Le 03-janv.-07, à 16:36, Stathis Papaioannou wrote (in more than one
posts) :
> Maudlin starts off with the assumption that a recording being
> conscious is obviously absurd, hence the need for the conscious
> machine to handle counterfactuals. If it were not for this assumption
> then there would not have been much point to the rest of the paper.
> Actually, Putnam and Chalmers also think that the idea of any physical
> system implementing any computation is absurd. I am not sure of
> Mallah's position (he seems to have disappeared from the list after I
> joined), but Hal Finney seemed to give some credence to the idea, and
> outside the list Hans Moravec and Greg Egan seem also to at least
> entertain the possibility that it is true. I would be interested if
> anyone is aware of any other references.
Hans Moravec has defended in this list indeed the idea that even a
teddy bear is conscious. You could perhaps search in the archive my
reply to him. I will try to sum up what I think about this, but other
things need to be clarified, perhaps.
> Although you have clearly stated that the two ideas - consciousness
> supervening on all physical processes and consciousness supervening on
> no physical process - are completely different I think they are
> related in that in both cases matter is irrelevant to consciousness,
> ...
If consciousness supervenes on all physical processes a case can be
made that matter could be relevant for consciousness. (I see Peter
Jones makes a similar remark).
> ... and we may as well say that what is important is computation as
> platonic object, not its accidental correlation with (putative) real
> world processes.
That's the point indeed. A consequence. Not something I assume. You put
it right below.
> They are also closely related in that the main argument against them
> is "that's absurd". A second argument against them is also the same:
> the difficulty explaining why we don't suddenly find ourselves in
> "white rabbit" universes.
My point was to show that comp transforms partially the "mind-body
relation" problem into an explanation of why such white rabbits are
rare. The non triviality of the mathematics of computers gives more
than a hint: in acomp (arithmetical translation of the UDA) we can
already show how a "quantum like quantization" occurs for the
observations having "probability one". I will insist on this later
perhaps. I must find a way to explain more the math without frightening
the inumerates :)
(Academically it has been the inverse: I had to motivate the mind-body
problem to mathematicians who are known to be generally allergic to
anything just looking like metaphysics. This has been far more
difficult!).
> Throwing out comp throws out physical supervenience as well, so it can
> eliminate the problem. Keeping comp and throwing out physical
> supervenience is the tricky thing.
I don't see why throwing comp throws out physical supervenience. On the
contrary, if my reasoning is correct, throwing comp is even the only
way to keep a physical supervenience, for example by asking for a non
computational physical process (like Penrose BTW but for very different
reasons).
Keeping comp we just have to throw out physical supervenience. It can
look tricky because, we are used to associate a particular mind/person
with (what we see as) a particular body. My opinion is that this is
just an aristotelian prejudice. Technically there is a priori no
problem (except the mathematical rarity of the white rabbit) given that
a computational supervenience thesis remain available. Consciousness is
associated with the infinity of equivalent (non distinguishable)
mathematical-computations. Time and space appears then to be internal
relatively stable constructions.
>
> I wonder if we are talking about the same thing, because it doesn't
> seem to me that an empirical proof is needed to show that any physical
> process could be seen as implementing any computation.
This is not obvious at all. You need both empiry + a physical theory
about that empiry, just to say that there is a physical process
implementing if only one computation (to say nothing about all
computations).
> For example, it is no problem if the physical process under
> consideration lacks a sufficient number of distinct states, because
> the one state can be reused to perform as many computations in
> parallel as you want, in the same way as a certain value in a given
> computer processor register will be the same in the course of
> countless possible computations.
But the state of the universal machine doing the computation will be
different. If the computer comes back to an earlier state it will loop,
and as such it will lose its universality. A state makes sense only
relatively to a universal machine ("material" or arithmetical say)
getting through that state.
> Of course this is completely useless as far as practical computation
> goes because to keep track of random 1's and 0's which, looked at the
> right way, are implementing a word processing program, you need to
> write the program and implement it on a conventional computer to keep
> track of everything. It is like saying that a page covered in ink
> contains any given text: true in a trivial sense, useless as far as
> helping you write a letter goes. However, what if computation can give
> rise to consciousness?
A computation is not just a collection of states. It is better
described as an ordering of discrete states relatively to a (universal)
machine computing producing those states. The "produce" here does not
to have to be a temporal or material processes, but can be described in
term of atemporal arithmetical relations. That is why we can expect non
trivial probabilities for going to one state to another.
> This would be like saying, what if a particular page of text were
> conscious? You would still need to remove all the unnecessary ink in
> order to read and perhaps interact with the text, but if you just left
> it alone it would still be conscious, dreaming away on the black page.
OK, but in that sense, assuming also a very speculative sort of
physical infinity, all computations will be there in a trivial sense:
all pieces of that paper contains not only UD* but the whole of
platonia (a non comp object). Without that infinity speculation I could
argue that most physical theories today will justify that among the UD
accessible states only a finite numbers of them will be accessed. That
is literaly nothing compared to what the whole of UD* needed for both
consciousness and then physical laws (assuming comp).
