From hpstapp@lbl.gov Thu Aug 2 19:24:31 2007
Date: Thu, 2 Aug 2007 19:24:31 -0700 (PDT)
From: Henry P. Stapp
To: sklein@berkeley.edu
Subject: Re: Article on Free Will and Determinism
On Thu, 2 Aug 2007, Stanley Klein wrote:
> Hi Henry,
> Do you know what 't Hooft is up to in the following article?
> Why is it that different from
> Bohm's deterministic theory.
>
> How was the Quantum Mind meeting in Europe?
> Stan
>
> From issue 2615 of New Scientist magazine, 01 August 2007, page 10-11
't Hooft's paper was a reply to the paper entitled
"The Free-Will Theorem" by John Conway and Simon Kochen.
So I will comment first upon that work.
This "Free Will Theorem" is based on "The FIN Axion" which asserts that:
"There is a finite upper bound to the speed with which
information can be effectively transmitted."
This "axiom" must be used with great care. It is well-known
that the formalism of Relativistic Quantum Field Theory (RQFT)
is 'Relativistic" in the sense that it allows no "signal" to be transmitted
faster than the speed of light. So RQFT does conform
to "The FIN Axiom" if by "effectively transmitted" one is referring
to the transmission of a "signal". Here a "signal" means a controllable
dependence of a faraway observable upon a sender's choices (of how he will
act); a dependence that affects the faraway situation in a way that would
allow the sender to convey information available to him, the sender, to a
faraway observer.
RQFT conforms to the demand of relativity theory that it allow no SIGNAL
to be transmitted faster than light (FTL). However, the formalisn of RQFT
(say a la Tomonaga & Schwinger) is incompatible with the seemingly quite
similar property that no INFORMATION about a choice that can treated as a
"Free Choice" can be transmitted FTL. Specifically, RQFT is incompatible
with the following two assumptions:
(1) What is observered and recorded in an earlier region is independent of
which experiment (say from some finite set of alternative possible
experiments) will be performed at a later time,
and
(2) information about a free choice made in one space-time region cannot be
transmitted to a another such region faster than the speed
of light.
Both of these assumptions are closely connected to, and probably entailed
by, the Fin Axiom, loosely interpreted.
Orthodox RQFT (with the collapse postulate) gives an explicit simple
example of a theory that conforms to the weaker no faster-than-light
SIGNAL velocity condition, while violating the stronger condition of
no FTL transmission of INFORMATION, provided the latter is interpreted as
entailing the two properties listed above. [In the final appendix of my
book, Mindful Universe, I give a simple proof that the
PREDICTIONS of RQFT (without using any reference to the way these
predictions are generated) are incompatible with the stronger "no FTL
transmission of information" property, provided that property is
interpreted as entailing the two properties (1) and (2) listed above.
An examination of the Conway-Kochen proof reveals that
it uses this stronger version of the no ftl condition,
and that their proof simply re-derives the inconsistency of their
two postulates, namely (A), the FIN Axiom (of no-ftl transmission
of INFORMATION---as opposed to SIGNAL), and (B), the assumption that the
choices of which experiments are to be performed can be treated as
"free choices", i.e., as "not determinded by the entire history of the
universe available to them". Here "availability" is limited by assumption
(A).
I conclude that the proper conclusion to be drawn from the
Conway-Kochen analysis is simply that their FIN axiom, as they apply it,
is incompatible with the conjunction of (a) certain predictions of RQFT,
and (b) the "free will assumption" that the choices made by the
experimenters can be treated as free variables.
Conway and Kochen assume the truth of their FIN axiom, applied in the
way that they do, and thus deduce from that assumption that if
our choices human choices are indeed free, in the sense
that they do not depend upon the information available to them, then some
actions of particles must also be free in this sense.
It seems clear to me their absurd-sounding conclusion is a consequence to their
failure to recognize the inconsistency---which they prove---of their premises.
Some confusion stems from a failure to properly define
the notion of "Free Choice".
