Re: Searle's Chinese Room Argument

From: HARNAD, Stevan (
Date: Fri Mar 17 2000 - 17:37:27 GMT

On Fri, 17 Mar 2000, Brooking, Stephen wrote:

> What I was trying to say was that I don't see that what is going on in our
> heads is like a program running on a computer, but the program and the
> computer are one and the same. Does that make it any clearer?

Not entirely. Computation is implementation-independent algorithms. If
you implement them, they are running on a computer. Most things
(including brains) are simulable, and the simulation is run on a

But what does it mean to say that in our heads, program and computer are
the same?

(1) Is the brain just running an implementation-independent algorithm or
not. (If yes, proceed to Searle. If not, proceed to (2))

(2) If the brain is not just running an implementation-independent
algorithm, is it running an "implementation-DEpendent" one? (Is that
what you mean by computer and program are one?) But then what does that
mean? Or do you just mean it's hybrid: part computational, part not?

If the latter, I would agree...

> > > Brooking:
> > > Why is the answer to "Could a machine think?" obviously yes? I would argue
> > > that it is no, and I certainly wouldn't say that it is obvious. I agree
> > > with Blakemore's point.
> >
> > Harnad:
> > Because it is not at all obvious that we ourselves are not machines! A
> > machine is just a causal system.
> Brooking:
> I accept that it is not at all obvious that we ourselves are not machines,
> but I don't see the progression to the claim that a machine could
> obviously think.

Because if we ARE machines (and since we DO think), that means we are
machines that think, and that means machines can think.

> Brooking:
> If we are machines, then you're making a human machine, but the point is
> that if it is possible to produce a human artificially, and it is done,
> then the result is a replication of human life, rather than just an
> intelligent machine. It seems to me that if you can make a human-like
> machine, then it would not have artificial intelligence but intelligence.
> There is a problem here with how we determine what artificial really means
> - if we can produce a human artificially, then it could be argued to have
> an artificial version of intelligence. But what does it mean to produce
> artificially a human-like machine? Presumably by any means other than the
> norm for reproduction, but if the machine was sufficiently like a human,
> made from the same materials, then is it artificial, or is it that the
> life (if it is acceptable to say that it is alive) has been started later
> than normal?

Two distinctions: machine vs. non-machine (you agree that if "machine" just
means a deterministic causal system, then we are probably machines.

man-made vs. non-man-made: If we made a T4 or T5 brain, it would be
man-made, but so what?

The relevant thing (and perhaps this needs to be stated very
explicitly): The assumption is that if you design a T2 or T3 or T4
system, you designed it out of an understanding of HOW it manages to
pass TX (algorithms only, if T2 by pure computation; hybrid systems if
any T3 or T4).

Obviously simply cloning a system is of no interest (nor is the clone's
passing of T-whatever), because we did not design it, and hence do not
know how it works.

This is all reverse-engineering: We want to figure out how things work,
not just clone them... Cloning leaves us as clueless as before.

> Brooking:
> It is difficult to argue that the brain would not work if implemented in
> silicon. But it is not so for other parts of the body - Chalmers' example
> of the digestive system for example, in which he said that if you slowly
> replace the parts of the digestive system with metal, a point would be
> reached where no food groups will be broken down and no energy extracted.

The only reason it SEEMS easier to imagine the brain implemented in
silicon is because you can't SEE thinking but you can SEE digestion.

But note that the "silicon" question has two components. (1) Is thinking
just implementation-independent computation (which can, among other
things, be implemented on a silicon-based digital computer) and (2) Even
if thinking is not just computation, could the HYBRID system be
implemented in some other materials (like an artificial heart).

The answer to (2) is: maybe it can (T4).


This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:36:27 GMT