From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!mcsun!uknet!edcastle!aisb!jeff Fri Jan 31 10:26:51 EST 1992
Article 3250 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!mcsun!uknet!edcastle!aisb!jeff
>From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.ai.philosophy
Subject: Re: Searle and the Chinese Room
Message-ID: <1992Jan29.015044.11105@aisb.ed.ac.uk>
Date: 29 Jan 92 01:50:44 GMT
References: <1992Jan23.230213.5114@bronze.ucs.indiana.edu> <1992Jan24.150146.4702@aisb.ed.ac.uk> <1992Jan24.224516.12635@bronze.ucs.indiana.edu>
Sender: news@aisb.ed.ac.uk (Network News Administrator)
Organization: AIAI, University of Edinburgh, Scotland
Lines: 105

In article <1992Jan24.224516.12635@bronze.ucs.indiana.edu> chalmers@bronze.ucs.indiana.edu (David Chalmers) writes:
>In article <1992Jan24.150146.4702@aisb.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:
>
>>I just don't see how if you line up programs and recipes and
>>compare them that there's anything in the program text that
>>corresponds to saying what the ingrediants have to be.  The
>>entire program corresponds to the instructions in the recipe
>>for how to manipluate the ingredients.  Eg, bake for 15 minutes
>>corresponds to sort into increasing order, or something.
>
>Recipes and cakes, like programs and computational systems, are
>connected by an "implements" relation, and it's a determinate
>matter whether a cake/system implements a recipe/program.  That's
>all the analogy needs.

The idea that cakes are implementations of recipes is sufficiently
strange that I don't see how it can help explain something about
programs.  I think it's explaining something reasonably clear by
something rather obscure, at best.

If you implement a recipe as if it were a program, maybe you get
a machine for making cakes, rather than a cake.  (Just as if you
implement a program you get a machine that performs the steps
specified by the program, or something like that.)

Well, that's just an example of an analogy that seems more natural
to me.  The particular analogy you want usually seems very unnatural
and other ones, that seem more natural, always suggest themselves
whenever I think about it.  Maybe your analogy really is the right
one, but I find it very hard to find a way to think about cakes
and recipes so that it seems right to me.

Anyway, there's probably not much more I can say about this.

>If you really want to find something that's analogous to
>"ingredients" (which isn't strictly necessary for the success
>of the analogy), consider the role that the clause "cherries round
>the edges" plays in the recipe.  It constrains the possible range
>of "implementations" in a particular way, by imposing the condition
>that implementations have to satisfy a particular physical condition
>(i.e., it must consist in part of cherries, and so on).

What makes the "cherries" mean the little red things is that
there's a person there who can interpret the word.  There isn't
any way to make express a similar constraint in a program.

On the other hand, the constraint that certain objects, treated
as "cherries", go around the edges doesn't suggest the same
difficulty.

>Similar,
>the clause "for i:=1 to 5 do a:=a+i" constrains the possible
>range of implementing computational systems, by imposing the
>condition that implementations have to satisfy a particular
>physical condition (here, this physical condition is a matter
>of causal organization).  There's the analogy, irrelevant though
>it is.

What, exactly, is analogous to "cherries" in this?  That you
can say "physical condition" about both isn't enough.  All sorts
of things might be physical conditons; not all are the sort
that are specified by programs.

>The division of the recipe into an "ingredient" section and a
>"process" section is just irrelevant here.  All that counts is that
>the recipe as a whole determines a range of possible physical
>systems, i.e. cakes, that satisfy it.

Even without specified ingrediants, it determines a range of systems
that satisfy it; but it's a larger range.  The reason I make the
division is that I can't find anything in programs that corresponds
to saying what the ingrediants are.  "for i:=1 to 5 do a:=a+i"
seems like "repeat 5 times", and not like "use a fresh egg, not
just any old egg you've had for a week or so".

>The only point of this argument is to show that the "syntax isn't
>sufficient for semantics" argument doesn't work, as the fact
>that programs are syntactic doesn't mean that *implementations*
>of programs are syntactic (any more than recipes being syntactic
>means that cakes are syntactic) -- rather, implementations are real,
>physical, complex causal systems.  Now, it may be the case that
>"causal organization isn't sufficient for semantics", but that
>would require a separate argument.

Well, as I thought, the direct statement is clearer.

In any case, I'm not sure it really addresses Searle's point.
At least some of the time, what he seems to be saying about syntax
and semantics is that programs process the Chinese symbols in the
questions and answers only in syntactic ways, and not that
implementations of programs are syntactic.

>>That said, however, I think it would be much more fun, and
>>the world would be a better place, if Searle were defeated
>>by the Crumbly Cake Reply than if it was by the boring old
>>Systems Reply.
>
>Thanks, but the Crumbly Cake Reply addresses a different argument,
>i.e. the syntax/semantics argument.  The Systems Reply is still the
>right reply to the Chinese room argument, of course.

A quibble, to my mind.  A defeat on either front might bring
him down.

-- jd


