Le 24-juil.-06, à 04:23, David Nyman a écrit :
>> Bruno: And this is perhaps the very root of a possible disagreement.
>> I would
>> not compare "mathematical" with "tautological", nor with
>> "conventional". This should be clear after the Godelian fall of
>> logicism. We know today that even just the "arithmetical" realm
>> (studied by number theorist) is not compressible into a unique theory.
>> Actually a complete theory of the number realm has to be infinite, and
>> even of a type of infinity not nameable in any effective way. Math is
>> as full of surprises (and even of contingent facts when seen from
>> inside) than physics is. As David explained in FOR, mathematical
>> reality quick back. Still I appreciate your quote in 'physical", but
>> the whole ambiguity (relatively to the comp emulation of the mind)
>> will
>> rest in what you mean exactly by those "quote" and what you mean by
>> the
>> difference you are making between "efficacious" and "descriptive".
>
> David: Bruno, your observations go to the heart of the matter! I
> think I
> could be clearer on some of these points - let me try, but forgive me
> if I don't adopt precisely the order of your comments. First, you are
> correct that my starting point is in assigning primacy to what
> (conventionally or otherwise) we think of as 'physical'. I don't
> intend this in any 'mystical' sense, but rather as a hopefully robust
> epistemological point of departure.
It seem a little bit ontological for me.
> So, I am claiming that the
> structures and relationships elucidated within this domain
But "this" domain is full of obscurity. Now, with the QM and the MW,
some light appears at the horizon, but even just among physicist
sincerely interested in conceptual issues, it is hard to say there is a
consensus.
> are what is
> efficacious in producing both 'objective' and 'subjective' phenomena -
> a distinction, I suggest, that is dependent on 'point of view'.
Well, I do agree with you, here. But I still find your notion of
"efficacious" unclear. Linking it with a putative 'physical' thing
makes things harder imo.
>
> Consequently, I also claim that the type of emulation, and associated
> isomorphism, necessary to be causally efficacious either 'physically'
> or 'experientially' must be 'physically constructable'.
>
> A Turing emulation, despite the fact that it runs on a 'physical' TM,
> does not 'physically construct' the subject of the emulation in this
> sense, because the necessary isomorphisms exist merely descriptively.
> I meant that the mathematical form of this description is in this
> sense 'tautological'. However, I wouldn't press this point. The
> crucial issue is that the description is quite different in form from
> the thing described. And the description has, if any, a different
> causal efficacy than the thing described.
In the case of a simulation, I would agree with (quasi by definition),
but the fact is that computer science provides that non trivial notion
of emulation.
To be sure I don't really believe in any basic notion of causality
other than IF p divides aq, and if p and q are primes then p divides a.
Things like that.
>
> Of course, we as observers perceive the isomorphisms as metaphors, for
> example when we read the listing of the relevant programs, but what is
> 'constructed' on the physical substrate of the run-time system is a
> syntactical narration of the program, not a physical isomorph of the
> subject of the narrative. This is not dissimilar to what occurs when
> a story is 'instantiated' in a storybook. The 'emulation' at the
> level of the pages of the book is decriptive and syntactical. When we
> read the narrative, we physically/ experientially construct an
> isomorph of the subject, and so 'bring it to life'. Unless a TM/
> digital computer possesses the same physically constructive
> capability, it cannot achieve this.
I don't pretend you are wrong. But I think that if you are right here,
you will need to assume actual uncomputable operations to actualize the
mind in your physical constructions (not to mention making that mind
unique).
>
>> Of course you may be true, but from what you say here, it could even
>> more speculative in the sense that with such a proviso, after you are
>> digitally duplicated, you could tell me that the copy is a zombie. You
>> will agree the doppelganger looks like you and behave like you because
>> the "descriptive (intensive)" sense has been preserved, but it would
>> be
>> a zombie because the copying was not "efficacious" in the "(extended)"
>> sense. This means, assuming comp, that your doppelganger will have
>> some
>> hard time to be respected as person ...
>
> This depends on whether a 'description' of a person will turn out to
> be sufficient to completely emulate the behaviour of a 'real' person,
> let alone 'be' one. This IMO is still an open empirical question.
I agree, but probably for a different reason. I think that IF we are
capable of surviving a copying process (including annihilation of the
"original") from a reconstitution made itself at some correct level of
substitution, then if we observe ourself or our neighborhood close
enough (meaning at a level equal or least than our substitution level)
then we must see the trace of the many computations. In *that* sense,
modern or pure QM (= QM without collapse) already confirms that comp
prediction. Would we have strong argument that physical reality is
unique, then we would have empirical support for a primitive actually
infinite and non-Turing-emulable matter playing a role in our
brain/mind processes.
