Brent Meeker wrote (through many posts):
> I won't insist, because you might be right, but I don't think that is
> proven. It may
> be that interaction with the environment is essential to continued
> consciousness.
Assuming comp, I think that this is a red herring. To make this clear I
use a notion of generalized brain in some longer version of the UDA.
See perhaps:
http://groups.google.com/group/everything-list/browse_frm/thread/
4c995dee307def3b/9f94f4d49cb2b9e6?
q=universal+dovetailer&rnum=1#9f94f4d49cb2b9e6
The generalized brain is by definition the portion of whatever you need
to turing-emulate to "experience nothing" or to survive in the relative
way addressed through comp. It can contain any part of the environment.
Note that in that case, assuming comp, such a part has to be assumed
turing-emulable, or comp is just false.
Of course, if the generalized brain is the entire multiverse, the
thought experiment with the doctor is harder to figure out, certainly.
But already at the seventh step of the 8-steps-version of the UDA, you
can understand that in front of the infinitely (even just potentially
from all actual views) running UD, comp makes all your continuations
UD-accessed. It would just mean, in that case, that there is a unique
winning program with respect of building you. I doubt that, but that is
not the point.
By the same token, it is also not difficult to get the "evolution of
brain" into the notion of generalized brain, so that "evolution" is
also a red herring when used as a critics of comp, despite the
possibility of non computationnal aspect of evolution like geographical
randomization à-la Washington/Moscow.
> I would bet on computationalism too. But I still think the conclusion
> that every
> physical process, even the null one, necessarily implements all
> possible
> consciousness is absurd.
OK, but the point is just that comp implies that physical processes
does not implement "per se" consciousness. They implements
consciousness only as far as making that consciousness able to manifest
itself relatively to its most probable computational history (among a
continuum).
>>
>> Reductio ad absurdum of what? Comp or (weak) Materialism?
>>
>> Bruno
>
> Dunno. A reductio doesn't tell you which premise is wrong.
Nice. So you seem to agree with the UDA+movie-graph argument, we have:
not comp v not physical-supervenience.
This is equivalent to both:
comp -> not physical supervenience, and
physical supervenience -> not comp
Now I agree that at this stage(after UDA) it would be natural to
abandon comp but then computer science and the translation of the UDA
in the language of a universal turing machine (sufficiently rich, or
lobian) such an abandonment could be premature (to say the least).
Incompleteness should make us skeptical in front of any intuitive and
too rapid conclusion.
> That's generally useful; but when we understand little about
> something, such as
> consciousness, we should be careful about assuming what's
> "theoretically possible";
> particularly when it seems to lead to absurdities.
Mmh.... If we assume "theoretical possibilities" and then are led to
absurdities, then we have learned something: evidences against the
theoretical assumptions. If the "absurdities" can be transform into
clear contradiction, perhaps by making the theoretical assumptions
clearer, then we have prove something: the falsity of the assumptions.
I think you know that, and you were just quick, isn't' it?
>
>> Stathis: In discussing Tim Maudlin's paper, Bruno has concluded
>> that either computationalism is false or the supervenience theory is
>> false.
>
> As I understand it Bruno would say that physics supervenes on number
> theory and
> consciousness supervenes on physics. So physics is eliminable.
Note that Maudlin's arrives at the same conclusion than me: NOT comp OR
NOT physical-supervenience. Mauldin's concludes then, assuming
sup-phys, that comp is problematic (although he realized that not-comp
is yet still more problematic). I conclude, just because I keep comp at
this stage, that sup-phys is false, and this makes primary matter
eliminable. Physics as a field is not eliminate of course, but is
eliminated as a fundamental field. It is not so astonishing given that
physics does not often seriously address the mind/body puzzle, and when
it does (cf Bunge) it still uses the aristotle means to put the problem
under the rug.
> That interpretation can be reduced to computation is implicit in
> computationalism.
> The question is what, if anything, is unique about those computations
> that execute
> interpretation.
