RE: consciousness based on information or computation?

From: Gilles HENRI <Gilles.Henri.domain.name.hidden>
Date: Tue, 2 Feb 1999 13:47:50 +0100

À (At) 14:43 +0000 1/02/99, Higgo James écrivait (wrote) :
> There is no difference between noise and signal except in the eye of
>the beholder. No integer describes anything except in the eye of the
>beholder.

This is not very satisfactory...Does it mean that consciousness would exist
with any permutation of particles, for example in the first moments after
the Big Bang? (see below my remarks on Hal's reply)

 I am not sure what you mean "The simplest program just enumerates
>all integer numbers."
>
In Schmidshuber theory, a state of the Universe is a finite substring
(between two commas), that is an integer. As the information on past is
entirely stored in each present state, i.e. in each substring, it is
independant on the rest of the string. A UD is of no use (I think the
problem comes from the confusion between the time of the "Big Computer"
that generates the bits and the time associated with a state of the
Universe, which has nothing to do with the former one. Of course with REAL
computer both times coincide!). To generate all possible substrings, you
only have to enumerate all integers. "0,1,10,11..." What I think, however,
is that a simple integer can not represent alone a Universe:

>James Higgo, <james.higgo.domain.name.hidden>, writes:
>> There is no difference between noise and signal except in the eye of
>> the beholder. No integer describes anything except in the eye of the
>> beholder.
>
>I would say that there is a sense in which an integer describes something
>in an objective sense, rather than just having it be subjective and in the
>eye of the beholder.
>
>Specifically, if the integer is very large, trillions of trillions of
>trillions of bits in size, and if it can be compressed down to a much
>smaller value via computational techniques, then that compression is an
>objective way of summarizing the information in the integer.
>
>Suppose that the integer can be best compressed by treating it as the
>output of a program which implements some kinds of "laws of physics",
>starting from some specified initial condition. This integer could then
>be considered as the result of actually running a universe simulation.
>This would be an objective interpretation, not dependent on a particular
>beholder.
>
>Looking at this objective interpretation of the integer as the history of
>a universe, we might then be able to see space and time dimensions,
>an analog to entropy increase and an arrow of time, causation, evolution,
>and information flow. We might even be able to identify structures which
>process information and which behave in the same manner as intelligent,
>conscious minds like our own. In that case we would have reason to say
>that, in an objective sense, this integer represents a universe which
>has conscious minds.
>
>Even our own universe can be considered as a four-dimensional space-time
>structure, which is in effect a static block of information. It is not
>known at this point whether the universe is fundamentally discrete or
>continuous, but certainly we cannot rule out the possibility that it is
>discrete at the lowest level. In that case our own universe can be
>fully represented as an integer.
>
>Hal

It is almost true, but only almost. The Universe could be discrete. Assume
you have discrete cells at the Planck scale, 10^{-33} cm. You would like to
describe at least the visible Universe, about 10^{30} cm. If theory of
inflation is correct, it could be 10^{50} times as large, so 10^80 cm So
you need at least about N=10^400 cells for a 4-D space time. Well there is
no limitation of available computing time. The problem is not here. The
problem is that if the principle of equivalence is true, any cell is in
principle equivalent to another one. You have no natural ordering. So if
you want to describe the Universe by a N bits string, you have to assign
arbitrarily a cell to some rank in the string. But you can do also any
permutation in this assignement. It means that each of n1!*(N-n1)!
different strings with n1 "1" and (N-n1) "0" are equivalent. Basically you
are left with N different states instead of 2^N as expected, because of the
high degeneracy introduced by the freedom of the assignement. This is
because you have no relationships between the bits that reproduce the
topological structure of the Universe and its physical laws. These are not
specified by a string (or equivalently an integer) alone, as long as it can
be. So I don't see how the compressibility of integers (which is by the way
generally unprovable) could be linked with the structure of space-time,
since there is no one-to-one correspondance between them.

You can specify such a relationship, but it means that the state of the
Universe is not fully represented by the string but by the string PLUS the
mapping. The physics is not only in the string, but in the formal
relationships between the bits. But it is not conceptually simpler that
ordinary physics, because it doesn't say anything on the origin of the
relationships. You have only done a discretization of space-time, by the
way an interesting idea whose consequences would deserve discussion.
Computation, as meant as a succession of string, doesn't mean anything
since the information of time enclosed in each substring has nothing to do
with the other strings (as said above). As I said before, it is somewhat
curious to explain a fundamental reality by the help of an apparatus that
exists only within this reality.

Gilles
Received on Tue Feb 02 1999 - 04:50:28 PST