The interpretation of quantum mechanics: where

do we stand?
GianCarlo Ghirardi∗
Department of Theoretical Physics of the University of Trieste, and
arXiv:0904.0958v1 [quant-ph] 6 Apr 2009

the Abdus Salam International Centre for Theoretical Physics, Trieste, Italy, and
the Istituto Nazionale di Fisica Nucleare.

Abstract
We reconsider some important foundational problems of quantum me-
chanics. After reviewing the measurement problem and discussing its
unavoidability, we analyze some proposals to overcome it. This analysis
leads us to reconsider the current debate on our best theory, i.e. quantum
mechanics itself. We stress that, after the remarkable interest and the
many efforts which have lead, in the last years of the past century, to a
revival of the subject, and, more important, to new interesting results,
we are now witnessing a re-emergence of the vague and unprofessional
positions which have characterized the debate in the second quarter of
the XXth century. In particular we consider as extremely serious the fact
that a completely mistaken position concerning the real meaning of Bell’s
theorem seems to have been taken by many scientists in the field.

1 Introduction
As well known, almost one century after its formulation and in spite of its un-
precedented predictive successes in accounting for physical processes, quantum
mechanics is still at the centre of a lively debate. Everybody knows very well
the disputes between some of the great founding fathers of the theory, from N
Bohr to A Einstein to E Schrödinger to W Heisenberg, which have marked the
twenties and the thirties of the past century. The debate has been less heated in
the subsequent years, perhaps because, as stated by M Gell-Mann in his Nobel
acceptance speech in 1976:
Niels Bohr brainwashed a whole generation of physicists into believ-
ing that the problem (of the interpretation of the theory) had been
solved fifty years ago.
Actually I remember well that, at the beginning of the sixties when I started
my scientific carreer, to work on foundational issues was considered by a great
part of the scientific community a loss of time, a choice to pay more attention
to (irrelevant) philosophical issues than to precise scientific problems. Luckily
enough, the sixties were also the years in which another deep thinker, John S
∗ e-mail: [email protected]

1
Bell, by deriving the celebrated inequality that bears his name [1], has given a
tremendous imput to our understanding of reality by making clear that nonlocal
features characterize most natural processes. However, this great scientist did
not limit his considerations to the peculiar aspects of the theory stemming from
one of its most surprising features, entanglement, but has devoted many impor-
tant papers to give voice to his unsatisfaction with the conceptual and logical
status of the theory, in particular with the so-called measurement or macro-
objectification problem. His immense prestige has pushed many physicists to
reconsider such a problem and during the fall of the past century there has been
a re-flourishing of foundational investigations.
In this paper we will first of all concentrate our attention on the macro-
objectification problem and on some recent proposals to overcome it, and then
we will briefly review and analyze the positions which characterize the present
debate on this problem.

2 The macro-objectification problem

The problem we are interested in is usually presented by making reference to
the so-called von Neumann ideal measurement scheme. Synthetically it goes as
follows.
Suppose one is interested in measuring a physical observable F of a micro-
scopic system S. Let us denote as F the (self-adjoint) operator associated to
it, as fi its eigenvalues and as |φi i the corresponding eigenvectors. Obviously,
since S is microscopic we have not direct access to it and therefore, to actually
perform the measurement, we assume, with von Neumann, that we have at our
disposal a macroscopic quantum system in a “ready state” |AR i which we put
into interaction with S and that the system apparatus interaction which lasts
for a certain time interval is such that the initial state |φi i ⊗ |AR i evolves into
the final state |φi i ⊗ |Ai i. Here the final states of the apparatus |Ai i are orthog-
onal and macroscopically different. In this way, by looking at the final state of
the apparatus we can deduce in which of the eigenstates of F the microsystem
was before the measurement.
Now comes the problem: quantum mechanics is a linear theory, implying
that if various states are possible states for a system S then also any normalized
linear combination of them is a possible state for it. Moreover, the evolution
law is linear, so that a linear superposition of states evolves into the same linear
superposition of the evolved of such states. In formulae and with reference to
our example:

X X
|φi i ⊗ |AR i → |φi i ⊗ |Ai i ⇒ ci |φi i ⊗ |AR i → ci |φi i ⊗ |Ai i, (1)
i i

where the simple arrow denotes the time evolution and the double one denotes
the implication.
As Eq. (1) clearly shows, in the final state a superposition of macroscopically
different states appears. Now, what meaning whatsoever can we attribute to
a state which is a superposition (and not a statistical mixture) of such states?
Actually the states |Ai i of the apparatus might correspond to different locations

2
of a macroscopic pointer. How can then a macroscopic object be in a state which
does not correspond to a macroscopically definite position?
We all know the answer that the so-called ortodox interpretation of quantum
mechanics gives to our question: wave packet reduction takes place during the
process and the final state is not the one appearing at the end of Eq.(1), but,
with a probability given by |cj |2 , the state |φj i ⊗ |Aj i of the superposition.
This assumption is fundamentally inconsistent since it contemplates two
contradictory evolution laws for physical systems: one, which is linear and de-
terministic, governing the evolution of microsystems, and one, which is nonlinear
and stochastic, holding when macroscopic systems enter into play. Why should
macrosystems not to be governed by quantum mechanics? Are they not built
up of elementary constituents (electrons, protons, neutrons) which are subjected
to the laws of quantum mechanics? But this is not the whole story. In fact,
on the one hand, we know very well that there are macroscopic systems whose
properties can be understood only in terms of quantum theory, while, on the
other, even if we would be inclined to accept two different evolution principles,
we cannot avoid to recognize that there is nothing, absolutely nothing, in the
theory which allows us to locate the split identifying which systems obey the
linear equation and which ones do not. As lucidly summarized by Bell [2]:

