On 13 May 2014, at 04:26, Pierz wrote:
On Tuesday, May 13, 2014 3:39:21 AM UTC+10, Bruno Marchal wrote:
On 12 May 2014, at 03:01, Pierz wrote:
I've been following the "Is consciousness computable?" thread and
it occurs to me that there may be a contradiction in the UDA. Step
6 introduces the idea that we can teleport a "brain" (i.e.
digitally instantiate a set of memories, predispositions etc) into
a computed virtual environment.
OK. Precisely we bet that there is a description level of the brain
which will preserve its functioning, and manifest the person
consciousness with intact memories, etc.
Yet according to the final conclusion of the UDA, physics is
necessarily non-computable, because it arises from an infinity of
computations. If step 6 is to work, ISTM that physics has to be
computable.
That does not follow.
The virtual environment simulating Moscow and Washington can even be
very crude, so that you know in advance that you will be aware of
being in a video game. But once you bet on a description level, you
can bet you will survive in such environment, and discover that they
are fake, indeed, if you have enough time, and fair ways to explore
the environment. Step six is only about the invariance of the
(local) probability calculus for the real, and locally virtual
change, *for a second*.
It will not be enough that we approximate physics computationally,
because we can always imagine teleporting the brain from and into a
physics lab where advanced particle experiments are being carried
out. We can imagine here an arbitrarily advanced physics lab of the
future capable of carrying out the most advanced experiments that
are theoretically possible. The simulated lab must reproduce the
exact same results as the actual lab or the teleportation fails -
the "brain" can tell there's been a switch. If the conclusion of
UDA is non-computable physics, but the reasoning to reach that
conclusion depends on it, then clearly the argument is faulty. This
might even constitute a real argument for primitive matter (not
that I'm a fan of it), since primitive matter stops us proceeding
at step 7, thus saving us from the contradiction.
Bruno, how are you getting out of this one?
Well tried :)
It was a good try, huh! But yeah, I get it. The point is really that
it is per se possible for one's environment to be virtual rather
than physical. Damn! Another lance broken jousting your windmill... :)
But good point, UDA benefits of two readings. In the second reading
you can keep in mind that we reason assuming those normal histories,
in which the duplication experiments, and experiences, are done.
This does not solve the rabbit problem, but then that's why I
interview the machine on it.
The rabbit problem?
That's the name I used in "Concience & Mécanism" for the arithmetical
or digital measure problem. It is a common 'reality' inflation problem
in modal realist theories. Infinities crops everywhere. Also,
consistency is very cheap arithmetically, making *many dreams* locally
consistent, with a priori too much white rabbits, (an allusion to
Lewis Carroll). It might be related to Russell's Occam catastrophe in
a wider context.
Calicivirus is quite effective I believe. :) You're referring to the
measure problem?
Well, to be sure, the one on the sigma_1 sentences, or piece of
proofs. Yes.
How does the machine interview help?
By the second theorem of Gödel, and more generally Solovay theorem,
the Löbian machine knows at the start what she can and cannot
consistently assert, know and observe.
Machine are aware of their limitation, it is their most natural state.
Just that evolution has needed adults and local authoritative
arguments, but even that is changing and can last long.
Look, those machine have not (yet) been baptized by any church, nor
adopted by many sects, and by the virtue of their few and simple
initial beliefs (RA or PA axioms), they have less prejudices than most
humans.
Sure, they have their certificate of "arithmetical realism/platonism",
but here this mean only that they believe that (A v ~A) applies to any
arithmetical close sentence A. Soon enough the correct machine get
humble with respect to truth.
By UDA, physics is given by the measure one on the consistent
extensions, (a self-referential notion that the machine can talk
about), which are computationally accessible, and this is translated
in arithmetic, by the logic of []p & <>t (and other variants) on the
sigma_1 propositions.
Starting from G, we get a knower, the modal logic S4Grz of the logic
of []p & p. And we get logic of alternative, with []p & <>p, but on
the sigma_1 propositions, we get already a logic of alternative, and
proximity, on S4Grz.
More on this in the math thread, I guess.
Bruno
http://iridia.ulb.ac.be/~marchal/
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.
