Re: On the computability of consciousness
On 20 March 2010 18:22, Bruno Marchal wrote: > Well, if by 3-p Chalmers you mean some 'body', such a body *is* a zombie. > The 1-p Chalmers is Chalmers, the person. Its body does not think, but makes > higher the probability that the 1-p thoughts refers to the most probable > computations you are sharing with him. Well, if its body does not think (which of course Chalmers assumes that it does, even though he says from his epiphenomenalist-dualist standpoint that this does not logically entail consciousness), just how does it increase the probabilities in the way you say above? IOW, what is the systematic correlation supposed to be between the "physical events" in its brain and the 1-p thoughts of 1-p Chalmers? This, after all, is a major aspect of the mind<-->body problem, and it's one thing for the explanation to be counter-intuitive, but right now I'm not sure I could claim any firm intuition about it at all. Let me try to tease this out, giving your words, as you say, their most favourable interpretation - for me, that is. By "the computations you are sharing with him" I assume you to refer to the 1-p-plural computations (as you reserve 3-p for the arithmetical reality). IOW the UD generates (amongst everything else) the 1-p-plural appearances that constitute all possible perceptions of our "shared environment" in all its possible extensions. Included in these, of course, are our bodies, and we expect - pace white rabbits - our bodily activities (including, naturally, our brains) to be consistent with our thoughts and with the behaviour of the rest of the environment. Nonetheless, presumably it is the case that there are "white rabbit" extensions in which my response to the pain of being burned is to do something pathological such as thrust my hand further into the flame. But the effect in experience of even a high measure of divergent pathological extensions is hypothesised as being damped by the convergence of "normal" behavioural extensions (maybe corresponding to some version of the least action principle, a la Feynman - and perhaps illuminating also the unreasonable a posteriori effectiveness of Occam). So the effect is to make it very much more likely that my actions will be consistent with my thoughts, including the actions of my brain. Does this mean that there may be "white rabbit" extensions in which the behaviour of my brain is grossly inconsistent with my thoughts? I suppose so. In this view, the concept of causation, if it is valid at all, must be reserved for the 3-p arithmetical operators - for the internal computational relations themselves. The higher-order relations between computations are rather correlative, and the appearance of causation in the correlative domain that we inhabit is that of consistency with expectation - normal, or non-pathological behaviour, IOW. So it isn't a case of the brain causing thoughts, or thoughts causing the brain, but rather a question of which thoughts emerge as being consistent with which brains. The remarkable thing then would be that we seem to find ourselves only in situations where most (perhaps all) brains are consistent with most (perhaps all) thoughts. The dreams of the machines, finally, seem to have converged on shared "physical" universes of staggering complexity and consistency. Is this anything like what you were trying to convey (interpreted favourably, of course)? David > > On 20 Mar 2010, at 16:56, David Nyman wrote: > >> On 24 February 2010 17:57, Bruno Marchal wrote: >> >>> Please, keep in mind I may miss your point, even if I prefer to say that >>> you >>> are missing something, for being shorter and keeping to the point. You >>> really put your finger right on the hardest part of the mind-body >>> problem. >> >> Bruno, I've been continuing to think, and meditate, about our recent >> discussions, and have been re-reading (insofar as I can follow it) >> your on-line paper "Computation, Consciousness and the Quantum". I >> feel I have more of a sense of how the aspects I've been questioning >> you about fit together in the comp view, but if I may, I would like to >> press you on a couple of points. >> >> My original post on "non-computability" was motivated by re-reading >> Chalmers and struggling again with his assertion that a zombie (e.g. >> including the "3-p Chalmers" that wrote The Conscious Mind!) could >> nonetheless refer to "consciousness" and hence be behaviourally >> indistinguishable from a conscious entity. > > So you talk here on the philosophical zombie which is counterfactually > correct. > Well, if by 3-p Chalmers you mean some 'body', such a body *is* a zombie. > The 1-p Chalmers is Chalmers, the person. Its body does not think, but makes > higher the probability that the 1-p thoughts refers to the most probable > computations you are sharing with him. > > > > >> I realise, by the way, >> that when considering thought experiments, including your own, one >> should not treat them in a naively re
Re: On the computability of consciousness
On 20 Mar 2010, at 16:56, David Nyman wrote:
On 24 February 2010 17:57, Bruno Marchal wrote:
Please, keep in mind I may miss your point, even if I prefer to say
that you
are missing something, for being shorter and keeping to the point.
