From: Hal Finney (hal@finney.org)
Date: Wed Jun 12 2002 - 11:41:16 MDT
Lee Corbin writes:
>suppose that the re-enactment of the original torture
> session is *almost* identical, except that one atom in one of
> the eyelashes of the little girl is absent.
John Clark replies:
> And if I said it still made no difference you'd ask about 2 atoms then 3, 4,
> 5,6...
> In general how many trivial changes does it take to make a profound change
> in something? I know a 70 pound man is thin and a 700 pound man is fat but I
> can't tell you exactly where the dividing line is.
I don't think Lee's question can be dismissed as just another Sorites
paradox. In the case of thin vs fat, these are fundamentally a matter
of definitions. We may choose to draw the line anywhere.
But in the case of conscious observers, we believe it is a matter of
fact whether there are one or two observers. It's not something we can
decide arbitrarily. We may not know how many there are, but in some
sense we believe that there is a fact of the matter. (Of course, not
everyone believes this, but I think those of us in this discussion do.)
Suppose we accept, with John, that two *identical* simulations represent
exactly one conscious observer. And presumably we all accept that
two different simulations represent exactly two conscious observers.
Then we have a problem if we can move smoothly between two identical
simulations and two different simulations.
Of course, technically we can't move smoothly in this way. Identity is
identity, and there's a big difference between "the same" and "almost
the same". In Lee's example, changing 1 simulated atom means that the
sims are no longer the same. On the other hand, it seems that this
small change might not affect the course of the simulated brain at all.
So is it the same, or isn't it?
I'd suggest two points. First, a change doesn't count if it doesn't
affect the brain simulation at all. With digital computers we have an
easier time of judging this than with physical systems. We can tell
whether two simulations had the same numeric values for all the simulated
neurons (or equivalent). Maybe changing one atom in an eyelash would
lead to no change whatsoever in the neural behavior.
Second, it seems likely to me that brain behavior is nonlinear and
chaotic. And maybe we could even stipulate that this is an inherent
property of consciousness as we know it. Hence even a tiny change to
a brain parameter in the sim would soon be amplified (by the "butterfly
effect") into a macroscopic change where everyone would agree that there
are two people involved.
So in principle we can solve the paradox by saying that any change to the
neural state, even one which is so small that it is presumably initially
imperceptible, does in fact cause two people to begin to exist. There are
two different simulations being run at that point, with different numeric
values. Chances are the differences will spread and grow. There is no
way, just from running one sim, that we can know how the changes will
occur in the other. So the information is no longer redundant between
the two sims.
For all of these reasons I would propose that even a minute change, as
long as it affects the numeric values relevant to the neural network, does
cause there to be two people. Even if the change is not big enough to
be noticed by the simulated people initially, the information processing
is different and we have to believe that there are two people present.
Hal
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:45 MST