> It
> is felt by some to entail answering the riddle of whether
> 'consciousness' is causally efficacious or in some sense a disposable
> 'epiphenomenon'. For what it's worth, my own view is that this is a
> category error and that the necessary move is to see that reality and
> unmediated self-intimacy are equivalent.
That's not so clear...
> The alternative is an
> infinite regress of 'observers'.
Most infinite regress have fixed point in computer science.
> Consequently, a 'zombie', in the
> sense of a structural isomorph that somehow lacks an experiential
> field, is a bad case of the grin without the cat. If he exists
> physically, then IMO he is by this token self intimate.
Should I understand you would accept that a robot or a 100% cyborg
would be conscious with a physical body, but not with a virtual body
implemented with a "real concrete" (which makes sense with your
"primitive matter" hypothesis) computer?
Or just (what is more common) that consciousness will be there in each
case where the simulation are eventually grounded in some physical
universe (as opposed to a purely arithmetical reality)?
In both case that would violate the consequences of the comp hyp. The
first case is easy to show, the second case is less easy to show ... (I
mean I don't pretend all this is obvious).
> However, the
> content of his self intimacy - his experiential field - depends
> critically on his 'physical construction', not merely his ability to
> embody a Turing emulation.
>
> Notwithstanding this, it may turn out that Turing's vision of a
> behaviourally indistinguishable intelligent being is possible, without
> the accompanying experiential constructs (or at least, with different
> ones). Or it may not. In the case that it is, the question then
> arises: are either we, or our emulated doppelgangers, intelligent
> enough to deduce from an examination of our internal representational
> specifications which type of entity we are?
It is bit like asking a doctor for knowing if you are in pain, it seems
to me.
> If so, then our
> doppelganger would realise that although he could get 'd-wet' from a
> 'd-typhoon', he wouldn't need to go to the expense of buying a real
> umbrella.
>
>> Somehow your argument is similar (corect me if not) to the argument of
>> those that a computer cannot really simulate a typhoon because you
>> will
>> never be wet due to a computer simulation of a typhoon. But the comp
>> point is that you need to feel wet by a virtual typhoon only when you
>> are emulated yourself in front of the typhoon. Again, this can be
>> shown
>> not making you feeling to be wet, only if you have not been simulated
>> at the right level or if that level does not exist, and it is up to
>> you
>> to explain us what specifically there is not turing emulable.
>
> Have I done this? As I suggest above, I agree that a Turing emulation
> could embody a narrative at any arbitrary level of detail, but I deny
> that it is constituted so as to construct the content or subjects of
> that narrative. So the narrative may contain the element that 'I
> believe there is a typhoon and that I am wet', but this does not
> entail that either the typhoon or my structural/ experiential isomorph
> is thereby constructed and in fact feels, and gets, wet. The 'right
> level' - and this is the critical difference I am claiming in our
> views - is a level unattainable to comp.
All right. That could exist if matter matters indeed. I think only the
relations are needed.
> It is a level attainable
> only by construction.
I think you will have an hard time defining "construction" from your
"matter".
> It may be that there is a belief that comp
> simply IS construction because of an underlying assumption that the
> 'physical' is a computation and that its structural relationships are
> fundamentally computationally/ syntactically established.
Not necessarily. If comp implied physics, comp will implied the
existence of internal notion of construction. I do believe it is
actually already the case.
> But even in
> this case there would still exist IMO a real distinction between
> 'comp' (within comp reality) and 'construction' (within comp reality).
Ah! But OK then.
>
>> Sure. That is why I propose in some thought experiments to consider a
>> "generalized brain", so as to put any part of the (multi)verse
>> necessary for the emulation of your consciousness. The question
>> remains
>> for this generalized brain: is it Turing emulable. If yes, you will be
>> driven the "immaterialist" consequence of comp, imo. If not, I will
>> ask
>> what is the precise non-comp processes you think about, and how will
>> you ascribe consciousness to it.
>
> As I have remarked above, I prefer to think in terms of reality as
> entailing unmediated self-intimacy. The question of 'consciousness'
> then collapses into how reality self-structures its physical/
> experiential fields.
All right.