You
really put your finger right on the hardest part of the mind-body
problem.
Bruno, I've been continuing to think, and meditate, about our recent
discussions, and have been re-reading (insofar as I can follow it)
your on-line paper "Computation, Consciousness and the Quantum". I
feel I have more of a sense of how the aspects I've been questioning
you about fit together in the comp view, but if I may, I would like to
press you on a couple of points.
My original post on "non-computability" was motivated by re-reading
Chalmers and struggling again with his assertion that a zombie (e.g.
including the "3-p Chalmers" that wrote The Conscious Mind!) could
nonetheless refer to "consciousness" and hence be behaviourally
indistinguishable from a conscious entity.
So you talk here on the philosophical zombie which is counterfactually
correct.
Well, if by 3-p Chalmers you mean some 'body', such a body *is* a
zombie.
The 1-p Chalmers is Chalmers, the person. Its body does not think, but
makes higher the probability that the 1-p thoughts refers to the most
probable computations you are sharing with him.
I realise, by the way,
that when considering thought experiments, including your own, one
should not treat them in a naively realistic way, but rather focus on
their logical implications.
Indeed! Absolutely so. I thought this was obvious (it should be for
deductive philosophers).
The problem with Chalmers' logic seems to
me to be that he has to assume that his zombie will have formal access
to what AFAICS are non-formalisable states.
Well, assuming comp, if the zombie has the "right" computer in its
skull, it has access to the non formalisable propositions, notions, 1-
states etc. (the 3-states are always formal).
But if the zombie skull is empty, then its counterfactual correctness
is just magical, and it makes no sense to say it accesses some states
or not. There are no 1-person state (because it is a zombie), nor 3-
person state, because there is no digital machine in its (local) body.
Now, in CC&Q, and in
discussion, you appear to say that Lobian machines can in fact refer
formally to what is non-formalisable. This could at first glance seem
to support Chalmers' argument (which I assume is not your intention)
unless you also mean that the formal consequences ("extensions") of
such non-formalisable references would somehow be characteristically
different in the absence of the non-formal aspect (i.e. zombie-land
would in fact look very different). IOW, consciousness should give
the appearance of exerting a "causal influence" on the physical, in
(naive) everyday terms.
Yes indeed. Except that "appearance" applies on the "physical". The
"causal" is the real thing, here, and it is incarnated, or implemented
with infinite redundancy (like the M set) in elementary arithmetic.
In CC&Q you point out that "we must not forget that the extensions
must not only be consistent, but must also be accessible by the
universal dovetailer". Hence, which extensions are accessible by a
conscious (non-formalisable) decision-maker would appear nonetheless
to be formalisable.
Indeed, by the UD, or by that tiny (but sigma_1 complete) fragment of
arithmetic, like Robinson arithmetic. It does not need to be Löbian.
The UD is NOT a Löbian entity. It is much logically poorer.
Again, my question is: how would the range of
accessible extensions for a zombie (purely formal) decision-maker be
characteristically different? For example, you cite the
"self-speeding-up" effect of consciousness with respect to the
organism's relation to its "neighbourhood" as a pragmatic argument for
the selective utility of consciousness. I assume this implies that a
conscious decision-maker would be likely to find itself in
characteristically different extensions to its "environment" as
compared with a non-conscious decision-maker, but some clarification
on this would be very helpful.
This is not entirely clear for me. For a non-conscious decision-maker,
there is just no sense at all to say that he could find itself (in the
first person sense) in some particular environment.
There is a sense in which it can find itself in the third person
sense, in some particular environment, but consciousness is a first
person notion, and it makes sense only when you ascribe it to the
(genuine) abstract computational states occurring infinitely often in
the UD*. It makes sense for a first person to find itself in an
infinite ensemble of computations/continuations.
