On Thursday, April 9, 2015, Bruno Marchal <marc...@ulb.ac.be> wrote: > > On 07 Apr 2015, at 20:48, Stathis Papaioannou wrote: > > > > On Wednesday, April 8, 2015, Bruno Marchal <marc...@ulb.ac.be > <javascript:_e(%7B%7D,'cvml','marc...@ulb.ac.be');>> wrote: > >> >> On 07 Apr 2015, at 15:06, Stathis Papaioannou wrote: >> >> >> >> On Tuesday, 7 April 2015, Bruno Marchal <marc...@ulb.ac.be> wrote: >> >>> >>> On 06 Apr 2015, at 01:22, Stathis Papaioannou wrote: >>> >>> I agree that if comp is true then consciousness cannot supervene on >>> physical activity, for the reasons in the MGA thread. >>> >>> >>> OK. Thanks for making this clear. >>> >>> >>> >>> The only way out of this conclusion is to deny comp, which means to deny >>> CT. >>> >>> >>> I guess you are too quick. We can still deny comp, without denying CT, >>> for example by pretending that no copy will get the right behavior, or even >>> that the copy will be dead, and cannot be made moving at all, perhaps >>> because we believe in some magical God which would not allow it, or >>> whatever, or that all copies will be mentally impaired, etc. It is only in >>> the case where the copies behave the same as the original, and claim they >>> have no change in qualia, that comp is follows from CT with the >>> "no-partial-zombie" argument. >>> >>> This does not imply CT is false, as the "magical soul", or the >>> "primitive matter", or the "infinitely low subst level (actually >>> infinite)", used to make someone saying "no" to the doctor, might not add >>> any new computability power, only that it would be needed to remain alive >>> and have the relevant behavior. >>> >>> I guess you agree with this remark, as we were in the context of copies >>> having the right behavior and pretending to survive perfectly. Obviously, a >>> believer in CT, and not in comp, needs some amount of magic, and perhaps we >>> can derive comp from CT, if, like in the MGA, we can show that indeed we >>> need to add something magical. I have to think more on this, as I might be >>> quick again. >>> >>> Hmm... A model could be given with having an infinite low substitution >>> level. When using a digital brain, people would survive ... for some period >>> of time, and then problems would add up, due to truncation error, decimals >>> incorrect, etc. The brain would be a truly infinite machine, but without >>> giving the person new computability power. It seems to me right now. >>> >> >> What I intended by "CT" is the narrower physical version, which says that >> all physics is computable. >> >> >> OK. This clarifies your point. But the original CT has nothing to do with >> physics. Also Deutsch's form of CT (everything physical can be quantum >> Turing emulated (perhaps in polynomial time)) is not equivalent with the >> original CT, and might be in conflict with it. >> >> All physics might be computable, without the entire physical universe >> being computable (which I thing is "figital physics". >> >> With computationalism, a priori, the physical should not be computable, >> but it has to be enough computable to disallow too much white rabbits, >> something that QM seems to do remarkably well, but it is an open problem >> with arithmetic. The reason is that the indeterminacy on the computational >> histories might be too much big. >> > > At least the physics in the brain must be Turing emulable, or the whole > enterprise falls down. > > > I don't think so. The relevant part of the brain activity, relevant for > consciousness to be able to manifest itself, must be Turing emulable, but > the brain itself does not. > Computationalism presupposes only the existence of a level of description > such that we can truncate the (possibly not entirely computable) physical > description. >
That's true, but it can't be guaranteed that an artificial neuron will fire at the right time (to give a concrete example) if its behaviour depends on non-computable functions, such as true randomness or real numbers. If you don't get this right then the artificial brain won't work properly, and the recipient won't be able to walk, talk or think properly; maybe more like a movie zombie hat a philosophical zombie. I suspect such functions can be approximated and the neuron will function appropriately, but there is no guarantee. One good thing, however, is that the problem is an empirical one. If the C. elegant model in the OpenWorm project behaves just like a real worm, that will be evidence in favour of comp. > Worst: computationalism suggests that the physics (notably of the brain) > cannot be exactly Turing emulable. Indeed the physics will be the FPI > calculus on, somehow, the infinite unions of all finite pieces of > computations going through "my" relevant state at the substitution level. > Today, we have no reason to believe that this will be computable, and > worst, that too much white rabbits will not crop up. > > A brain, or any piece of matter cannot be entirely computable and still > have a statistical behavior which is computable, and indeed sometimes > Turing universal, like a brain, or a cell. But the "real" (with comp) > physics of some piece of matter can only be a map of the accessible worlds. > An electronic orbital is plausibly exactly that. If you look at an electron > in an orbital, corresponding to some energy level/eigenvalue, what you see > is the map of the set of computation/continuation in which your mind will > remain locally invariant when moving to the next worlds. As only the energy > level matters in your computation, it does not matter for you if the > electron is here or there, and your consciousness/first-person view is in > the two (arithmetical) relative reality at once. > > Physics is one of the way to consider the border between the computable > (sigma_1) and the non computable (The pi_1, sigma_2, pi_2, sigma_3, > pi_3, sigma_4, pi_4, sigma_5, pi_5, ...part of arithmetic)(*). > > A priori, the FPI confront the machine's 1p to the full complexity of > arithmetic. How to avoid oracles? There is no way, so physics results not > only from the competition between all universal machine, but also all > oracles. > > I do suspect "phase randomization" killing the oracle