> A TM engaged in Turing emulation is of course a
> physical/ experiential structure by this token, but the crucial
> question is whether what it constructs has equivalent causal efficacy
> to the subject of emulation. I say that it cannot have this unless it
> constructs equivalent physical/ experiential isomorphisms, which in
> virtue of its syntactical mode of operation, it does not. So the
> 'non-comp' processes I have in mind are physically constructive ones,
> not merely descriptive or narrative.
> This is analogous to pointing
> out the difference betweeen describing in complete detail the building
> of a house, and actually building one.
... where I think this is relative.
>
>> But if the digital transformation does preserve the relevant morphism
>> it will either mean that the level of digital substitution has been
>> chosen incorrectly, or that such a level does not exist, in which case
>> computationalism is just false.
>
> Well, it may be that computationalism in the context we are discussing
> is just false. IMO, the rock upon which it may founder is the issue of
> construction versus emulation. Computational emulation is successful
> if the behaviour of the emulated system (transformation of inputs to
> outputs) is invariant in the face of differences in the internal
> processing that mediates them. But if you hold, as I do, that the
> content of reality is determined by the structure of the physical/
> experiential field, then the form of the processing is critical to the
> experience, even if the input/ output relationships are preserved. My
> claim is that computational isomorphisms are thus merely
> transactional, but that the type of isomorphism required for
> experiential equivalence is structural.
Mmmmmmhhh.... I think you have a good intuition but that you are dead
wrong when you want attach physical and experiential in one field. I
can understand the willingness to proceed in that way, but you will
have to introduce actual infinities. With comp you do inherit of actual
infinities (indeed), and actual indeterminacies, but only from some
first person point of views. And then you don't need to introduce them.
"First person" is what Everett called the "subjective".
>
>> ... only if the "generalized brain" is not turing emulable. With both
>> comp and the quantum, only the "geographical" or the contingent is not
>> computable * from a first person point of view*, like observing a
>> electron spin up in the z direction after having prepared it in the up
>> + down state, or finding oneself in Washington after a
>> Washington/Moscow self-duplication.
>
> Again, I think you are relying on a transactional level of equivalence
> here if you claim that the Turing emulation of 'reality/ experience'
> effectively constructs it. I disagree with this view. I think that
> what you are calling here a 'first person point of view' is rather
> more accurately 'a description of a first person point of view'.
Ah, now I must "cc" this to the everythinglist where, if you are
interested, you can follow sequences of attempts to explain this in
more details.
But the beauty here is that computer science can explain why, although
the first person admits some description, such a description is
inaccessible from the person itself and from others person of similar
complexity (alas some technics are needed to believe this weird fact).
Comp makes any first person as unnameable as the big one itself ...
> My
> claim is that an embedded first person perspective is possibly
> re-constructable, but not computable.
Believe it or not, but this is a consequence of comp. It follows from
some incompleteness-like results that accepting comp entails accepting
many non computable things. Incompleteness itself is a consequence of
the law of Post (Church thesis).
> This may well be a semantic/
> syntactic distinction, in the case that 'semantic' implies the
> construction of the full blown structural network of physical/
> experiential relationships, and syntactic merely a specification of
> the same.
I can accept this.
> Of course, I agree in either case that the contingent or
> indexical are neither computable nor constructable.
>
>> You should elaborate on this, but again your argument will need an
>> actual non-computable invariant (as opposed to uncomputable
>> contingencies) to make it "critically different" with respect to
>> turing
>> emulability.
>
> My non-computable invariant is that of structural isomorphism within
> the physical/ experiential field. This is non-computable because it
> is not preserved within a purely transactional emulation. It is
> however re-constructable (just as we can re-construct a house from the
> plans, even though we can't live in the plans).
I think you miss the incompleteness phenomena, which somehow makes us
more humble (especially with comp) with respect to what is possible or
not with machines. I agree with a lot of what you say at some perhaps
deep level, but I don't see any reason why the comp hyp could
jeopardize your intuition. On the contrary, it could make it more
precise.
>
>> the whole ambiguity (relatively to the comp emulation of the mind)
>> will
>> rest in what you mean exactly by those "quote" and what you mean by
>> the
>> difference you are making between "efficacious" and "descriptive".
>
> Is this any clearer now? (as mud, probably!) It rests on the
> fundamental distinction I have attempted to draw between
> 'construction' and 'emulation'.
This is akin to the important distinction between first and third
person discourse (and the content of these discourses). If that is the
case I would say comp is on your side. The price seems to be: many
many many many differentiating realities in all directions. Do you
think *this* should be wrong?
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 Jul 24 2006 - 11:07:48 PDT