Empirically we share a lot of very similar computations, and this
makes us believe that physics describes some local 3-reality, but comp
makes it describe only a sharable infinite set
Re: Jack's partial brain paper
You will find them by clicking on "publications" on my home page http://iridia.ulb.ac.be/~marchal/ The main one is "informatique théorique et philosophie de l'esprit" (theoretical computer science and philosophy of mind). Toulouse 1988. Like my thesis I have been asked to do in french (alas). You can ask any question, in case you have difficulty of understanding, or critics (I mean on the available english versions: sane04 is rather complete, if you accompany it with some books in mathematical logic. Best, Bruno On 19 Mar 2010, at 18:36, L.W. Sterritt wrote: Bruno, Your response is most appreciated. Your publications will keep me busy for while. You also mentioned earlier some of your publications that are not on your URL. That reference has gone missing in my labyrinthine filing system. Would you please post those references again. William On Mar 19, 2010, at 2:11 AM, Bruno Marchal wrote: William, On 18 Mar 2010, at 18:06, L.W. Sterritt wrote: Bruno and others, Perhaps more progress can be made by avoiding self referential problems and viewing this issue mechanistically. I don't see what self-referential problems you are alluding too, especially when viewing the issue mechanistically. Self-reference is where computer science and mathematical logic excel. A self-duplicator is just a duplicator applied to itself. If Dx gives xx, DD gives DD. Note the double diagonalization. That basic idea transforms mechanically "self-reference problem" into amazing feature about machines. The most in topic, imo, is that it leads to two modal theories G and G* axiomatizing (completely at the propositional level) the provable and true, respectively, logics of self-reference. Machines can prove their own limitation theorems, and study the productive geometry of their ignorance, and indetermination. They can easily infer a large class of true but unprovable propositions, and used them in different ways. Useful when an argument (UDA) shows that matter (physical science) are a product of that indetermination reflexion. It makes comp testable. Actually it leads to a general arithmetical interpretation of Plotinus neoplatonist theory of everything (God-without, God- within, the universal soul, intelligible Matter, sensible matter (qualia) etc.). The theory is there. It is also the theory on which converge the self-referentially correct machines which look inward. It is computer science. The key of comp. Where I start: Haim Sompolinsky, "Statistical Mechanics of Neural Networks," Physics Today (December 1988). He discussed "emergent computational properties of large highly connected networks of simple neuron-like processors," HP has recently succeeded in making titanium dioxide "memristors" which behave very like the synapses in our brains, i.e. the memristor's resistance at any time depends upon the last signal passing through it. Work is underway to make brain-like computers with these devices; see Wei Lu, Nano letters, DOI:10.1021/nl904092h. It seems that there is a growing consensus that conscious machines will be built, and perhaps with the new Turing test proposed by Koch and Tonini, their consciousness may be verified. Then we can measure properties that are now speculative. I think the contrary. If a scientist speculates that consciousness can be tested, he has not understood what consciousness is. We may evaluate it by bets, and self-identification. Any way, this is the strong AI thesis, which is implied by comp (*I* am a machine). With *I* = you, really, hoping you know that you are conscious. Tononi has interesting ideas, typically he belongs to comp. He is not aware, or interested, in the body problem to which comp leads (and he is wrong on Mary). But the comp body problem is not just a problem. Like evolution theory, it is the beginning of an explanation of where the appearance of a material world comes from, and why it is necessary, once you believe in 0, 1, 2, 3, ..., and addition and multiplication. I guess I'm in the QM camp that believes that what you can measure is what you can know. What I say depends only of saying yes to a doctor at some level. No problem if you choose the quantum level. In all case physics has to be derived, in a precise way (based on the logics of self- reference) from arithmetic (see my url for the papers). Bruno William On Mar 18, 2010, at 1:44 AM, Bruno Marchal wrote: On 17 Mar 2010, at 19:12, Brent Meeker wrote: On 3/17/2010 10:01 AM, Bruno Marchal wrote: On 17 Mar 2010, at 13:47, HZ wrote: I'm quite confused about the state of zombieness. If the requirement for zombiehood is that it doesn't understand anything at all but it behaves as if it does what makes us not zombies? How do we not we are not? But more importantly, are there known cases of zombies? Perhaps a silly question because it might be just a thoug
Re: On the computability of consciousness
On 24 February 2010 17:57, Bruno Marchal wrote:
> Please, keep in mind I may miss your point, even if I prefer to say that you
> are missing something, for being shorter and keeping to the point. You
> really put your finger right on the hardest part of the mind-body problem.