There is a fundamental ambiguity in quantum mechanics, in that

nobody knows exactly what it says about any particular situation, for
nobody knows exactly where the boundary between the wavy quantum
world and the world of particular events is located.

To conclude this section we remark that the above ideal scheme of measure-
ment (such measurement processes are usually denoted as ‘of the first kind’) is an
extremely idealized one: in practice, the final apparatus states are not strictly
orthogonal, we cannot control all degrees of freedom characterizing them, a
macroscopic object cannot be isolated from the environment, no apparatus is
exempt from malfunctioning, etc. Some scientists maintain that it is just re-
sorting to such an idealized description as the one of Eq.(1) which leads to
the inconsistencies we have just discussed. As a paradigmatic example we will
mention a sentence [3] by H. Primas:

If you really want to discuss measurements of the first kind you may
try your luck. But remember that measurements of the first kind are
unrealistic and completely irrelevant for experimental science. Re-
place in all books and philosophical treatises on quantum mechanics
the word ‘measurement’ by the proper expression ‘measurement of
the first kind’ and add a footnote: measurements of the first kind
are idealizations which never play any role in experimental sciences.

Our next step is to prove that this position is plainly wrong and that the
macro-objectification problem cannot be avoided. To this purpose we will briefly
recall the argument of a recent paper [4] in which the problem has been tackled
in its full generality.

3
3 A completely general measurement scheme
To exhibit an explicit proof of the last statement of the previous section, let
us consider a microsystem which we assume can be prepared in two orthogonal
states |ui and |di (typically we can think of the two spin states of a spin 1/2
particle corresponding to a definite value of σz ), and in their equal amplitudes
superposition |u + di = √12 [|ui + |di]. We consider also a statistical ensemble
E{|AR , αi, p(α)} of apparata, devised to measure precisely the observable σz .
We have resorted to an ensemble of apparata just because we cannot have full
control of all their degrees of freedom. The state |AR , αi corresponds to the
‘ready state’ of the apparatus, let us say one for which the macroscopic pointer
of the apparatus ‘points, at 0’, while the index α distinguishes situations differing
for uncontrollable degrees of freedom such as the precise state of the molecules
of the pointer, and so on1 . The quantities p(α) are the weights of the various
subensembles of the statistical ensemble characterized by the macroproperty
AR . Completely in general, when we will resort to expressions of the type
|X, µi we intend to indicate a situation of the macro-apparatus corresponding
to what we perceive2 as ‘the pointer point at X’ and all other relevant variables
are summarized by the symbol µ. Let us denote as X the set of all such states for
all possible values of µ. We will not require orthogonality of states corresponding
to macroscopically different situations, but, obviously, we have to impose that
two such states are, to a large extent, “distinguishable”, a fact that we will
characterize by imposing:
√
infµ∈X ,ν∈Y k |X, µi − |Y, νi k≥ 2 − η, η ≪ 1. (2)

Let us go on with our assumptions. We put the system in one of the two
states |ui or |di into interaction with the specific apparatus in the state |AR , αi,
and we assume that the quantum evolution during the measurement is governed
by a unitary operator U (ti , tf ). Let us denote as |F, u, αi and |F, d, αi the
final states obtained when we trigger the apparatus in the state |AR , αi by the
microstates |ui and |di, respectively:

U (ti , tf )[|ui ⊗ |AR , αi] = |F, u, αi; U (ti , tf )[|di ⊗ |AR , αi] = |F, d, αi. (3)

Let us now consider some physically meaningful sets:

JU− = {α : |F, u, αi ∈ −
/ U}; JD = {α : |F, d, αi ∈
/ D}, (4)
and their complements JU+
= CJU− , +
JD = CJD−
. The physical meaning of these
−
sets should be clear. For instance JU represents the subset of those apparata
(actually the set of the α’s) for which, in spite of having being triggered by the
microstate |ui, their final statevector does not belong to the set U, i.e. the one
for which the pointer ‘points at U’. For α in such a set the apparata did not
register correctly the fact that the system we are interested in is in the state
|ui. The meaning of the other three sets, is obvious. We can now formulate
1 Note that we can include in α also degrees of freedom which are external to the apparatus,

typically, those referring to the environment.

2 It goes without saying that all statements concerning the position of the pointer do not

make reference to a situation in which the position of the pointer is a geometrical point, but
to one in which it lays in a quite narrow interval on a graduated scale.

