Re: Human mind not Turing computable according to Eliezer?

From: Christian Szegedy (szegedy@or.uni-bonn.de)
Date: Thu Oct 07 2004 - 09:56:12 MDT


Bill Hibbard wrote:

>Penrose makes a very specific mistake: he uses infinite
>Turing machines to model human brains whereas they should
>be modeled by finite state machines. As I show in:
>
> http://www.ssec.wisc.edu/~billh/g/appendix.html
>
>his argument breaks down if the Turing machines are
>replaced by finite state machines.
>
>Cheers,
>Bill
>
>
>
>
>
I have explicitely stated that I am not convinced by the arguments of
Penrose.
However, it find it not just an overstatement but a completely irrelevant
statetement that he is hopelessly confused about the Theorem of Goedel.
It is a very simple theorem and I am quite sure that
he understands it at least as well as we do. Of course, the way he
applies it
to the real world can and should be disputed. I am not irritated by
criticisms on established scientists, but I don't think it is justified if
*anyone* states that the things are so and so and therefore someone else
(in this case, Penrose, who is a brilliant mathematician) is completely
wrong. A correct attitude is to say, that the observations A,B and C
combined with arguments D,E and F speak against his position on X.
(Exactly the way you did.)

If I am on it, I am not at all sure that the finite-state-machine model is
much better than the Turing machine model. The human mind is in
interaction with a practically infinite universe, so I think that both
models of computations have their specific flaws.

I found your last argument most convincing that the human thinking is
not based on any consistent formal model, but on a combination of
experimentation, probabilistic reasoning and formal logic - an inconsistent
system (not even a model in mathematical sense) but still effective for
a lot of tasks.

However I would not rule out the opinion of Penrose, after all his position
can be correct, even if some of his arguments are not convincing.



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:49 MDT