Bruno, I've been continuing to think, and meditate, about our recent
discussions, and have been re-reading (insofar as I can follow it)
your on-line paper "Computation, Consciousness and the Quantum". I
feel I have more of a sense of how the aspects I've been questioning
you about fit together in the comp view, but if I may, I would like to
press you on a couple of points.
My original post on "non-computability" was motivated by re-reading
Chalmers and struggling again with his assertion that a zombie (e.g.
including the "3-p Chalmers" that wrote The Conscious Mind!) could
nonetheless refer to "consciousness" and hence be behaviourally
indistinguishable from a conscious entity. I realise, by the way,
that when considering thought experiments, including your own, one
should not treat them in a naively realistic way, but rather focus on
their logical implications. The problem with Chalmers' logic seems to
me to be that he has to assume that his zombie will have formal access
to what AFAICS are non-formalisable states. Now, in CC&Q, and in
discussion, you appear to say that Lobian machines can in fact refer
formally to what is non-formalisable. This could at first glance seem
to support Chalmers' argument (which I assume is not your intention)
unless you also mean that the formal consequences ("extensions") of
such non-formalisable references would somehow be characteristically
different in the absence of the non-formal aspect (i.e. zombie-land
would in fact look very different). IOW, consciousness should give
the appearance of exerting a "causal influence" on the physical, in
(naive) everyday terms.
In CC&Q you point out that "we must not forget that the extensions
must not only be consistent, but must also be accessible by the
universal dovetailer". Hence, which extensions are accessible by a
conscious (non-formalisable) decision-maker would appear nonetheless
to be formalisable. Again, my question is: how would the range of
accessible extensions for a zombie (purely formal) decision-maker be
characteristically different? For example, you cite the
"self-speeding-up" effect of consciousness with respect to the
organism's relation to its "neighbourhood" as a pragmatic argument for
the selective utility of consciousness. I assume this implies that a
conscious decision-maker would be likely to find itself in
characteristically different extensions to its "environment" as
compared with a non-conscious decision-maker, but some clarification
on this would be very helpful.
David
>
> On 23 Feb 2010, at 22:05, David Nyman wrote:
>
> Bruno, I want to thank you for such a complete commentary on my recent
> posts - I will need to spend quite a bit of time thinking carefully
> about everything you have said before I respond at length.
>
> Thanks for your attention, David.
> Please, keep in mind I may miss your point, even if I prefer to say that you
> are missing something, for being shorter and keeping to the point. You
> really put your finger right on the hardest part of the mind-body problem.
>
>
> I'm sure
> that I'm quite capable of becoming confused between a theory and its
> subject, though I am of course alive to the distinction. In the
> meantime, I wonder if you could respond to a supplementary question in
> "grandmother" mode, or at least translate for grandma, into a more
> every-day way of speaking, the parts of your commentary that are most
> relevant to her interest in this topic.
>
> I am a bit panicking, because you may be asking for something impossible.
> How to explain in *intuitive every-day terms* (cf grandmother) what is
> provably counter-intuitive for any ideally perfect Löbian entity?
> Bohr said that to say we understand quantum mechanics, means that we don't
> understand.
> Comp says this with a revenge: it proves that there is necessarily an
> unbridgeable gap. You will not believe it, not understand it, nor know it to
> be true, without losing consistency and soundness. But you may understand
> completely while assuming comp it has to be like that.
> But I will try to help grandma.
>
> Let us suppose that, to use the example I have already cited, that
> grandma puts her hand in a flame, feels the unbearable agony of
> burning, and is unable to prevent herself from withdrawing her hand
> with a shriek of pain.
>
> OK.
>
>
> Let us further suppose (though of course this
> may well be ambiguous in the current state of neurological theory)
> that a complete and sufficient 3-p description of this (partial)
> history of events is also possible in terms of nerve firings,
> cognitive and motor processing, etc. (the details are not so important
> as the belief that such a complete history could be given).
>