4
our final assumption. By taking advantage of the fact that the function p(α)
is a probability measure on the set on which the index α runs, we can define a
measure by the formula:
Z
µ{X } = p(α)dα. (5)
α∈X

We now make a natural reliability assumption concerning our ensemble of ap-

paratuses:
µ{JU− } < ǫ, µ{JD
−
}<ǫ i.e. µ{JU+ } > 1 − ǫ, µ{JD
+
} > 1 − ǫ. (6)
Note that these relations imply that the set of all α’s for which the apparatuses
give the right answer both when triggered by |ui or by |di has a measure:

µ{JU+ ∩ JD
+
} > 1 − 2ǫ, ǫ ≪ 1. (7)
which is near to 1.
Having made precise our assumptions and notation we can now proceed as
follows. Let us trigger those apparatuses for which α belongs to JU+ ∩ JD +
(i.e.
those which register correctly both the state |ui as well as the state |di) by the
state |u + di. The linear nature of quantum mechanics implies, from Eq. (3):

1 1 1
U (ti , tf ) √ [|u+di⊗|AR , αi ≡ √ ]U (ti , tf )[|ui+|di]⊗|AR , αi = √ [|F, u, αi+|F, d, αi].
2 2 2
(8)
Let us now raise the question: does the final state of Eq.(8) belong to the
set U of states? The answer is no, because its distance from the state |F, d, αi
is:
1 1 1 1
k |F, u+d, αi−|F, d, αi k=k √ |F, u, αi+(1− √ )|F, d, αi k≤ √ +1− √ = 1,
2 2 2 2
(9)
while, according
√ to Eq.(2) a state belonging to U must have a distance almost
equal to 2 from any state belonging to the set D of states corresponding to a
macroscopically and perceptively different situation from the one which is, by
assumption, associated to the perception ‘the pointer points at U’. The same
argument can be followed to prove that the final state cannot belong to D or to
any other set of states corresponding to a definite position of the pointer.
The conclusion should be obvious. Having assumed that we have an en-
semble E{|AR , αi, p(α)} of macro-apparata which allows us reliably (but not
perfectly) to know whether the microstate triggering them is either |ui or |di,
then the large majority of such apparata (i.e. all those characterized by an α
belonging to the set JU+ ∩ JD +
, whose measure is near to 1), when triggered by
the superposition |u + di ends up in a state which cannot correspond to any
definite position-perception concerning the macroscopic pointer. The macro-
objectification problem cannot be overcome invoking measurement processes
more realistic than the von Neumann one: it is the very possibility of mea-
suring an observable with an acceptable level of reliability and the assumption
of the unrestricted validity of quantum mechanics which imply that the same
micro-macro interaction leads almost always to a perceptually nondefinite sit-
uation when the triggering state is the superposition of different eigenstates of
the observed quantity.

5
4 Decoherence
A quite popular proposal which has been put forward to overcome the macro-
objectification problem is based on the consideration of the unavoidable and
uncontrollable interaction of any macroscopic body with its environment. A
sketchy way of summarizing this proposal is the following. Let us consider a
state which is a linear superposition of two macroscopically different states of a
macro-object S:
|Ψ(sys) i = α|M i + β|N i. (10)
One then takes into account the above mentioned unavoidable interactions
of S with the environment, which we suppose to be described by the state |E0 i
at the initial time. Such interactions and the subsequent evolution lead to a
state of the system+environment which we will express as:

|Ψ(sys+env) i = α|M i ⊗ |EM i + β|N i ⊗ |EN i. (11)

Now we have to take into account two important facts. The states |EM i and
|EN i appearing in the above equation turn out to be practically orthogonal and
they are essentially out of the control of the experimenter. The second fact is
obvious, the first is made understandable by taking into account that, e.g., the
interaction of a macrosystem which is, let us say, located in a certain position
(when the state is |M i) interacting with a single molecule of the environment
can easily throw it into an excited state which is orthogonal to the state of
the same molecule when it has not interacted with the macrosystem since it
is differently located (as, we suppose, is the case for the state |N i). But this
is not the whole story; even if for the two macroscopically different states of
the macrosystem no one of the microsystems of the environment is in one of a
pair of orthogonal states, for sure many microsystems will be in different states
in the two cases. Such states, being different, have a scalar product which is
smaller than 1. If one then takes into account that we can easily have, let us
say, an Avogadro number of such states (corresponding to the many particles
which have interacted with the macrosystem), one easily realizes that the two
states |EM i and |EN i are, de facto, orthogonal.
As already stated, however, the experimenter has no control over the states
of the particles of the environment: he is only interested in the physics of the
macrosystem he is studying. Now, quantum mechanics teaches us that to de-
scribe the properties of a subsystem of a composite system, the most efficient
and formally direct way is the one of resorting to the statistical operator formal-
ism and to take the partial trace on the degrees of freedom we are disregarding.
Let me summarize the steps which are involved using both the language of the
statevectors and of the statistical operators:

|Ψ(sys) i → |Ψ(sys+env) i (12)

(sys) (sys) (sys) (sys+env) (sys+env) (sys+env)
ρ = |Ψ ihΨ | → ρ = |Ψ ihΨ |

We recall now the explicit form of ρ(sys+env) following from Eq.(11):

ρ(sys+env) = |α|2 |M i ⊗ |EM ihEM | ⊗ hM | + |β|2 |N i ⊗ |EN ihEN | ⊗ hN |(13)

+
+αβ ∗ |M i ⊗ |EM ihEN | ⊗ hN | + α∗ β|N i ⊗ |EN ihEM | ⊗ hM |.