In my way of thinking the notion of "Free Choice" only arises
within a non-reductive framework: i.e., within a framework in which
psychological/mental processing is NOT reducible to physical
process. In such a framework the idea that a choice is "free"
means that it is determined by a psychological process involving
reasons and psychologically felt values, AS CONTRASTED TO
mechanical/mathematical laws of the kind exemplified by classical
physics, and by the closely associated mathematical (Hilbert Space) aspects
of QM.
In a deterministic reductive framework this contrast between psychological
and mechanical is not maintained,
and the notion of free will becomes meaningless!
Now on to 't Hooft!
't Hooft has been working hard to derive QM from a deterministic
classical-type underpinning. His program is, I believe, not yet
complete. When is completed it will be interesting to see how his
theory accommodates the proven failure of RQFT to satisfy the property
of no FTL transmission of information.
The failure is entailed by the conjunction of some simple
predictions of RQFT with the notion of "Free Choice". But
't Hooft's mechanistically deterministic theory has no "Free Choice"
so he is, in some sense, off the hook. But not completely!
In the involved experiments the pertinent predictions are essentially
that if one performs a certain experiment in one region and gets a certain
outcome then if one performs a certain experiment in another region---
which is spacelike separated from the first---one will certainly get
a particular *specified* outcome.
Of course, the actions of the experimenters are certainly influenced by
their pasts, and are therefore not really "free". However, the experimenters
can let the choice of experiment be determined by the detection of
cosmic rays from distant sources, or by detections of local radio-activity,
or by weather reports from airports etc. etc.
YET THE PERTINENT PREDICTIONS ARE SPECIFIED TO BE COMPLETELY INDEPENDENT OF
HOW---FROM AMONG THE MYRIADS OF SEEMINGLY WHIMSICALLY SELECTED POSSIBILITIES
---THE EXPERIMENTS WERE DETERMINED FROM THE PAST. So there is an
important sense in which the past is really not relavant to these
choices: even if the choice of experiments are acually fixed by the past,
the causal chain can be so varied and circuitous, *with no effect at all on
the prediction*, that the idea that choices effectively could go either way,
independently in the two regions, with the predictions borne out no
matter which combination of experiments is picked, is hard to avoid.
So even in a strictly deterministic setting the idea that the
choices can be treated *as if* they are free variables is not easily
dismissed. More specifically, one can ask: How does 't Hooft's deterministic
classical-type theory actually accommodate the logically required violation
of the no FTL transmission of information condition?
It is of course true that the choice of the experiment
can, in a deterministic theory, be regarded a condition on
an "unconstrained initial state" as 't Hooft notes.
But the idea that all of the myriads of ways that the experimenter
might choose to determine the experiment in his region would all
---for one choice of experiment---lead to initial conditions that
would all produce some specified outcome in the faraway region is not
something that can simply be assumed: it would be a seemingly profound
extremely nontrivial consequences of 't Hooft's theory, which he needs
to explain. I would regard a successful explanation, via conditions
on unconstrained initial conditions, of this apparent FTL effect
as an acid, and extremely illuminating, test of t'
Hooft's finished theory.
My expectation is that his finished deterministic classical-type theory
will not be able to explain this logically required FTL effect while
rigorously forbidding FTL signalling.
I'll tell you about the Quantum Mind Conference in Salzburg in a separate
message.
Best,
Henry
> From issue 2615 of New Scientist magazine, 01 August 2007, page 10-11
>
> REDEFINE the concept of free will? Only a Nobel laureate would have
> the nerve. Last year, the Dutch physicist Gerard 't Hooft announced
> that the weird effects that spring from quantum mechanics arise from
> a deeper deterministic reality based on classical physics. People
> objected that his theory appeared to rob us of free will, and now 't
> Hooft has responded by moving the goalposts. No, we don't have free
> will as it is commonly understood, he says - but that's because the
> way it is commonly understood is wrong.
