On Sun, 22 May 2005, rmiller wrote:
I'm approaching this as a sociologist with some physics background so I'm
focusing on what the behavior system perceives ("measures"). If all possible
worlds exist in a superpositional state, then the behavior system should
likewise exist in a superpositional state.
First, it looks like you are confusing the multiverse of QM with the
plenitude of "all theories" or all UTM programs (Level 3 with Level 4
multiverse in Tegmark's terminology). Different level 4 worlds do not
superpose, they don't relate to each other in any way, by definition.
Second, in QM you need to distinguish between two kinds of superposition:
those which cause interference effects (e.g. 2-slit experiment), and those
which don't (because the wave functions of the superposed "worlds" don't
overlap or are incoherent). "Behaviour systems" are complicated enough
that it is a mathematical certainty that they fall in the second class. In
which case there is no way to detect that the superposition is happening;
for all practical purposes each world goes its own sweet way.
If there are say, 10 possible
"worlds" available to the behavioral state (percipient) but each world
differs from the other by elements that are not observed by the percipient,
then the behavior system is under the assumption that interaction is taking
place with a single, unified environment.
Recalling the Copenhagen interpretation: does Chicago exist if you happen to
be by yourself in a hotel room in Des Plaines, IL? The answer is irrelevant
until the behavior system begins to experience some aspect of Chicago.
The superposition properties depend on the information available in the
whole system (e.g. your hotel room), not just the mind of the observer.
The world is constantly in close touch with itself. For instance, if
Chicago vanished in a large quantum fluctuation, photons which would
otherwise have been reflected from its streets to the clouds would be
different. Hence photons leaving the clouds that land in fields 40 miles
away would be different and so on. Very soon (within microseconds) the
photons comint through your hotel window are affected, and you become 100%
correlated with the state of Chicago. From your point of view, Chicago is
either there or not.
What if Deutsch is incorrect about contact between the various worlds?
i.e., what if quantum theory is wrong and a different theory applies?
But the only reason we have to believe this stuff is the evidence in
favour of QM (which is pretty overwhelming).
Suppose the behavior system normally exists across a manifold of
closely-linked probabilities, with the similarities forming a central
tendency and the differences existing at each edge of the distribution?
Again, QM makes definitive (but difficult-to-understand) predictions about
this. Yes there is a manifold of possibilities, in fact an infinite
number of them, for instance "configuration space" which is the manifold
of all particle positions (3N dimensional for N particles), or "momentum
space", the manifold of all particle momenta (also 3N dim). According to
QM, the probability distributions in these manifolds are not independent,
e.g. config and momentum wave functions are related by a Fourier
Your "central tendency" is just the wave function, which is peaked around
some configuration of particles in any given branch. What people generally
don't factor into this is just how *BIG* 3N dimensional spaces are, when N
is macroscopic. Even apparently minor differences, such as the presence or
absence of a speck of dust, correspond to enormously large separations in
configuration space. Although technically there is some (usually
infinitesimal) amplitude for all configurations, the only way you can get
a useful amplitude for two macroscopically different "worlds" is to
amplify some quantum behaviour, in which case the wave function splits
into 2 or more branches, each of which behaves more or less according to
classical physics. The "width" of the distribution for a single branch
corresponds to ordinary microscopic quantum fuzziness. Hence the branches
don't overlap in configuration space (or in the space of any other
macroscopic variable), and so can't communicate.
If the behavior system can perceive only a small chunk of information at
a time, then it may be possible that each percipient really does live in
his or her own little world---a small island of similar probabilities
made"real" from the larger cloud of probabilities.
We are all in our own little worlds, but in an objective sense; the same
is true for "non-behavior" systems, e.g. rocks.
If we quantify a behavior system in terms of elements and interactions
between elements, we arrive at a complex, but definable state. If that
behavior system exists across multiple worlds that differ in minute details
(i.e. a unobserved kitchen saucer moved an inch to the side) then the
behavior systems would exist as identical entities (or, as my friend Giu P.
would say, *shadows*) across the similar "sections." Employing a little
math, the behavior system could exist as an object in Z space--not too
different than a fibre bundle in topology. Differences among the realized
probabilities among these "shadow worlds" might show up at each end of the
normal distribution, but may be still be perceived by the behavior system as
guesses or hunches, depending upon where the primary centre of the behavioral
bundle is at the time. Psychology experiments in the 1980s suggest (to me
anyway) that a psychological mechanism has evolved that helps the behavioral
system "negotiate" this territory.
If this has anything to do with the basic laws of physics we have to
replace QM. Personally it sounds to me like great SF and very duff
Bottom line, it may be useful to take a step back and challenge some of our
primary assumptions---namely, that we exist in a discrete world in the
multiverse and that we can never "step" into the one next door. That is, we
may be wondering why we can't visit the next room, when in fact, we inhabit
the entire neighborhood.
This is not so much an assumption as a fact. Schrodinger's cat is either
alive or dead and once the experiment is done we can't change the outcome.