6
If we now evaluate the reduced statistical operator, taking into account the
orthogonality of the states of the environment we have:

ρ(Red,sys) ≡ T r(env) ρ(sys+env) = |α|2 |M i⊗|EM ihEM |⊗hM |+|β|2 |N i⊗|EN ihEN |⊗hN |.
(14)
Such a statistical operator is the one associated to the statistical mixture of
the states |M i and |N i with statistical weights |α|2 and |β|2 , a fact which would
render legitimate the replacement of the original pure state of the system with
the corresponding mixture:

|Ψ(sys) i = α|M i + β|N i ⇒ E{|M i, |N i; |α|2 , |β|2 }, (15)

with obvious meaning of the symbols.

The argument seems appealing and actually puts into evidence one relevant
fact, i.e., that to experimentally distinguish the pure state from the statistical
mixture is an extremely difficult task. But if one looks at it from the conceptual
point of view one immediately understands that it is seriously misleading for at
least two reasons:
• In the previous section we have proved that actually linear superpositions
of macroscopically different states occur and that their occurrence leads
to a deadlock. To replace them with a statistical mixture is a trick which
can have only a FAPP (for all practical purposes) validity - to use a spec-
ification due to Bell [5].
• There is also a much more serious objection to the now described game.
Most of the proponents of this solution to the puzzling aspects of linear
superpositions of macroscopically distinguishable states seem to ignore a
fundamental fact, i.e. that, within quantum mechanics, the correspon-
dence between statistical ensembles and statistical operators is infinitely
many to one. Thus, even if one would accept that the statistical operator
which must be used is the one of Eq.(14), one has no reason to interpret
it as describing the statistical ensemble at the r.h.s. of Eq.(15). In fact,
just to mention a trivial example, one immediately realizes that, e.g., also
the following statistical mixture of three pure quantum states:

1 1
E{ √ [|M i + |N i], √ [|M i − |N i], |M i; |β|2 , |β|2 , (|α|2 − |β|2 )}, (16)
2 2
corresponds precisely to the same statistical operator of Eq.(14), but it still
contains embarassing superpositions of macroscopically distinguishable states.
Actually, the decoherence approach to the measurement problem has been
repeatedly criticized (see, e.g. [6] ) and, moreover, the most serious proponents
of this solution have plainly recognized that the situation which we have outlined
is puzzling, and have proposed (vague) ways out from the puzzle. The most
paradigmatic example can be found in an important paper [7] by Joos and Zeh,
which clearly admidts the arbitrariness of choosing one statistical mixture (the
most reasonable from a perceptual point of view and the one corresponding to
wave packet reduction), among the infinitely many which are possible:

7
 • The local description is assumed and the specific choice of a basis can
perhaps be justified by a fundamental underivable assumption about the
local nature of the observer ... and his way of perceiving,
• No unitary treatment of the time dependence can explain why only one of
these dynamically independent components is experienced,
• The difficulty in giving a complete derivation of classical concepts may as
well signal the need of entirely novel concepts.

As a final remark we would like to stress that the most recent attempts to
achieve important technological improvements based on quantum mechanics -
such as Quantum Cryptography, Quantum Teleportation and Quantum Com-
putation - deal unavoidably with individual physical systems and they make a
systematic use of wave packet reduction at the individual level as an important
resource. Unless one ignores these fundamental aspects of modern research one
has to face the necessity of breaking the linear nature of quantum mechanics at
an appropriate stage.

5 The primitive ontology of a theory

Many scientists maintain that the purely technical, formal and logical aspects of
a theory represent all what deserves attention. We share with J S Bell and many
others the opinion that further requirements must be imposed to any theoretical
scheme to be considered as a fundamental account of natural processes. We do
not want to spend many words on this point; for a deep analysis we refer the
reader to a recent lucid paper [8] in which the demand for a richer elaboration
of the meaning of the formal scheme one is considering has been put forward.
The authors of this paper have stressed the necessity of equipping any theory
with what they call ‘the Primitive Ontology’ (PO) of the formalism. In brief,
the PO consists in the clear and precise specification of what the theory is
fundamentally about.
As already stated we will go quickly through this point, limiting ourselves
to illustrate it with reference to two theories, two proposals for the solution of
the macro-objectification problem which, at least at the nonrelativistic level,
are fully consistent. We have in mind Bohmian Mechanics and the Dynami-
cal Reduction Models, in particular the so-called GRW theory. Their different
status deserves to be stressed. In fact while Bohmian Mechanics is, by assump-
tion, predictively identical to quantum mechanics the GRW theory represents
a modification of the standard theory. They correspond to the two alternatives
which Bell [9] has indicated as the only two possible ways out of the macro-
objectification problem:
Either the wavefunction, as given by the Schrödinger equation, is
not everything, or it is not right.
Let us come to analyze these theories and to discuss their primitive ontolo-
gies.

