meekerdb wrote:
On 5/11/2015 6:54 PM, Bruce Kellett wrote:
meekerdb wrote:
On 5/11/2015 12:14 AM, Bruce Kellett wrote:
Bruno Marchal wrote:
On 07 May 2015, at 14:45, Bruce Kellett wrote:
......
Now, having read this many times, and looked at the other
summaries of the MGA, I still feel that something crucial is
missing. We go from the situation where we remove more and more of
the original 'brain', replacing the removed functionality by the
projections from the movie, which, it is agreed, does not alter
the conscious experience of the first person involved, to the
conclusion that the physical brain is entirely unnecessary;
indeed, irrelevant.
Hmm... On the contrary: the brain is necessary. It is the primitive
physicalness of the brain which is not relevant.
That is not what you say in the paper. "Hence, consciousness is not
a physical phenomenon, nor can it be a phenomenon relating to
observed matter at all." You go on to say that the appearance of
matter cannot be based on a notion of primitive matter. But these
are different things. Elsewhere you appear to agree that
consciousness does depend on the observed physical brain. In fact,
it would be foolish to deny this given the weight of physical
evidence that shows this to be the case.
Now that I have had a couple of days away from the internet to think
about this, and have read other comments on this thread, I think I
understand better the point that was not clear to me from the
COMP(2013) paper. What your intuition claims to be absurd in the MGA
is that replaying the film can instantiate consciousness. The reason
for this is based on your belief that replaying the film is not a
computation, and since the basic assumption is of comp is that
consciousness is Turing emulable -- is in fact a computation -- we
cannot have consciousness without the associated computation.
I think this obfuscates the point. One says yes to the doctor not
because one's conscious thought is a computation, but rather because
the doctor proposes to replace part of your brain with something that
will perform ALL the computations that part of the brain could do.
It is not that consciousness is a computation, rather it is a class
of computations that will map all possible (not just actual)
environmental inputs into outputs. And that's why a recording is
ruled out - whether it would be conscious or not; it is not
counterfactually adequate.
The recording is not supposed to instantiate a fully conscious person,
capable of actiang normally in a changeable environment. All is was
ever presumed to do was replace just the one conscious moment (or
string of moments) that were originally recorded. This whole argumetn
about counterfactual correctness is a total red herring.
Why? Have you proven that consciousness supervenes on a record?
Have you proven that it does not? The assumption of the argument was
that consciousness supervenes on the brain state. If that is the case,
then reproducing the brain state reproduces the consciousness. (Not a
brain replacement, but the consciousness of that recorded moment or
moments.)
The claim that the film (and projection) is not a computation is
thus false.
No, I think it's true because it's not counterfactually correct.
Whether you call it a computation or just and look-up table is, as
Russell points out, a matter of intuition about size. How many
counterfactuals must it deal with? Whether the ultrafinitism is true
or not, our theory of the world and consciousness should not depend
on there being infinities. So within ultrafinitism all TM's can be
replaced by lookup tables. Or looked at the other way around, a
sufficiently enormous lookup table is a computer.
As stated above, counterfactual correctness is not required to
reproduce just the one original conscious moment.
But how do you know this. Consciousness is somewhat mysterious. Bruno
starts with idea that you can replace part of the brain with something
that is I/O functionally identical. Saying yes to this doesn't commit
you saying that a recording is functionally identical.
You are conflating two issues. "Yes Doctor" is not about recordings, but
fully functional general computers that can reproduce all the functions
of your brain. We are talking here about a recording of one set of
conscious moments.
You would very
likely only say yes if the device were counterfactually correct for at
least a large range of inputs. So it certainly doesn't follow from
"saying yes to the doctor" that you must also agree that a recording
will instantiate consciousness.
That a recording is conscious is plausible because supervenience on a
sequence of physical states is plausbile. But it has its own problems:
Like the rock that computes everything, the sequence of states may be
conscious of everything.
The same goes for the existence of normal numbers, which must contain
within themselves all possible states of the dovetailer, including all
possible relations to the universal numbers that Bruno insists are
essential for there to be a computation.
So conterfactual correctness is not important for the single conscious
moment. It might be one way of saying that a conscious person is one
who can respond, more or less appropriately, to a range of physical
circumstances (external inputs), but it says nothing about separate
conscious moments.
I think the concept of separate conscious moments is incoherent.
Conscious "moments" need to have duration and to be in overlapping
sequences, and they need to occur in reference to an environment.
That is why I try to be careful to refer to a conscious moment *or moments*.
But it is by no means clear that a need for counterfactual correctness
can be concluded from the computations of the dovetailer. The idea
there seems to be that the same conscious instant (or sequence of
instants) is reproduced many times in the dovetailer, and many of
these will lead to different continuations, implying that each
instance has to be more flexible than is required for *that* instant.
I think you are equating a conscious moment with a state of the
dovetailer. I think a conscious moment, a thought, must correspond to a
long sequence of dovetailer states which may not be identical but only
'similar enough' at a classical level.
I think this is where your concern about the need to include quite a bit
of the external world in the set of states that instantiate a
consciousness comes to the fore. The assumption is that consciousness
supervenes on the physical brain. That brain obeys deterministic
physical laws, so the set of dovetailer states that gives a
consciousness must also give a coherent world that obeys a set of
deterministic physical laws. Since the same laws obtain over the whole
universe that is accessible to us (I do not comment on type II or IV
universes), then the relevant set of dovetailer states must cover the
whole history of the observable universe. If they did not, there would
be no requirement that the laws apply consistently across the whole
universe. We have to have a theory that agrees with observation at at
least this level. SO I am beginning to think that the dovetailer states
that sustain one consciousness must also sustain all other
consciousnesses that exist, or have existed, as well as the whole
physical universe from the time of the big bang.
OK, this might be possible, but I think it reduces the set of
dovertailer states over which one must sum to a set of essentially zero
measure. Does this make sense? I struggle to think that it does.
But I have a problem with this if the dovetailer is instantiating a
classical physics model of consciousness. There is an important
theorem in complex analysis that states that if two analytic functions
coincide in a neighbourhood, no matter how small, then the two
functions are equal everywhere in the region over which they are
analytic. This has its parallel in classical physics, where it can be
shown that if one is given initial data over some Cauchy surface, then
the complete past and future of that system is determined, and
calculable in terms of known physical laws.
This all depends on continuous functions. I don't think anything
similar applies to digital computations.
I think that, in effect, it does. See above.
This seems to imply that if two computations coincide for some
sequence of conscious states, then the continuations of those
computations must be identical. If they are not, then the computations
do not instantiate consciousness that is governed by deterministic
physical laws.
That's why Bruno wants to model consciousness as a bundle of threads of
computation. The bundle can divide and diverge presumably modeling
Everett's MWI.
But Everett's MWI is entirely deterministic and law-governed. It cannot
be reproduced by random variations from dovetailer state to dovetailer
state. I think the importance of the deterministic nature of the MWI is
frequently overlooked.
Bruce
--
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 [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.
