On 14 Aug 2014, at 10:34, Russell Standish wrote:
On Thu, Aug 14, 2014 at 09:59:31AM +0200, Bruno Marchal wrote:
A human being or any physical system reacts to the world in one
way or another. What was asked was for was counterfactual
correctness, i.e. the the MG reacts the same as would the
conscious being emulated - which might be no change at all.
I agree with you. The counterfactualness needs "If we change the
input then the output will change in the relevant way". But I am
not sure that we need the actual (physical) counterfactual behavior.
I might differ from Russell here.
Counterfactual correctness is needed as part of the computational
supervenience thesis in order to forbid supervenience on recordings.
Counterfactual Correctness (hereafter CC) is an attribute of a machine
(or a program) which compute some function. In a first approximation,
the machine is CC is it computes the right function whatever the input
is.
We can extend the sense of CC on a computation, and say that a
computation is CC, if it is the computation (done by some universal
machine) of some CC program. We suppose by defalut that the universal
machine executing that CC machine/program is itself CC.
So I think that CC is already needed for having a computation,
although in practice, the CC character can be restricted on its domain
(the machine computes correctly the function, unless some input is too
big, for example).
Let me try to be slightly more technical. People can ask question if
they don't remind the meaning of a term here.
Let phi_i be an enumeration of all partial computable functions (=
programmable on a computer).
I will call a computation "raw" if it is described by a sequence
phi_i(j)^n. So it is the steps of the computation of the UD itself,
when computing the n first steps of the computation of the function
phi_i (= executing the first n step of program i) on the input j. OK?
Here, it is the program or machine i which is CC, and this makes sense
only relatively to some semantics. A program computing wrongly the
factorial function, might be said to compute correctly some variant of
the factorial function!)
Now the computationalist supervenience thesis will associate
consciousness to an abstract entity, called the first person (and
approximated by the believer/prover + truth). Here the truth (as we
assume comp) is (notably) that the relevant computational state
belongs to an infinity of raw computations made by the UD. By
definition they are all CC (in the extended sense). When they diverge
on different inputs (like Washington or Moscow) they both do the
relevant corresponding behavior.
So, counterfactualness is "in" the program (even before it run), and
is kept in the raw computations corresponding to the relevant program
in the UD.
To sum up: the computationalist supervenience thesis associate a
conscious state (including its feeling being at this time at this
place) to an infinity of computations, which are CC by definition of
computations.
Now, a record of a computation, is obviously not CC. I would say that
it does not compute at all. It is a description of the sequence of
steps of a computation, but there is no universal machine going
through those steps in virtue of being itself a universal computer.
The movie projector, in particular does not compute (or just in a weak
sense unrelated to the computation it projects the movie of).
I disagree with your idea that to have counterfactual correctness we
need the actualisation of the diverging computations. We do have them
in UD*, so that is not a problem. But we don't need them.
If I give a classical well defined input to a classical (non quantum)
program, it will computes the same output, from the same input in all
the multiverses, except for the non normal "white rabbit" worlds. The
quantum counterfactualness (which exists and can be related to the
multiverse structure) does not seem relevant here, except that UDA
imposes such an actualization of different computations below our
level of substitution (and this will be confirmed by the machine's
talk about the []p & <>p, which gives a quantum logic (and quantum
logic are sort of logic of counterfactual or conditionals, as shown by
Hardegree).
The problem with physical supervenience, is that we can build a record
of a computation (thus non conscious), but becoming CC when inactive
physical stuff is in its neighborhood, making weird the role of the
physical with respect to consciousness. This suggests that the (often
considered immaterial) consciousness is related to the immaterial set
of raw computations going through that states, or up to that states
(ate least something like that), and not to any of its particular
emulation by the UD, or another program (non raw computation, itself
done by a raw computation).
Physical supervenience is something observed.
I don't understand what you mean by that. I think you might mean that
we do attribute consciousness to "normal brains", and we can do that
indeed with comp. But the real person is in Platonia and is
distributed to infinitely may emulations in Platonia or UD*. In this
case it is a little, and computer generable platonia, as it is the
effective enumeration of the true sigma_1 sentences. It is equivalent
with UD*).
Again, in order to
prevent supervenience on physical recordings, actual physical
counterfactuality is required - which is basically the quantum nature
of physical reality.
This is close to something I find deep (and plausible) but you are too
quick, at least for me. And you do use, like deutsch the idea that the
counterfactuals need to be realized to make sense, but I disagree with
this, even if they $are* all realized in the sigma_1 arithmetic.
This reminds me that I do think we can extract the quantum nature of
reality from MGA+Maudlin, in a sense deeper than in step 7 (where
physics has already a quantum aspect due to the infinitely many
computations which "interfere" statistically below our substitution
level, especially compared to the Feynman sum of history formulation
of QM).
But for this I need Hardegree result showing that QL, with the Sazaki
hook implication, which is nice because the orthomodularity becomes
equivalent with the modus ponens, provides a logic of conditional/
counterfactuals, and apparently the logics []p & p, []p & <>p, []p & p
& <>p do that too, when p is restricted to the sigma_1 sentences.
But we need, if you are correct, only *that* sort of actualization of
the counterfactuals, and that can't be used to attribute a
"primitively material" character to it.
OK?
We might need to work on this. The problem is that counterfactual are
counterintuitive and that is why I tend to trust more the logics
G1(*), S4Grz1, Z1(*), X1(*) than my intuition on this.
Cheers
Bruno
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.