny at all.  I say "I don't get it".  In fact, I have considered several
possible interpretations, but rejected them on the grounds of non-humourousness.
Later, it may prove that one of my rejected interpretations was in fact
the "meaning" of the joke, so that I had really understood it, after
all.  I was misled by my (mistaken, in this case) expectation that a
joke, when understood, will be funny.

Conclusion: An objective test is the only reliable measure of the understanding
of a system.  If the system claims to understand X, but nonetheless fails
the standard test, we are justified in rejecting its claim.  There is
no "incorrigibility" associated with understanding.

All this is, of course, unnecessary to definitively refute the Chinese
room "argument".  As a previous poster pointed out, *Searle's* understanding
or lack thereof is totally irrelevant, since he is merely a *component*
of the room.  Searle's argument (however deeply thought about) reduces
to the claim that "The mechanism is understood, therefore there is no
understanding", which is absurd on the face of it.

Cheers,

Bo "Think deep, dig hard" Curry
curry%hplchm@hplabs.HP.COM