Perhaps you believe in physical real number (almost all real number
contains arithmetical truth coded in a non readable way). If physically
real numbers exist (why not, they don't really exist in loop gravity,
but they do exist in string theories), then any analog (non static)
machine embed arithmetical truth (much more than all computations) in
all of its activity. But now, it can be shown that such machine are not
turing universal: there are many problem that such machine cannot even
address. This is not an obvious result. Tarski showed it by proving
that the first order theories of the reals can be made complete and
decidable (and are thus logically poorer than Peano Arithmetic).
This can also be explained in term of Polynomial Equations with integer
coefficient. That is, equation like the following:
3xyzt + 24yx^3 -7x^45 - 4u - 12 = 0
If the variables or unknown x, y, z, t u belong to the reals, it can be
shown (by Sturm + Tarski) that there is an algorithm capable of finding
the solutions; but if you ask that the variables are natural number or
integers then there is no such algorithm. If this seems weird just
think about Fermat which is trivial on the reals and very difficult
with the natural numbers (but recently solved by Wiles & Co.).
If you want, the problem with the "black pages" is that there is no way
(or all ways!) to distinguish (internally or externally) the difference
between an "incarnation/implementation" of the number 1 and the number
1,
000000000000000000000000000000000000000000000000000000000000000000000000
0000000000 ... 000100000....
This changes once you add the sinus function. Trigonometric or periodic
functions are a way to reintroduce the natural numbers in the frame of
the reals (by sinus(2PIx) = 0 for example). But to show that the black
page implements the sinus in a genuine computational (even in not
external user friendly way) is again not obvious and presuppose many
physical things. We can come back on this later perhaps.
> One argument against this idea is that it is not enough to leave the
> computation (or the text) hidden in randomness, because it is only
> through interaction with the environment that consciousness manifests
> and only conventional brains and computers can interact with the
> environment. But this excludes the possibility that the conscious
> entity is dreaming or interacting with a virtual reality, isolated
> from the substrate of its implementation due to the impenetrable
> randomness.
OK.
[Stathis wrote also:]
> Peter Jones writes:
>
>> > SP: 'using the term "comp" as short for "computationalism" as
>> something
>> > picked up from Bruno. On the face of it, computationalism seems
>> quite
>> > sensible: the best theory of consciousness and the most promising
>> > candidate for producing artificial intelligence/consciousness'
>> What Bruno calls comp isn't standard computaitonalism, it has
>> an element of Platonism.
>
> Standard computationalism involves the idea that consciousness can be
> captured by a computer program running on a computer. Bruno keeps the
> first part but challenges the second, suggesting that the idea of the
> physical process in the computer actually causing the conscious
> experience is flawed, as per Maudlin's paper. Thus he does not begin
> with the idea that conscious computations exist as Platonic objects
> (although I think he did suspect this all along) but ends with it as a
> conclusion from examining the claims of standard computationalism.
> Always risky to summarise someone else's ideas when they're watching,
> but perhaps Bruno could comment if I have it wrong.
No you are right. By comp I really mean "standard computationalism". It
is equivalent with the "saying yes to the doctor who proposes some
artificial digital (generalized(*)) brain substitution for some level
of description". To make the "digital" condition transparent and
machine or formalism independent, I put Church thesis in "comp". Then
the arithmetical realism (platonism) is just the acceptance of the
excluded middle principle (A or ~A) for the arithmetical propositions A
(actually I need only the Sigma1 one, those with shape A = ExP(x) with
P decidable).
It is the platonism of the usual working classical mathematician.
So comp is really "standard computationalism", probably made a little
more precise than usual. The idea that Matter (primary matter) has no
role at all in science and consciousness is just an (admittedly non
obvious) consequence of standard comp (cf. UDA). Of course I did
suspect this all along. I am also glad with it in the sense than
physicist does not presuppose primary matter except in a meta-heuristic
way when they want to explain their subject matter without digging too
much in the theoretical interpretation problems. But most physicist
does use, without any problem of conscience, the excluded middle
principle on arithmetical propositions. In acomp (arithmetical version
of comp where it is explained "how" to derive the physical laws from
the number theoretical truth) I don't even need any "ontological"
platonism: I interview at the start machine who "believes" (= prove
here) the excluded middle principle, that is I am using a more
epistemological weaker form of "platonism". But for the UDA (which
explains *why* we have to derive the physics from the number) you have
to accept that for all natural number i and n, either Fi(n) is defined
or Fi(n) is not defined. The UD itself would have no sense in the
contrary case. Fi = the ith one-variable program in an universal
enumeration of all partial computable function (cf Church thesis).
(*) The concept of generalized brain (the portion of whatever is needed
to be Turing emulated so that my consciousness is locally recreated) is
necessary to prevent the Colin Hales (but see also Johnson Laird in the
literature) move consisting in saying that "the environment" plays a
role in my consciousness: if that is the case put the environment in
the brain (by definition). If the environment is both needed and not
turing emulable then comp is just *already* false.
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?hl=en
-~----------~----~----~----~------~----~------~--~---
Received on Thu Jan 04 2007 - 08:45:43 PST