8
5.1 Bohmian Mechanics
The formal aspects of the proposal put forward in his celebrated papers [10] by
D Bohm in 1952 can be summarized as follows:
• The quantum description of the state of a physical system is incomplete;
for a complete specification one has to add to the wavefunction also the
positions of all particles which constitute the system under examination
(the positions are the hidden variables of the theory because we are able
to prepare any desired wavefunction, but we have no way to control at
which point of the support of the wavefunction the particle actually is -
and the theory assumes that it has a precise position),
• The initial state of a system is fully specified by the assignment of the
initial wavefunction in configuration space Ψ(q1 , q2 , ..., qn , 0) ≡ Ψ(qi , 0),
and the initial positions Qi (0), (i = 1, 2, ..., n) actually occupied by the
constituents particles,
• There are two linked evolution equations in the theory. The wavefunction
evolves according to the standard quantum equation. Then one defines,
in terms of the wave function, a velocity field vB,k (Qj , t) for each particle
and considers a first order (in time) differential equation for the positions
themselves. In formulae:

∂Ψ(Qi , t) dQk (t)

ih̄ = HΨ(Qi , t); = vB,k (Qj , t); (17)
∂t dt
h̄ Im[Ψ∗ (Qj , t)∇k Ψ(Qj , t)]
vB,k (Qj , t) = .
mk |Ψ(Qj , t)|2

The theory exhibits some nice features. First of all, in virtue of the continuity
equation satisfied by Schrödinger’s wavefunction one immediately proves that if
one defines the position density distribution ρ(Qk , t) = |Ψ(Qk , t)|2 one has the
so called equivariance property:

ρ(Qk , 0) = |Ψ(Qk , 0)|2 ⇒ ρ(Qk , t) = |Ψ(Qk , t)|2 . (18)

This equation tells us that if one has an ensemble of particles whose initial posi-
tions are distributed according to the modulus square of the initial wavefunction,
then, by letting the particles evolve along the uniquely defined trajectories en-
suing from the differential equations for the positions, one recovers at any time
the position density distributions predicted by quantum mechanics.
The theory is purposedly equivalent to standard quantum mechanics but it
can be shown to imply that macroscopic objects end up, with the correct prob-
abilities, in those positions that are assigned to them by the standard theory
enriched with the reduction postulate. In short the theory solves the measure-
ment problem in a perfectly consistent way.
The primitive ontology of the theory emerges clearly from the given picture:
all particles of the Universe have at all times perfectly defined positions, they
move along precise trajectories in such a way to reproduce the position density
distribution of standard quantum mechanics. In the case of micro-macro inter-
actions, macroscopic systems end up, in turn, in the precise positions predicted

9
by quantum mechanics plus the wave packet reduction postulate. Obviously,
one has to assume that all measurements reduce essentially to position mea-
surements, just as it happens when we consider a system with a macroscopic
pointer, from the position of which we can infer the outcome of the measure-
ment. What the theory is about are the positions and they match perfectly
our perceptions concerning macroscopic systems. Using Bell’s terminology, the
positions are the beables of the theory, and everything is built up from them.

5.2 The dynamical reduction program

This attempt to account for all natural processes in terms of a unique dynam-
ical principle is based on two important remarks: first of all the characteristic
trait which makes the standard evolution and wave packet reduction radically
different is that while the first is linear and deterministic, the second one is
nonlinear and stochastic. Secondly, if one wants to have a mechanism leading
to the macro-objectification of some variable, one has to make a precise choice
about it since different quantities are associated to noncommuting operators and
one cannot ask that incompatible observables become simultaneously objective.
The most natural candidate, if one takes into account our definite perceptions,
are the positions, in accordance with the remark by Einstein [11]:
A macro-body must always have a quasi-sharply defined position in
the objective description of reality.
Accordingly, in the first model of this kind which has been proposed [12] (see
also the comprehensive review[13]) one adds to Schrödinger’s equation nonlinear
and stochastic terms affecting all elementary constituents of any system and
involving a universal localization mechanism. Formally one assumes that, at
random times, with an appropriate frequency λ ≈ 10−16 sec−1 , the wavefunction
suffers a process corresponding to a localization of one of the constituents. When
this process involves the i − th particle the wavefunction changes according to:
Λi (x)Ψ(q1 , q2 , ..., qn )
Ψ(q1 , q2 , ..., qn ) ⇒ (19)
k Λi (x)Ψ(q1 , q2 , ..., qn ) k
1 (qi −x)2
Λi (x) = 2 3/2
· e− 2σ2 ; σ ≈ 10−5 cm.
(2πσ )
A last specification concerns the positions at which collapses occur. It is embod-
ied in the assumption that the position x is chosen at random with a probability
distribution:
dP (x ∈ [x, x + dx]|Ψ(t), i) =k Λi (x)Ψ(q1 , q2 , ..., qn ) k2 dx. (20)
The model exhibits some physically interesting features:
• The spatial probability density of the localizations, according to Eq.(20),
practically coincides with the one which the standard theory attributes to
position measurements,
• It is immediately proved, by passing to the centre-of-mass and relative
coordinates, that the theory entails the so called Trigger Mechanism: the
localizations of the c.o.m. are amplified with the number of particles, actu-
ally of nucleons (because the frequency λ is assumed to be proportional to
the mass of the constituents), the value given above referring to nucleons,

