From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!usc!wupost!uunet!mcsun!uknet!edcastle!aiai!jeff Tue Apr  7 23:22:09 EDT 1992
Article 4707 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rutgers!usc!wupost!uunet!mcsun!uknet!edcastle!aiai!jeff
>From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.ai.philosophy
Subject: Re: The Systems Reply I
Message-ID: <6517@skye.ed.ac.uk>
Date: 24 Mar 92 20:09:44 GMT
References: <44765@dime.cs.umass.edu> <6422@skye.ed.ac.uk> <1992Mar18.064723.6873@ccu.umanitoba.ca>
Sender: news@aiai.ed.ac.uk
Organization: AIAI, University of Edinburgh, Scotland
Lines: 119

In article <1992Mar18.064723.6873@ccu.umanitoba.ca> zirdum@ccu.umanitoba.ca (Antun Zirdum) writes:
>In article <6422@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:
>>In article <44765@dime.cs.umass.edu> orourke@sophia.smith.edu (Joseph O'Rourke) writes:
>>>	It seems to me that the way in which a program manipulates
>>>its symbols shows that it has attached some type of meaning to them:

>>Or, to go back to your approach to "understanding X" as being able to
>>answer questions about X.  Now, the Geometry Room, for example, can
>>answer questions aboue geometry (even if the person in the Room hasn't
>>a clue).  But is that because this system understands geometry or
>>because the programmers (or the mathematicians they consulted)
>>understand geometry?

This sort of idea ought to be fairly easy to understand.  For
instance, if I pass a test in geometry, you might conslude that I
understand geometry.  But if you discover that I answered the
questions by pointing a little camera hidden in my glasses at the page
and haveing someone else read the questions and tell me the answers
via a tiny microphone, you would no longer conclude that I understood
geometry.  However, you might well conclude that the person who told
me the answers understood geometry.

Similar conclusions might be reached if the person gave be a book
of instructions that told me how to produce answers to most of the
likely geometry questions and the book was such that I could use
it wihtout understanding geometry.  

For some reason, Antun Zirdum seems to be confusing this idea with the
claim that someone who _learns_ goemetry doesn't understand it -- only
their teachers do:

> [...]  If we follow
>your argument above to its relevent conclusions then nobody
>understands mathematics, it is always the teachers that they
>consulted with that understand mathematics!

It should be clear that in some cases the person answering
the questions understands, but not in others.  

Anyway, here's what filled the "..." above:

>I am capable of programming a computer program, that when run,
>is able to manipulate mathematics that is beyond my level.
>So clearly there can exist a program that functions in ways
>that its designers never could have forseen.

It's certainly clear that there can exist programs that function
in ways their designers did not forsee (eg, programs with bugs).
Whether they _couldn't_ have forseen it is a more difficult question.

>I do not think that you have a clear understanding
>of what it means to understand! For example, Subjectively
>you understand X! Now what does that mean?

As I tried to explain in other messages, I'm not going to play the
game where other people demand definitions and I have to produce
them.  If you want to make progress on this point, you might try
telling me what you think it means to understand, and why you
think "the system" satisfies this definition.  Of course, I will
understand if you too decline to play this game.

>>But the situation is a bit different when we come to the "syntax
>>isn't enough for semantics" arguments.  The person in the Room is
>>performing some manipulations that depend only on "sytax" such
>>as the shapes of the squiggles.  And the person doesn't understand
>>what the symbols mean.  How is it that the system can do any better?
>>
>If we are to go in that direction then I must argue that
>syntax is the symbols, semantics is the actual manipulation
>of the syntax. After all, the CR is not merely a collection of
>symbols, it also encompasses a manipulation of those symbols
>- that cannot be reduced to mere syntax! (however you cut it!)

So the symbols acquire meaning by being used in certain ways.
There might be something to that.  On the other hand, the mere
fact that the symbols are being manipulated does not give them
meaning.  For instance, I could manipulate them by shaking them
up and throwing them on the floor.

There's an important distinction here (or somewhere around here).
You may recall that Dave Chalmers has argued the programs specify
a causal structure; consequently, an implementation of a program
isn't purely syntactic.  You may be making a similar point here,
about "manipulation".

This sort of argument may answer some of Searle's points, such as
when he says (in the 2nd Reith Lecture) "programs are defined purely
syntactically".  But when he addresses the systems reply in the same
lecture, he says something different:

  There's no way the system can get from syntax to semantics.
  I, as the CPU, have no way of figuring out what any of these
  symbols mean, but then neither does the whole system.

And earlier in the lecture:

  The rules specify the manipulations of the symbols purely
  formally, in terms of their syntax, not their semantics.
  So a rule might say: `Take a squiggle-squiggle sign out of
  basket number one and put it next to a squoggle-squoggle 
  sign from basket number two.'

  ...

  There you are, locked in your room, shuffling your Chinese symbols,
  ...  On the basis of the situation as I have described it, there's
  no way you could learn any Chinese simply be manipulating these
  formal symbols.

The rules that specify the manipulations refer only to syntactic
properties of the symbols (not to meanings).

Now, it turns out that I don't think Searle has quite succeeded in
showing that the system doesn't attach meaning to the symbols (see
my article of 23 Jan in which I quote the same passeages as above).
But neither have I seen a convincing explanation of how the system
could attach meaning.

-- jd


