Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: Physics is a human construct



In essence, Penrose argues that a computer
could not simulate human consciousness
because the mind is not a digital device.

I'd have to say you missed the point. His argument (convincing to me) is
that a machine can never be said to understand anything, though it may
give answers to many problems. I can recommend Shadows of the Mind, in
which Penrose uses a cleverly chosen Goedel example that we can
understand, but which no algorithmic-driven computer could. He also
speculates on how the mind/brain differs from any possible computer.

How do we assess that anyone understands anything? In university
we try to do this by giving examinations. Using these we can
clearly demonstrate things that are not understood, but we can't
yet determine objectively that something is understood because a
student has submitted what is, on its face, a correct answer.
Every teacher has, to his dismay, seen counterexamples.

I don't believe that it is practical to ask for an operational
definition of "consciousness" or "understanding" beyond the
depth of satisfaction of the Turing test, since that is really
the only test we use on our students and peers. Penrose didn't
get beyond it; I doubt anyone will within present limitations.

I will now state that a machine can be said to understand
something*. My machine understands my intent when I throw its
power switch to the "on" position. Everything beyond that is
merely a matter of degree.

Leigh

* I thereby disprove by example your contention that a machine
can never be said to understand anything; I just said it!
Penrose is a very bright fellow, but he's out of his depth here.
I've heard of no one who has got nearly this far out without
drowning.