[Wittrs] Re: Nominalism / Neil

  • From: "jrstern" <jrstern@xxxxxxxxx>
  • To: wittrs@xxxxxxxxxxxxx
  • Date: Thu, 10 Sep 2009 16:00:22 -0000

--- In Wittrs@xxxxxxxxxxxxxxx, "iro3isdx" <xznwrjnk-evca@...> wrote:
>
>
> --- In Wittrs@xxxxxxxxxxxxxxx, "jrstern" <jrstern@> wrote:
>
>
> > Be very careful, or you'll simply be describing nominalism!
>
> I am not troubled by that.

Excellent.


> > However, I believe Turing, Church, and Godel, and maybe Post,
> > include full enumerations in their theories, including all
> > possible errors as well as correct theories.
>
> Not the kind of errors that people make.  For example, people
> will make ambiguous marks, then come back and fix them to
> remove  the ambiguity. And the way they do this is probably
> not rule-based.

There are "problems of error" I worry about,
but these ain't those.


> >> The computer is an electro-mechanical device. You, and
> >> others, make a computational interpretation of what that
> >> device is doing.
>
> > I believe that is too simple, that is, untrue.
>
>
> > Comes down to the paint on Searle's wall.
>
> Searle's wall is not running the WordStar program, no matter
> how you look at it.

Quite.  And no other interpretational view is
significantly better.

I mean, sure a cat is a cat only if I interpret it
as such, if I interpret it as a dishrag, well, it
may make a nice dishrag in a certain context.  But
it's still a cat, too.

Hey, look at me, I made an ontological statement!
Don't like doing so, better go reconsider this.


> AI folk could avoid the "problem" by calling themselves
> mechanists, rather than computationalists.

Well, both, I suppose.

>They talk about cognition being
> mechanism anyway, so this should not be a problem for them.
> And it might avoid having to argue about what is or is not
> computation.
>
> Of course, it might turn out that we don't have a clear
> definition of what mechanism is, either.

I approve of "mechanism".  Shanker calls the entire Turing
project "mechanist".  I think that's right.

But just slapping the label on it, is not enough.  How is
it a mechanist system works?  "Derived intentionality" just
pushes the problem around, doesn't touch it.

And you could have all sorts of mechanistic systems
that are irreductionist in one or another ways, perhaps
someone will say it's impossible to reduce from letters
to bits, something essential would be lost.  Or, how about
Penrose's microtubules - quantum "mechanist", doncha know.
Yes, I think we'd better keep the "computational" part.


> > That seems odd. Even by your theory, I put symbols in
> > there. So, they are symbols only when I'm watching?
>
> You are interpreting that as symbols.  The computer itself
> is not interpreting it that way (or not interpreting it at
> all), so nothing  in the computer's physical behavior
> depends on whether you consider  parts
> of what is happening to be symbolic.

But wait, if I write a mundane payroll program, and
have a variable "person", and a statement person = "John",
and the computer executes that, you want to say the computer
is not using symbols?

That's a very common view of computing.  In fact, it's
exactly the compsci eliminative view that Searle uses as
the basis for his entire work.  Now, I guess you're being
a little more eliminative, but again, the Good Old Fashioned
Computational (GOFCO) view is that computation "is functionalist",
that Searle appropriately savages.  GOFCO pushes an abstraction
as its theory, functionalism being that abstraction.

It does not work.

In fact, it does no work.

So, here's the spectrum.

On the west horizon, that computation works by correspondence.

On the east horizon, that computation works by derivation.

Right in the middle, is the idea that computation works
by mechanism.

Each view has at least some utility, but the mechanist
view, I believe, is both the most important (by far),
and the one least explored, just perhaps because we
are standing right on top of it.


> >> That seems misplaced. I would think of eliminativism
> >> as the denial that there is any belief, and perhaps the
> >> denial that there is any discourse. I don't see that
> >> in Wittgenstein. Surely, he is just pointing out what
> >> should be obvious, namely that the rule based
> >> explanations that are traditionally given are failures.
>
> > Well, that is problematic.
>
> > It suggests rule-based would not work for machines,
> > either, when obviously it does.
>
> I'm not sure where you are getting that.

Well, how else can we look at it?

If LW tells us rule explanations are insufficient for how
Johnny extends the sequence 2,4,6 ..., just why would it
be any better for a machine?

Because, if it were better for a machine, just what would
be wrong with using it for a human?

Oh, sure, you can probably make up an answer for this either
on your own, or synthesized from other LW statements.
"People are not computers!" you say, shocked, sarcastic,
and sanctimonious.  But that is never the question.  The
question is about the grammar, and whether a grammar can
be a rule - in fact, whether a grammar must be a rule.
LW makes his points, but only by trying to have things
both ways, because calling it a "grammar" does not really
answer the basic rule question, just tries to dodge it.


Josh


Group Home Page: http://seanwilson.org/wittgenstein.discussion.html
Group Discussion Board: http://seanwilson.org/forum/
Google Archive: http://groups.google.com/group/Wittrs
FreeList Archive: //www.freelists.org/archive/wittrs
FreeList for September: //www.freelists.org/archive/wittrs/09-2009
FreeList for August: //www.freelists.org/archive/wittrs/08-2009
Group Creator's Page: http://seanwilson.org/
Today's Messages: http://alturl.com/whcf
Messages From Last 3 Days: http://alturl.com/d9vz
This Week's Messages: http://alturl.com/yeza
Yahoo Archive: http://groups.yahoo.com/group/Wittrs/

Other related posts: