From stapp@thsrv.lbl.gov Tue Apr 25 14:27:36 2000
Date: Tue, 25 Apr 2000 14:20:00 -0700 (PDT)
From: stapp@thsrv.lbl.gov
Reply-To: hpstapp@lbl.gov
To: kleinlist , bdj10@cam.ac.uk, brings@rpi.edu,
brucero@cats.ucsc.edu, chalmers@paradox.ucsc.edu, ghrosenb@ai.uga.edu,
hameroff@u.arizona.edu, hpstapp@lbl.gov,
Jeffrey M. Schwartz , keith@imprint.co.uk,
klein@adage.berkeley.edu, patrickw@monash.edu.au, phayes@ai.uwf.edu,
Kathryn Blackmond Laskey
Subject: Re: Deutsch claims Q-info is local (fwd)
---------- Forwarded message ----------
Date: Tue, 25 Apr 2000 14:16:21 -0700 (PDT)
From: stapp@thsrv.lbl.gov
Reply-To: hpstapp@lbl.gov
To: augustyn@erim-int.com
Subject: Re: Deutsch claims Q-info is local
On Mon, 24 Apr 2000 augustyn@erim-int.com wrote:
> Dear Henry,
>
> http://xxx.lanl.gov/abs/quant-ph/9906007
>
> In this June 99 paper, Deutsch claims that "all information in quantum systems
> is, notwithstanding Bell's theorem, localised." He claims there is a wide
> range
> of misconceptions stemming from belief in the contrary. He defines a new
> concept, locally inaccessible information, as "information which is present
> in a
> system but does not affect the probability of any outcome of any possible
> experiment on that system alone."
>
> Unless I've once again misunderstood something, this paper seems to be in
> direct
> conflict with your theory.
>
> Ken
>
>
>
Dear Ken
Deutsch adopts the Heisenberg picture, which is a good thing to do:
I used to do that myself. Then he says that he will keep the
state vector constant. This means he is adopting basically
the Everett view, where there are no "reductions". All nonlocal
effects are consequences of reductions, so by adopting a "no
reduction" (many-worlds or many-minds) approach he avoids
all nonlocal effects. The avoidance of nonlocality is THE
huge virtue of the "no reduction" approach.
Within the "no reduction" approach one is NOT entitled to assume
that when an experiment is performed, and an observer observers the
outcome, that some actual choice has been made that picks out that
one outcome and eradicates the other: both outcomes exist.
A second key point is that a correlation between the far-apart events
is not measured, or actualized, until the outcomes of the two far-apart
measurements are brought together (by classical communication mode, and at
speed not greater that the speed of light)) at some later time. I
stressed these points in my 1980 paper "Locality and Reality" (Found. of
Phys. 10, 767) about the local character of the "no reduction" approach,
and John Bell told me, to my surprise, that he thought that I was the
first one to mention the latter key point. Deutsch makes important use
of it in his treatment, as well as the key result in the field, which is
that the PROBABILITIES for outcomes in one region do not depend on
which experiment is performed in a spacelike separated region.
Deutsch's paper is basically just a restatement of these well
known things, specialized to the particular sort of interactions
that normally occur in quantum computing.
The main point is simply that the "no reduction" (many-worlds or
many-minds) theory is certainly local. The question, then, is whether
this virtue of being local is sufficient to make it viable?.
The troubles come in only when one wants to bring
in the human observers. If one brings them in then one has to
reconcile the fact that a human observer experiences only
one classical outcome with the fact that the theory has no
corresponding "reduction": both classically represented outcomes
must be carried along as existing actualities.
But if there is no reduction then what prevents both one brain-state
projector P corresponding to an experience, and then a very slightly
different P' also corrsponding to an experience, and both experiences
occurring. But then probabiliies for the different possibilities will not
add to unity.
The vN/W theory invokes the key vN process I (von Neumann
reduction)
S--> PSP + (1-P)S(1-P),
to make the probability calculus work. I follow THAT approach.
Henry
From stapp@thsrv.lbl.gov Thu Apr 27 11:26:05 2000
Date: Thu, 27 Apr 2000 11:24:08 -0700 (PDT)
From: stapp@thsrv.lbl.gov
Reply-To: hpstapp@lbl.gov
To: Aaron Sloman
Cc: hpstapp@lbl.gov, bdj10@cam.ac.uk, brings@rpi.edu, brucero@cats.ucsc.edu,
chalmers@paradox.ucsc.edu, ghrosenb@ai.uga.edu, hameroff@u.arizona.edu,
Jeffrey M. Schwartz , keith@imprint.co.uk,
klein@adage.berkeley.edu, patrickw@monash.edu.au, phayes@ai.uwf.edu
Subject: Re: Deutsch claims Q-info is local (fwd)
On Thu, 27 Apr 2000, Aaron Sloman wrote:
> Henry responded to this:
>
> > On Wed, 26 Apr 2000 augustyn@erim-int.com wrote:
> >
> > > Dear Henry,
> > >
> > > Thanks for clearing that up. The many-worlds approach appears be
> > > a reactionary head-in-the-sand way to avoid the problem of
> > > consciousness. There doesn't seem to be anything (to my meager
> > > knowledge) that explains why we are on the branch
> > > we find ourselves on. I'm surprised it is so popular.
> > >
> > > Ken
>
> That objection is just silly. On the many-worlds theory (e.g. as
> explained in Deutsch The Fabric of Reality (I have not yet tried to read
> the more recent paper)) in each branch we, or rather our counterparts,
> are ON that branch, so there is nothing to explain.
>
Yes. That is correct.
> The question (why we are on the branch we find ourselves on) presupposes
> that on the many worlds theory we are only in ONE branch, which would
> have to be explained. But that would be totally inconsistent with the
> theory.
Yes. There are, according to MWI, other branches of Ken, and of Aaron,
and each branch includes the conscious experiences that goes with it.
So Aaron is quite right to point out that the question
" why we are on the branch we find ourselves on" is not a problem for
MWI.
>
> > Locality is hard for physicists to give up.
>
> Deutsch gives up a kind of locality insofar as he explains interference
> phenomena as involving interactions between particles in alternative
> universes, if I've understood him.
I doubt that Deutsch explains interference in that way: in MWI
the interference phenomena are between parts of a single branch.
Indeed, if these physical phenemena were caused by interactions between
different branches then their physical occurrences would be evidence
for the existence of the other branches. But there is no such evidence.
The reasons for assuming that other branches do exist are purely
theoretical: that assumption saves locality and it saves the idea
that the deterministic Schroedinger equation of motion (or the
Heisenberg equations of motion, if one chooses to use the Heisenberg
picture, as Deutsch does) always holds; that the basic eqn. does not
suddenly fail when "mind" enters into the dynamics. The attitude of MWI
theorists, like that of classical-physics theorists, is that mind
is purely a local deterministic consequence of the physical structure
described by quantum theory, and that the Copenhagen approach
(or the vN/W approach) of bringing mind into physical theory was a
VERY bad idea.
>
> When I hear scientists debating such options without any possibility of
> settling which is correct, apart from personal preferences, I
> immediately suspect that neither has the truth, which needs to be
> formulated in such a way as to subsume both options.
The differences in the "explanatory powers" of the two theories is
very large. That difference provides one way for scientist to decide which
theory is better, in the sense of being more useful: the capacity to
provide "understanding" is a valid criterion for deciding between
theories. Reduction theories can give mind real power over brain events,
whereas MWI has no fixed place for mind and makes mind epiphenomenal:
it says that psychological processes do not really matter.
But then that theory is very flabby in terms of its predictive and
explanatory power in regard to the empirical findings pertaining to
how psychological processes to enter into brain dynamics, in contrast
to vN/W QT, which specifies in a fair amount of detail how mind controls
brain dynamics.
>
There is also the question of whether MWI is logically coherent
as a description of the universe, including the human observers
and their experiences.
Deutsch is mainly interested in quantum computation, and thus he consider
situations that are very specialized and idealized.
But how well does MWI work for human brains and the streams of
consciousness associated with them.
If the Schroedinger Eqn. govern everything then the state (in the
Schroedinger picture) is a superposition of a "continuum" of
possibilities. For each P there are a continuum of different nearby P's
that are not orthogonal to it: P P' not zero. But in a universe that is
essentially a continuous superposition of all possibilities there is no
way to specify, just on the basis of the evolving state, that P
corresponds to a possible experience but P' does not. The problem,
basically, is that in a realistic MWI universe the separation of the
state into a superposition of distincts orthogonal "branches" specified by
the state itself, at finite times, with no outside "observing system" or
arbitrary rule, cannot be achieved: there is no intrinsic-to-the-state
difference between P and P' if the state is a smeared out superposition of
all possibilities. But if P and P' both correspond to allowed experiences,
then the probability calculus fails.
von Neumann's process I, which is a reduction process, is needed to make
the probability calculus work. Each allowed experience must be associated
with a P that allows the state to be decomposed into PSP + (1-P)S(1-P),
where the two terms correspond to the experience either happening or not
happening. If there is no such reduction and another experience
corresponding to a slightly different P', with PP'not 0, can also
occur, then the probability calculus collapses in ruins: neither in
projections into the future nor surveys of the past will the quantum
probability rules hold. For the probability for some experience to
occur will be infinite, and there will be contribution to the brain
state in which certain experiences have both occurred and not occurred.
So I believe that MWI fails for technical reasons connected to
the "choice of basis" that is needed to implement the probability
calculus. This "choice of basis" enters vN/W QT as
the "asking by the `person' of some particular question".
This process must, I believe, be nonlocal because, in contrast to
the process described by the Schroedinger equation, it
concerns the (necessarily) nonlocal P. So my vN/W QT is built
upon a certain mind-based resolution to precisely the difficulty
that I believe wrecks the MWI. I was led to it by way of the MWI.
Henry