>
> 't Hooft, of the University of Utrecht in the Netherlands, shared a
> Nobel prize in 1999 for laying the mathematical foundations for the
> standard model of particle physics. Like Einstein, he was troubled by
> the indeterminism at the heart of quantum mechanics, according to
> which particles do not have clearly defined properties before you
> measure them, and you can never predict with certainty what the
> outcome of your measurements will be. So 't Hooft constructed a
> deterministic alternative which showed that fundamental states which
> exist on the smallest scales do start out with clearly defined
> properties. Information about these states gets blurred over time,
> until we are no longer able to tell how they initially arose -
> leading to their apparently probabilistic quantum nature, he says.
>
> However, mathematicians John Conway and Simon Kochen at Princeton
> University showed that if 't Hooft's theory is true, then people's
> ability to make instantaneous, unpredictable choices on a whim is
> similarly constrained - we don't have free will (New Scientist, 4 May
> 2006, p 8).
>
> The revelation has been a stumbling block for his theory, 't Hooft
> admits. "It's not the mathematics that loses other physicists," he
> says. "It's this metaphysical worry about free will. Why worry at all
> about a notion so flimsy as 'free will' in a theory of physics?"
>
> Imagine you are holding a cup of coffee. "I can't change my mind in
> an instant about whether to drink the coffee or hurl it across the
> room. My decision must have roots in brain processes that occurred in
> the past," he says. "What's important is that I have freedom to
> calculate what happens if I throw my coffee cup. Equally, I have the
> freedom to calculate the effects after I drink from my cup." What we
> lack is the freedom to instantaneously switch between which of these
> initial states we start from. 't Hooft calls his new formulation the
> "unconstrained initial conditions postulate".
>
> Hans Halvorson, a philosopher of physics also at Princeton
> University, agrees that our ideas of free will need to be revised.
> "It's likely that our naive gut reaction about what free will is may
> need to be radically rewritten in just such a way, if we really want
> to consider what's happening at the deepest levels," he says.
>
> Conway and Kochen say a deterministic theory denies us the freedom to
> choose what to measure about a particle's characteristics. The only
> way 't Hooft's theory matches experimental results, they say, is if
> nature is conspiring to prevent physicists measuring certain
> characteristics of a quantum particle by changing its properties at
> the same moment that they decide what to measure.
>
> 't Hooft sees nothing mysterious about this. Any decision about what
> to measure must have been influenced by environmental factors in your
> recent past, and it will take time to enact your choice as you modify
> your measuring apparatus. It's safe to assume that in this time, the
> particle you plan to measure will also be influenced by these
> environmental factors - a disruption that accounts for nature's
> ability to tweak what you are able to measure, he says.
>
> However, Antoine Suarez, a physicist at the Center for Quantum
> Philosophy in Zurich, Switzerland, remains troubled. "If 't Hooft is
> really correct, then the work for which he is famed was not carried
> out as a result of his free will. Rather, he was destined to do it
> from the beginning of time," he says. "In that case, maybe his Nobel
> prize should rightfully have been presented to the big bang instead."
>
> Suarez has performed an experiment that he claims proves 't Hooft
> wrong. 't Hooft's deterministic theory and his redefinition of free
> will rely on fundamental states obeying causal laws, so that a chain
> of events can be calculated precisely, given the starting conditions.
> By bringing the effects of special relativity into play in a standard
> entanglement experiment, Suarez and his colleagues were able to check
> how time flow interacts with the quantum world (see "Effects without
> causes"). "We tested the very concept of time," says Suarez.
>
> The result was a resounding success for quantum mechanics, says
> Suarez. His team showed that the well-behaved time-ordering 't Hooft
> needs simply doesn't exist: there is no causality at a deep level.
> Suarez is submitting his paper to Foundations of Physics, a journal
> that is edited by 't Hooft. "I think it will spark an interesting
> debate," Suarez says.
>
> 't Hooft is ready to meet that challenge. Although his theory cannot
> yet explain the results, he is confident that it will eventually do
> so. "After all, we know that quantum mechanics produces eccentric
> results," he says. "That's exactly why I am looking for an alternative."
>
> From issue 2615 of New Scientist magazine, 01 August 2007, page 10-11
>