Re: consciousness based on information or computation?

From: <hal.domain.name.hidden>
Date: Fri, 15 Jan 1999 17:04:09 -0800

Wei Dai, <weidai.domain.name.hidden>, writes:
> I don't think your position is consistent. If you apply what you said "All
> that really matters is whether the program is instantiated at all." to the
> wavefunction as a whole, then either a program is instantiated in the
> wavefunction or not, and therefore a conscious experience must have the
> same measure as any other conscious experience.

I meant that in the context of a single universe. In any given universe,
all that matters is whether a program is instantiated at all. Instantiating
exactly the same program multiple times in that universe does not increase
its measure, or the subjective probability of its occurance.

I say this not as an assumption, but because of the thought experiment
I described. Two computers, running in lock-step, performing exactly the
same calculation at every instant, seem to me to be effectivelly the same
as a single computer. Make a computer with extra wide processing elements
and data paths, then divide them all down the middle by an insulator.
You have turned one computer into two. I don't see how that can be a
subjective change. Make the insulator a variable resistor, and we can
vary smoothly between one and two computers. But it does not seem that
we should be able to vary smoothly between one and two conscious entities.

Perhaps there is some other resolution to this puzzle that I have
missed. But this is the basis for my conclusion that multiple identical
instantiations of a consciousness do not add anything.

> If consciousness is based on computation, than it must be that some
> computations are instantiated "more often" than others. What "more often"
> means depends on what "instantiated" means, but so far we don't have a
> formal definition of "instantiation of a computation," which is why I
> currently lean towards the position that consciousness is based on
> information.

I would say, not "more often", but "with greater measure", leaving open
the possibility that the increased measure comes from some other effect
than being run more often.

> Let me ask my question in another way that avoids this issue for the
> moment. In the MWI, there is a branch of the wavefunction where a spin
> measuring device outputs the brain state (or consecutive brain states if
> you like) of someone who does not exist in that branch as a flesh and
> blood being or as any other kind of computation. Now the question is does
> this branch make any contribution to the measure of the conscious
> experience that someone has in that brain state?

I would suggest that the answer is no, or at least, possibly not.

Just creating a state or series of states is not enough, because it is
not possible, even in principle, to interact with them. They are static
and just sit there.

I would like to ground the question of whether a consciousness has been
instantiated by asking whether you could possibly interact with it.
If the answer is, yes, then I'd say there must be a consciousness
instantiated, else what are you interacting with? On this basis I
would point to active computation as being a necessary ingredient for
being *sure* that consciousness exists.

With an inert state, we can't be sure one way or the other. There is
no way to know, no way to interact, even in principle. So the question
can, perhaps, be ignored. It is similar to the ephemeral conscious
brains which flash into existence all around us as air molecules move into
configurations which happen to match some kind of conscious brain state
(at least some subset of the air molecules do). I think we should try
to ignore these, although I don't have a good enough understanding to
say exactly why not. This seems to be one of the biggest puzzles.

Hal
Received on Fri Jan 15 1999 - 17:13:13 PST