role, and I do > suspect the winner(s) exploit the random oracle, but strictly speaking it > is a complex open problem if physics is computable or not (assuming > computationalism). The prediction from the map is not computable, but > irrelevant for consciousness, and is just the non predictability of the W > or M in the WM-duplication. > But for the shape of the measure, the "orbital", with comp, it is unclear > if it is computable, and with QM, the partial trace are computable, and the > known hamiltonian are computable, but physics itself, well, we still haven' > it, as we can't unify all forces, and agree on the interpretation. > > > Bruno > > > (*) By a theorem of Kleene and Mostowski, the hierarchy above is captured > by the syntax: > > A machine is sigma_1 complete if it can prove all true arithmetical > formula having the shape ExP(x), with P recursive, mechanically decidable. > An oracle is pi_1 complete, if it can "decide" the truth of an > arithmetical AxP(x) > > An oracle is sigma_i complete if it can "decide" any question of the shape > ExAyEzAr ... P(x), with i alternating quantifiers, and > pi_i is for AxEyAz...P(x), again with i alternation. > > It is nice and not so difficult to figure out why that sigma_1 > completeness, is equivalent with Turing universality (in a sense made > precise by Davis 1956, but the community use a slightly different sense of > Davis 1957). > > You might take a look here if interested: > > http://en.wikipedia.org/wiki/Arithmetical_hierarchy > > Forget the zero in the names, and I see it fails to explain why we can > contract ExEy into Ez, which is based on the existence of computable > bijections, so that in computer science dimension is a quasi arbitrary > notion (which still plays some role tough). > Well, it fails to make the link with computer science. > > > > > > If that is true then at least the behaviour of a person should be >> computable, though he may be a zombie if in fact consciousness has nothing >> to do with physics but occurs in a separate spiritual realm. >> >> What is incompatible is the following three beliefs: >> >> (a) all physics is computable, and >> (b) consciouness supervenes on brain processes, and >> (c) consciousness is substrate-dependent and so will not be reproduced >> even with a sufficiently fine-grained and perfectly well behaved brain >> simulation. >> >> >> OK. >> >> I think that (a) and (b') are already incompatible: >> >> (a) all physics is computable, and >> (b') consciousness supervenes on *digital* brain processes. >> >> But (a), (b) and (c) already make obligatory to derive physics from the >> FPI on the whole UD*. The "winning" computation(s) are plausibly the one >> with a linear symmetrical bottom, and which admit long (deep in Bennett >> sense) computational histories. This makes us very rare in the arithmetical >> reality, but also super-multiplied, and with natural ways to entangle many >> universal machines in a many video-game type of (observable, >> phenomelogical) reality. >> >> I will surely come back on Church's thesis. It is a quite strong thesis >> which implies incompleteness in one simple (double) diagonalization. The >> original thesis asserts only that lambda-calculus defines all intuitively >> computable function. It is provably equivalent with the same thesis with >> lambda-calculus replaced by any know (Turing) universal system. Does this >> implies comp? I doubt (given the counter-example), but might be closer than >> I thought. >> >> And if CT implies comp, or almost comp, then consciousness would be close >> to being equivalent with the ability to get troubled by the following >> sentence, which typically can neither be true nor false: >> >> "Anyone currently reading the present sentence will never know that this >> present sentence is true" >> >> In arithmetic this sentences does *not* belongs to G* \ G, as it is not >> even expressible (arithmetical truth is not expressible or definable in >> arithmetic). There is no problem with the self-reference, by using the >> Kleene's method sum up by if D('x') gives T('x('x')'), then D('D') gives >> T('D'D''). >> Approximations of truth can be made, and the corresponding sentences, >> when they exist, can be true, or false, according to the approximations, >> but then those approximations disconnects truth from consciousness. >> Hmm... apology because I am thinking aloud. >> >> Bruno >> >> >> >> >> >> -- >> Stathis Papaioannou >> >> -- >> 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 everything-list+unsubscr...@googlegroups.com. >> To post to this group, send email to everything-list@googlegroups.com. >> Visit this group at http://groups.google.com/group/everything-list. >> For more options, visit https://groups.google.com/d/optout. >> >> >> 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 everything-list+unsubscr...@googlegroups.com. >> To post to this group, send email to everything-list@googlegroups.com. >> Visit this group at http://groups.google.com/group/everything-list. >> For more options, visit https://groups.google.com/d/optout. >> > > > -- > Stathis Papaioannou > > -- > 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 everything-list+unsubscr...@googlegroups.com > <javascript:_e(%7B%7D,'cvml','everything-list%2bunsubscr...@googlegroups.com');> > . > To post to this group, send email to everything-list@googlegroups.com > <javascript:_e(%7B%7D,'cvml','everything-list@googlegroups.com');>. > Visit this group at http://groups.google.com/group/everything-list. > For more options, visit https://groups.google.com/d/optout. > > > 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 everything-list+unsubscr...@googlegroups.com > <javascript:_e(%7B%7D,'cvml','everything-list%2bunsubscr...@googlegroups.com');> > . > To post to this group, send email to everything-list@googlegroups.com > <javascript:_e(%7B%7D,'cvml','everything-list@googlegroups.com');>. > Visit this group at http://groups.google.com/group/everything-list. > For more options, visit https://groups.google.com/d/optout. > -- Stathis Papaioannou -- 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 everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/d/optout.
