From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!pindor Thu Feb 20 15:21:33 EST 1992
Article 3810 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: Re: Definition of understanding (Re: Evidence that would falsify strong AI.
Message-ID: <1992Feb17.225956.24338@gpu.utcs.utoronto.ca>
Organization: UTCS Public Access
References: <6185@skye.ed.ac.uk> <1992Feb14.181324.16278@psych.toronto.edu> <1992Feb14.221800.23311@gpu.utcs.utoronto.ca> <1992Feb16.182943.7817@psych.toronto.edu>
Date: Mon, 17 Feb 1992 22:59:56 GMT

In article <1992Feb16.182943.7817@psych.toronto.edu> michael@psych.toronto.edu (Michael Gemar) writes:
>In article <1992Feb14.221800.23311@gpu.utcs.utoronto.ca> pindor@gpu.utcs.utoronto.ca (Andrzej Pindor) writes:
>>In article <1992Feb14.181324.16278@psych.toronto.edu> michael@psych.toronto.edu (Michael Gemar) writes:
>>>With reference to the exchange Jeff Dalton has been having with folks
>>>regarding what "understanding" is:   
>>>
>>>When Steven Harnad came to the University of Toronto to give a colloquium
>>>on *his* solution to the Chinese Room, he noted, "Everyone thinks that
>>>defining understanding is so difficult.  Well, here..." at which point
>>>he spouted something entirely incomprehensible in a non-English language.
>>>"There," he said, "that was Hungarian.  Did you understand that?  If not,
>>>then you know what understanding involves..."
>>>
>>>- michael
>>>
>>
>>Good showmanship! Very handy at public lectures (or colloquia), but contributes
>>little to appreciating that 'understanding' has many facets.
>>
>
>Nonsense.  While it may be true that there are "degrees" of understanding,
>or that " 'understanding' has many facets," Harnad was merely demonstrating
>that there are some instances in which it is *clear* that no understanding
>exists.  He (and I presume Searle) would assert that that is all that is

The demonstration performed by Harnad is pure showmanship, because it has 
very little to do with a problem at hand. 
1. Even if he said for example (in English) 'fire' , are you sure that 
everyone would understand? Audience might have thought that he meant that the
building was on fire, whereas he might have meant 'fire' like in 'fire a gun'.
Do you agree? CR is input a _story_, structure of which can be analysed
and relationships between terms determined without knowing Hungarian, Chinese
or English.
2. The problem is not whether the person inside CR understands words used to
describe the story, but whether CR understands. By Searle's definition,
the person inside does not understand Chinese, so what is the point of
the demonstration? Is anyone questioning this? On the other hand, the 
demonstration reinforces the notion that whether CR understands is crucially
dependent on the understanding by the person inside. Assume that the person
inside understood the language used to code the story (English, Chinese or
whatever) in a sense of understanding what a hamburger is etc. but was dim
enough to be unable to answer the question (whether hamburger was eaten or not).
You agree that it is possible, right? Nevertheless, following the rules, 
he/she would produce the correct answer. Would Searle agree now that CR
(with the help of person inside) understands the story, because the person 
salivates when he/she sees 'hamurger' written down? Even if he/she would not
be able by him/herself answer the question? We first have to know what
'understanding' we are talking about, before making categorical statements
'understands' or 'does not understand'. Appreciating this fact is ill served
by the demonstration discussed above.

>necessary for the Chinese Room, since (according to both him and Searle),
>the person in the Chinese Room is in the exact same state as a monolingual
>English speaker hearing Hungarian.  One may want to argue about
>the *precise* nature of "understanding", but there are certainly cases
>in which we can agree it isn't present.
>
>- michael
>


-- 
Andrzej Pindor
University of Toronto
Computing Services
pindor@gpu.utcs.utoronto.ca