10
 • According to this prescription a nucleon (or an atom) suffers a localization
about every 108 years, and, consequently, microscopic systems are prac-
tically unaffected, while a macroscopic object, containing an Avogadro’s
number of particles, suffers a localization about every 10−7 sec.,
• The statistical operator obeys an equation of the Quantum Dynamical
Semigroup type,
• The theory is, in principle, testable against quantum mechanics.
The conclusion should be obvious: the proposed universal dynamics leaves
all quantum predictions for microsystems unaltered but it accounts for wave
packet reduction with probabilities in agreement with the quantum ones and
for the classical behaviour of macroscopic systems, as well as for our definite
perceptions with them.
In the simplified version we have presented the theory suffers of a serious
limitation: the localizations break the symmetry character of the wavefunction
for identical particles. This feature is easily corrected as firstly shown in[14].
Before coming to consider the ontology of the model we mention that much
more formally elegant (even though physically equivalent) variants of the theory
have been presented [14, 15]. They are based on the formalism of stochastic
differential equations of the Ito or Stratonovich type.
Let us now come to discuss the PO of the theory. A first proposal has
been put forward by Bell himself in his presentation of the GRW theory at the
Imperial College meeting celebrating the centenary of Schrödinger’s [9]:

There is nothing in the theory but the wavefunction. ... However,

the GRW jumps (which are part of the wavefunction, not something
else) are well localized in ordinary space. Indeed each is centred on
a particular spacetime point (x,t). So we can propose these events
as the basis of the ‘local beables’ of the theory. These are the math-
ematical counterparts in the theory to real events at definite times
and places in the real world. ... A piece of matter then is a galaxy
of such events.

This assumption that ‘what the theory is fundamentally about’ are the local-
izations themselves has been characterized as ‘The flashes ontology’ in refs.[8]
and [16, 17, 18]. An alternative ontology has been proposed in [19] and has
been subsequently characterized as ‘the mass density ontology’. This assumes
that what the theory is about, what is real ‘out there’, is the mass density in
3-dimensional space:
N Z
X
m(x, t) = dr1 , ..., drN δ(ri − x)|Ψ(r1 , ..., rN , t)|2 . (21)
i=1 R3N

This concludes our description of two ways out from the macro-objectification
problem. As already mentioned there have been many other interesting attempts
aiming at the same result. We have concentrated our attention on these two
because they are precise models which, at the nonrelativistic level, account on
the basis of a unique universal dynamics both for the behavior of microscopic
systems, as well as for the reduction process which takes place during mea-
surement procedures. Bell has qualified them [5] as ‘exact’, a term by which

11
he intended which are precisely formulated and which neither require nor are
embarrassed by a conscious observer. Bohmian Mechanics has been discussed
many times and there are also entire books devoted to it. Concerning the GRW
theory there is an extensive literature (which can be found, e.g., in [13]) in which
many physically interesting aspects of the theory are analyzed.

6 The recent debate

In what follows we will be specifically interested in reviewing the new trends
in the foundational investigations about quantum mechanics. In spite of many
new proposals, we witness a revival of the old positions taken by the Copen-
hagen school in the thirties. In particular some scientists have expressed various
criticisms concerning the two approaches we have just discussed and their con-
sideration will allow us also to call attention to the new trends in the field.

6.1 The case of Bohm theory

A quite widespread critical attitude towards Bohmian Mechanics makes refer-
ence to the fact that since this theory is empirically indistinguishable from the
standard theory, it should be considered an example of ‘bad science’, a ‘degener-
ate research program’ in the sense of I Lakatos. To stress this position I cannot
do better than quoting a sentence from a letter written in 1966 by S Weinberg
to S Goldstein which can be found in ref.[20]:
At the regular weekly luncheon meeting today ot our Theory Group,
I asked my colleagues what they think of Bohm’s version of quantum
mechanics. The answers were pretty uniform and much what I would
have said myself. First, as we understand it, Bohm’s quantum me-
chanics uses the same formalism as ordinary quantum mechanics,
including a wavefunction that satisfies the Schrödinger equation, but
adds an extra element, the particle trajectory. The predictions of the
theory are the same as for ordinary quantum mechanics, so, there
seems little point in the extra complication, except to satisfy some
a priori ideas about what a physical theory should be like... In any
case, the basic reason for not paying attention to the Bohm approach
is not some sort of ideological rigidity, but much simpler – it is just
that we are all too busy with our own work to spend time on some-
thing that doesn’ t seem likely to help us make progress with our real
problems.
In our opinion it is remarkable that various (presumably) high level scientists
and even a giant like S Weinberg do not consider important at least to mention
the puzzling aspects of the standard interpretation of the theory, the aspects
for which Bohm has been led to work out his proposal. Without paying the due
attention to this fundamental point one cannot appreciate the great merit of
the approach, i.e., the fact that it consistently solves the measurement problem
and leads to the classical behaviour of macroscopic objects, a feature which is
by no means trivial.
At any rate, if the predictive equivalence of the two theories is considered
as a drawback, people interested in foundational issues should not ignore that

12
the other ‘exact’ theory we have presented qualifies itself as a rival theory of
quantum mechanics which, in principle, can be tested against it. So, let us
pass to analyze some of the critical remarks adressing the dynamical reduction
program.