Interpretation are done by interpreter, that is *universal* (turing)
machine.
Perhaps we should agree on a definition, at least for the 3-notions: a
3-interpretation can be encoded through a (in general infinite) trace
of a computation.
With the [Fi, ...] and Fu being an universal function, and thus an
interpreter, a 3-interpretation of the program i on data j will be
given by a infinite sequence 1Fu(<i,j>), 2Fu(<i,j>), 3Fu(<i,j>),
4Fu(<i,j>), 5Fu(<i,j>); 6Fu(<i,j>), 7Fu(<i,j>), 8Fu(<i,j>), ... The
number x in xFu(<i,j>) represents the number of steps of the current
computation, and <i,j> is the output of a computable bijection from N
to NXN.
We must be careful to distinguish this 3-notion of interpretation and
any notion of human conscious interpretation which is a first person
notion (see below).
>> Stathis: You're right that the correct mapping is the one in which
>> you and the bat share the
>> environment. That is what interaction with the environment does:
>> forces us to choose
>> one mapping out of all the possible ones, whether that involves
>> talking to another person
>> or using a computer. However, that doesn't mean I know everything
>> about bats if I know
>> everything about bat-computations. If it did, that would mean there
>> was no difference
>> between zombie bats and conscious bats, no difference between first
>> person knowledge
>> and third person or vicarious knowledge.
>>
>
> I don't find either of those conclusions absurd. Computationalism is
> generally
> thought to entail both of them. Bruno's theory that identifies
> knowledge with
> provability is the only form of computationalism that seems to allow
> the distinction
> in a fundamental way.
Careful: I don't identify "knowledge of p" with "provability of p". I
do identify "knowledge of p" with "provability of p AND p". Cf: I apply
Theaetetus' definition on "provability".
The confusion here is easy because, concerning a correct lobian
machine, knowledge of p and provability of p are equivalent (almost by
definition), but, the correct lobian machine itself cannot know that.
In case I am a correct lobian, only my own G* can prove that my
knowledge and my provability abilities are identical, but G* can prove
also that I cannot prove that, nor can I know that in any scientific
way. I can intuit that or bet on that, but no more.
> But my point is that this may come down to what we would mean by a
> computer being
> conscious. Bruno has an answer in terms of what the computer can
> prove. Jaynes (and
> probably John McCarthy) would say a computer is conscious if it
> creates a narrative
> of its experience which it can access as memory.
OK, modulo what I say just above. And given that "proving" entails the
existence of a narration I think I am close to Jaynes and McCarthy,
Except that Jaynes thinks about a third person narrative about a third
person self (like a memory can refer too by comp). I take into account
the incompleteness phenomena which forces us to distinguish the
many-person point of views.
The provability of p, written Bp for short, can correspond to Jaynes
"narration", but consciousness is nearer the notion of Bp & p or Bp &
Dp & p (Theaetetical knowledge).
The big nuance between Bp and Kp = Bp & p, is that the lobian machine
can prove that knowledge is incorrigible (Kp -> p), and cannot prove
that provability is incorrigible (Bp -> p).
Proof: suppose that for any p, the correct lobian machine can prove
that Bp -> p. Then she will prove Bf -> f. But, by elementary
propositional calculus Bf -> f is equivalent with NOT Bf. But "NOT Bf"
is equivalent "I don't prove the false" which is equivalent "I am
consistent". So the correct lobian machine would prove its own
consistency, which is impossible for any correct lobian machine by
Godel's second incompleteness theorem. OK?
Must go now ... (Comments on Peter and Russell will follow asap)
Bruno
http://iridia.ulb.ac.be/~marchal/
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to everything-list.domain.name.hidden
To unsubscribe from this group, send email to everything-list-unsubscribe.domain.name.hidden
For more options, visit this group at
http://groups.google.com/group/everything-list
-~----------~----~----~----~------~----~------~--~---
Received on Mon Sep 11 2006 - 09:46:32 PDT