amoeba, decision and consciousness

From: Marchal <marchal.domain.name.hidden>
Date: Fri Jan 15 01:10:07 1999

Wei Dai asked


>Is consciousness based on information or computation? Let me give an
>example to explain what I mean. Suppose there is a computer running an AI
>program. Assuming computationalism, this computation should contribute to
>the measure on conscious experiences. Now suppose there are two computers
>independently running two identical AI programs with the same inputs. This
>should make twice as much contribution to the consciousness measure.
>
>But now suppose there is one computer running an AI program, and a second
>computer that makes a copy of the first computer's state after every
>operation. My question is how much contribution to the consciousness
>measure does this setup make, compared to the single computer setup? If
>consciousness is based on computation, then it makes the same
>contribution, since there is just one instance of the AI computation being
>run. But if consciousness is based on information, then it makes twice as
>much contribution, since there are two copies of the AI's state at any
>given time.
>
>So which is it?

I am not sure I understand why you choose the words computation and
information, but let use these words. Neither don't Ifeel easy with the
expression "measure of consciousness" : I understand it as a question of
measuring conditionnal probabilities. Let me put it in the following way :
You are computationnalist so you are willing to use (classical)
teletransportation. So your instantaneous description is read at
Amsterdam (let us say) and then you are annihilated (100% of success,
this is not possible with MWI, let us suppose it is possible for the sake
of the argument). Then you are build following the description at
Washington in one exemplar, and you are build in Moscow in 9 exemplars.
What is your expectation, when you are at Amsterdam of feeling yourself
appearing in Washington after the experiment ? I suppose the 9 exemplars
in Moscow are build in identical but separate closed rooms.
Would you say there is something like 1/10 of appearing in Washington and
9/10 appearing in Moscow ?
For computationnalism making sense, I believe there must be a difference.
If there is no difference there is a risk of collapsing all
computationnal state, at least from a first person point of view.
I think that this question is fundamental both for the computationnalist
and any many-worlder. Nevertheless it is not easy to formalize such
question. I think it is an important open problem.

Gilles wrote

>I personnaly deny the concept of "measure of conscious experiment". I deny
>the fact that consciousness is an *objective* property of matter, just
>because you can not define a physical property, measurable by an external
>apparatus, whose measure could determine the degree of consciousness (or if
>you can, let me know).

OK, but just replace "measure of conscious experiment" by "probability of
having a personal experience". See above.

>Much of the discussion about consciouness is plagued by this fact, because
>we include it in formalisms that have not been deviced to handle it. The
>only consciousness we know is our own one, by means that are different from
>those we use to interact with the outer world. We think that the other
>human beings are conscious because of the similarity of their behavior with
>ours, but it does not define what is consciousness. If one succeeds in
>building a computer with a human-like behaviour (which is quite possible in
>my sense), deciding if it is actually conscious or not is purely a matter
>of convenience, not an intrinsic property. In other words, I think the
>proposition "Another than me is conscious" is really unprovable.

That is a very important point. Note that the proposition "I am
conscious" is also unprovable. I am used to define quasi-axiomatically
consciousness by this very property. To be conscious = to know something
and to feel that it is not possible to prove it. This make it possible to
find objective root for subjectivity. The existence of unprovable (by a
machine x) but true (about machine x) proposition encourage to use logic
for the search of objective foundation of subjectivity.

James wrote

>So it is MEANINGLESS to think of 'choice' or 'determinism'. Two meaningless
>words can be as compatible as you like.

I think it is a question of level. At the arithmetical level or if you
prefer at the level of the wave function of the universe 'choice' has no
meaning. But at that level things seems to me to be deterministic
(Schroedinger equation is determinist). But at the personal higher order
level, choice has a meaning. The meaning is linked to the fact that I am
ignorant -for all practical purpose- of the low level, AND I can reflect
that practical ignorance and feel obliged to make choice.
If not, any lawyer could defend anyone very easily : O member of the
jury, please, indulge poor Bill Clinton, after all, all he has done is
just obeying physical deterministic equation ! And of course the jury can
propose the death penalty to poor Bill, justifying themselves by invoking
the same equation.
Well, by saying that 'choice' has no meaning, you take the risk of
denying the meaning of any high level psychological term, and to produce
something like an ethical collapse.
 
James wrote also

>In the MWI view, we have a 'block universe' made up of every possible
>universe and time is just a relationship between them. To speak of anything
>'happening' - let alone a choice - is meaningless when you take this
>'Archimedian' perspective in 'nowhen' (The time issue is mastered by Huw
>Price in 'Time's Arrow and Archimedes's Point' (1997 I think) - absolutely
>required reading.). Determinism, free will - all meaningless.

Right, but WE are NOT living at the Archimedian level. We do have free
wil, just because we cannot determine ourself completely, and we can
reflect that.
And free-wil could be an objective property of any sufficiently rich
machines embedded in that 'block universe'.

> There is NO reason to think that the arrow of time is an
>objective feature of reality.

Same remark. I agree with that (as a matter of fact I do think there is no
reason to think that any physical stuff has an objective feature in
reality).
BUT here too, I do think that there could be an objective feature to the
existence of the feeling of the arrow of time for a machine (or abserver)
which can only postulate the existence of the Archimedian level.
The intrinsical limitation of the observer can be founded on objective
law linking the observer and the archimedian level. Exactly like Everett
when he gives objective reasons (schroedinger equation) for subjective
probabilities (personal histories of automaton).

Bruno.
Received on Fri Jan 15 1999 - 01:10:07 PST