6.2 The dynamical reduction models

As already remarked, it is often claimed that one can very well adopt the or-
thodox interpretation of quantum theory provided one takes into account the
decoherence due to the environment. Typically, J Bub, in describing what he
calls ‘the new orthodoxy’, claims [21] that the superpositions of macroscopically
different states actually occur but we do not see them due to the coupling with
the environment. We have already discussed this aspect and we believe to have
proved that, even though one might adopt it FAPP, one cannot avoid facing
it when one is interested in the internal consistency of the theory. Moreover,
invoking decoherence leads to consider practically only the statistical aspects of
the theory: in a sense one claims that, when dealing with ensembles, one can
use the reduced statistical operator obtained by disregarding the environmen-
tal degrees of freedom. But, as we have already remarked, modern technology
deals more and more with individual physical systems and only if one resorts
to individual reduction processes one can go on. We have the impression that
the attitude of many of the adherents to the ‘decoherence’ approach make an
a priori and vague choice aimed to render nonrelativistic quantum mechanics a
theory which, from an experimental point of view, cannot be caught in telling
a lie3 .
Such, in our opinion, not well pondered attitude emerges also from a formal
aspect to which these scientists make often reference: the so-called ‘ancilla ar-
gument’. It is well known that any quantum evolution equation of the quantum
dynamical semigroup type for the statistical operator is physically equivalent to
a quantum mechanical theory with a unitary dynamics provided one considers
an enlarged Hilbert space. The game is quite simple: one introduces a new
quantum ancilla whose degrees of freedom are unaccessible. Then one cooks
up a unitary dynamics in the Hilbert space System ⊗ Ancilla such that by
partial tracing on the degrees of freedom of the ancilla one recovers the evo-
lution equation for the statistical operator of a dynamical reduction model of
the type we have considered. A natural question arises: what is the purpose of
this position besides the desire to protect the standard formalism come what
may? The ancilla has, by construction, no observable effect and the procedure
amounts simply to introduce hidden elements whose only role is to save the
formal structure of the standard theory.
In view of the fact that dynamical reduction theories qualify themselves as
rival theories of quantum mechanics, would it not be more serious, scientifically,
to try, as R Penrose [22, 23, 24], S Adler [25, 26, 27] and many others do, to
investigate the possibility of performing some crucial test? I do not intend to
suggest that the GRW theory has to be taken seriously in its present formulation
as a fundamental theory of natural processes, but I cannot avoid calling attention
to the fact that, being a precise theory, it can give some indications about where
to look for possible violations of the linearity of quantum mechanics.
3 For a brilliant discussion of the arguments presented in this subsection we refer the reader

to ref.[20].

13
7 Relativistic generalizations
Bell in one of his last papers, after having discussed Bohmian Mechanics and
the GRW theory has stated [5]:
The real problem now is which one of these two exact theories admits
a relativistic generalization

Concerning this fundamental problem it is useful to mention that Bohmian Me-

chanics admits relativistic generalizations of various kinds, from the original
attempts of Bohm and Hiley [28] based on a preferred Lorentz frame, to the
recent investigations by Dürr et al. [29] and Salmos [30], resorting to an arbi-
trary preferred space-time slicing. Berndl et al. [31] and Dewdney and Horton
[32] suggest a preferred joint parametrization of the world lines. We are not
interested here in the technical details of these approaches, what is relevant is
that they are characterized by a general feature which turns out to be unavoid-
able: they are not ‘genuinely invariant’ in the precise sense that they require a
(hidden) preferred reference frame. This is a consequence of the fact that [33]
any theory which violates Bell’s locality condition by violating the parameter
independence requirement, does not admit a ‘genuine’ relativistic generalization.
There have been also various attempts at a relativistic generalization of dy-
namical reduction models. The first is due to P Pearle [34] and it has been
discussed in detail and proved to be formally perfectly Lorentz invariant in
ref.[35]. The idea is quite simple but deep: one considers a fermion field which
is coupled to a meson field and introduces a reduction mechanism which for-
bids the superpositions of different mesonic states. Since the mesonic clouds
associated to different positions of a fermion differ, one induces in this way an
indirect localization of the fermions. However, a new problem arises: the intro-
duction of stochastic processes in a relativistic context leads to the emergence
of untractable divergences.
Other attempts should be mentioned, among them those of Dove and Squires
[36], of Dowker et al. [37, 38, 39], which are formulated on a discrete space-time.
Up to very recent times no real step forward has been done. In the year 2006,
R Tumulka has presented [16, 17, 18] a relativistic generalization of the GRW
theory for N noninteracting distinguishable fermions based on the consideration
of a multi-time Dirac equation. The theory sticks to what we have called the
flashes ontology4. It is particularly interesting to mention Tumulka’s conclusions
which he has reinforced recently [17]:
A somewhat surprising feature of the present situation is that we
seem to arrive at the following alternative: Bohmian mechanics
shows that one can explain quantum mechanics, exactly and com-
pletely, if one is willing to pay with using a preferred slicing of space-
time; our model suggests that one should be able to avoid a preferred
slicing if one is willing to pay with a certain deviation from quantum
mechanics.
4 Models based on the mass-density ontology require a serious reconsideration before one

can resort to them for relativistic generalizations.

