> Brief summary of Penrose's position (with which I strongly disagree):
> Computers can't have intelligence because they are specifically designed
> to avoid randomness due to quantum uncertainty.  This was a mostly accidental
> (but generally fortuitous) result of using large numbers of electrons, thus
> getting statistically predictable behavior.
> Humans, on the other hand, use neurons that might possibly demonstrate
> macroscopic behavior influenced by quantum uncertainty.  Therefore, they
> can't be simulated by computers.  Therefore computers can't be intelligent.

I read the book quite a while ago so I don't remember it well, but I
don't think Penrose's points were those. He rather has a belief (or
at least considers the idea) that our brains are non-deterministic,
and he uses quantum theory to back up this thought, showing a
hypotetical scientifically-approved source for this uncertainity.

> Obvious problems with his argument:
>         does human intelligence really depend on quantum uncertainty?

As I said, this is an hipotesis he uses to show that it is physically
possible to have a non-deterministic brain.

> As near as I can tell, Penrose believes quantum uncertaintly is necessary
> to intelligence because:
>         all known naturally-occuring forms of inteligence involve quantum
>         uncertainty
>         we've never developed intelligent computers (circular reasoning)
>         we can't predict the behaior of human beings, so they must be
>         non-deterministic (false, increases in complexity of a system tend
>         to cause exponential increases in the computation needed to model
>         the system, so inability to predict the behavior of a system does
>         not prove that it is non-deterministic)

You can say that if something can be predicted (to 100% accuracy) by
us then it is deterministic, but the counter-theorem is far from true
and not only because of what you're saying. I don't believe Penrose
will make such a big mistake, you must have misunderstood him. As an
example, the points that don't belong to a Mandlebrot set are
deterministic in nature but they can't be calculated.

Oh, here I am messing with this topic again! I don't think people in
the list will like it, sorry, I promise not to send another message like
this here.

Regards,

Andres Djordjalian
adjordj@aleph.fi.uba.ar