From: "Dan Minette"
> From: Robert Shaw
> > "Dan Minette" wrote
> > >
> > > How do these filters work? They cannot work algoritmically, that's
been
> > > proven rigorously.
> >
> > Has it? What is the proof?
> > All known physics is technically computable, so the brain can't
> > perform non-algorithmic processes without using unknown physics.
> >
> Right, that's Penrose's arguement for new stuff going on. If you apply
> quantum mechanics to the brain, you find that the state of the brain will
> quickly become indeterminate. Thus, the most precise scientific thing we
can
> say about human thought is that the precise results are indeterminate.
There isn't room in physics for new stuff on a scale that affects the
workings of the brain. Penrose appeals to nonlinearities in quantum
theory connected with gravity but there is no independant evidence
of such effects. Quantum gravity is even less likely to be relevant.
>
> We can, of course, give probabilities of this result or that result. But,
> we are very close to proving that it is theoretically impossible to
exactly
> predict future actions of humans by studying the present state of their
> brain and body and by controling their environment from time t to t+dt.
>
It's not possible in practice but proving it impossible in princple is
much harder. You need to discover new physics which has a significant
effect on processes at the energy levels and time scales found in the brain.
In known physics the movement of the human brain through its infinite
dimensional phase space can be precisely predicted by using the Hamiltonian
operater.
.> >
> > That's false. A brute force algorithm would work that way but the actual
> > algorithms used pare the tree. Some lines are explored 8 deep and others
> > only 3 deep.
> >
> That is true, but I specifically referred to the same line. Indeed, I can
> understand general strategy along a lone 20 or 30 deep, and yet miss a 1
> mover along that same line.
Those are two different lines which may diverge 5 or 10 moves deep.
I was using line to refer to a single sequence of moves. You appear
to be using it to refer to groups of such lines.
> I've done that a number of times. Karpov has
> missed a 2 mover. No chess program rated over 2000 does that, and he was
> rated 2700 at the time.
>
Chess programs don't use the same algorithms as humans. That doesn't
mean humans don't use algorithms.
> > > Humans can see a subtle 8 deep combo and miss a simple 1 mover at the
> same
> > > time.
> > >
> > That's because we don't consider all possible initial moves. We
recognise
> > patterns in the relative positions of the pieces. If a good move
doesn't
> >fit those patterns it doesn't get considered.
>
> But, that's not all we do. We generalize patterns in manners that do not
> seem to work algorithmically. We know that the stopping problem for a
> Turning engine cannot be solved in general. Yet, our minds makes
> generalizations akin to that.
Such as? And are they actually correct?
Algorithms with random elements can guess if turing machines will stop,
and they can be highly accurate, but not 100%.
>This is used as evidence that the workings of
> our mind are not reducible to physics...consistant with, mind you, but not
> reducable.
>
But your argument requires new physics, a claim that requires
much stronger evidence to support it.
--
Robert
