RE: consciousness based on information or computation?

From: Marchal <marchal.domain.name.hidden>
Date: Tue Feb 2 09:39:14 1999

Hi Gilles and James,

I want comment your last dialog.

James wrote:
>>Schmidhuber suggests that there is just one program, as simple as
>> possible,
>>and the simples program generates all possible universes with their
>>complement of consciousness or lack of it. So he and I agree that the
>>factthat there is an awful lot of 'noise' does not matter.

Gilles answered :
>James, the problem is not the amount of noise. The problem is how to
>distinguish a noise from a signal. Once again it has no meaning without an
>interpreter.
>There is also an inconsistency with time. In Schmidhuber's view, time is
>given by the ordering of outputs. But as you judiciously remark ( and in
>Deutsch's point of view I suppose) time is an illusion. Macroscopic worlds
>contain some structure that make them apparently related to other states
>that we call past and future. But this information is stored in the state
>itself. So the string in which this state would appear and its place in
>this string is immaterial. The simplest program just enumerates all integer
>numbers. If you can tell me which integer describes the existence of
>consciousness or to the existence of anything like an electron or whatever
>you want, I would be very grateful to you!

My (Bruno) comment :

Let UD*, (you can do the same with Schmidhuber's great programmer, or Wei
 Dai's counting algorithm) be the complete description of the execution of
 the UD (universal dovetailer : the program which run all programs). If UD
 is written in a two dimensional programming language (like Conway's game
of life), then you can see UD* as an infinite 3-dimensional cone.
In particular, it is an infinite and purely static object.

1) James seems to me to be right, the 'noise' doesn't matter.
Nevertheless,
 this remain to be proved or at least to be made more precise.
2) You cannot associate mind with any program which are runned by the UD,
you can only associate mind with the whole UD*. To make any precise
prediction,
 you must "integrate" the whole set of all computationnal histories. This
is a computationnalist analog of Feynman path integral. Schmidhuber missed
 completely that fundamental point. This follow by the fact that delay of
 reconstitution doesn't change the quantification of computationnalist
 indeterminism.
3) With COMP, time is an illusion (I don't like that word, but I will use
it
 until I find the opportunity to add some nuances). But with COMP, space
is also an illusion, as matter and causality. In fact, all "physicalities"
 became internal mode of UD*, or even of just arithmetical truth. For
exemple, causality relations are reductible to infinite set of
 counterfactuals (which can be seen as coded in the UD*).

Sorry to be so short. Unfortunately I will be very buzy the next three
 weeks. I will continue the reading of the everything-list, but no more
participate. If I do, I
 will be very short. Nevertheless, about Schmidhuber, although there is
 some superficial resemblance with my thesis, I realise Schmidhuber has
an inconsistant u
Received on Tue Feb 02 1999 - 09:39:14 PST