14
8 Quantum computation, nonlocality and real-
ism
Quite recently, an attitude which we see as an attempt to revive the Copenhagen
position emerged in connection with two quite different subjects, i.e., on the
one side, the investigations concerning the promising and interesting field of
quantum computation and, on the other, some theoretical and experimental
analyses of quantum nonlocality. Let us proceed to discuss them.
People working in quantum computation have repeatedly put forward the
idea that what quantum mechanics is about is not ‘something existing out there’,
but only and exclusively ‘our information’. A paradigmatic example of this
attitude appears in a paper by A Zeilinger [40]:
• The distinction between reality and our knowledge of reality, between reality
and information, cannot be made,
• Bell’s result suggests that the concept of reality itself is at stake,
• There is no way to refer to reality without using the information we have
about it.
Leaving apart the tautological structure of the last sentence, I believe that
the best response to the first has been given in a recent paper [41]:
The distinction between reality and our knowledge of reality, not only
can be made; it must be made if the notions of knowledge and infor-
mation are to have any meaning in the first place.
To further illustrate the discussion about the recent positions taken by some
scientists we consider it appropriate to mention a recent exchange of views
between D Mermin and the present author [42, 43]. Mermin asserts that the
present status of quantum computation is completely satisfactory on the basis
of the fact that:
• ... all gates, including the measurement gate, alter the state associated
with the incoming Q-bits in a well defined, generally discontinuous manner
which is precisely defined by the state,
• ... in spite of the fact that there remains the distinction between linear,
invertible, unitary gates with output state fully determined by the imput,
and nonlinear, irreversible, measurement gates, with output state stochas-
tically determined by the input, ..., the action of both types of gates are
fully determined, with nothing left to the discretion of the theoretical physi-
cist.
We perfectly agree on these statements, but we stress that they amount
simply to a way of evading the real problem. In fact Mermin himself stresses
that:
it is through the readings of 1-Qbit measurement gates, and only
through such readings, that one can extrect information from a quan-
tum computer ... measurement is essential at both ends of a quantum
computation. ... the ‘user’ which looks at a visual display or a print
out [plays the fundamental role in the process].

15
 Here one can see, on the one side, the plain recognition that the very practice
of quantum information and computation requires to accept the validity of two
different and incompatible dynamical principles. On the other side, one ignores
completely the fact that while we know everything about the working of the
unitary gates used to implement our algoritms, we have an extremely vague idea
of why and when individual physical processes occur which are not governed
by the (usually) assumed universal laws of quantum mechanics and actually
contradict it. This is, in essence the criticism to Mermin that we have raised.
He has replied [44] and, besides making various remarks he has made quite clear
his position:
I stress that the state and the unitary transformations it is subject
to are mathematical abstractions that enable one to compute ... the
probabilities of the readings of the final measurement gates. Math-
ematical abstractions do not require stochastic ‘hits’ originating in
unknown physical processes or interactions with gravitons to be reset;
they are reset by us, when we acquire more information and want to
calculate what we expect to experience next.
This sentence makes explicit reference to the GRW theory and to the pro-
posals by Penrose of relating wave packet reduction to gravity. We believe that
one cannot ignore that these alternative theories represent serious attempts to
account consistently, in physical terms and not by vague verbal assumptions,
for what ‘we will experience ... next’.
To conclude this first part of our comments we would like to mention that the
idea that quantum mechanics deals simply with ‘information’, is not new. For
instance it was put forward during the discussions concerning the decoherent
histories approach to quantum mechanics when it has been suggested to look at
them exclusivively with reference to IGUSes (information gathering and using
systems). Bell has considered this position and he has made clear [5] that he
was inclined to reject any reference to information unless one would, first of
all, answer to the following basic questions: Whose information?, Information
about what?
This remark allows to focus in a precise way the two incompatible positions
which characterize the present debate which, in our opinion, echoes the one of
more than 60 years ago. In fact, e.g., Mermin himself, has recently taken a quite
precise position [45] concerning the issue under discussion:

The question of ‘information about what’ is a fundamentally meta-

physical question that ought not to distract tough-minded physicists.
This concludes our first set of remarks.
We come now to the second point, i.e., to the recent debate concerning non-
locality. An anticipation of this attitude is already contained [40] in the second
sentence reported at the beginning of this section. The story goes as follows:
A Leggett has considered [46] a class of theories, referred to as nonlocal realis-
tic theories, which exhibit nonlocal features but satisfy a particularly natural
(in his opinion) set of conditions. Leggett himself has shown that, in spite of
the fact that one could not derive a Bell’s type of inequality for such theories,
they were at variance with the predictions of quantum mechanics and had to
be abandoned. Subsequently various authors [47, 48, 49, 50] have considered

16
the same problem both with reference to Leggett’s proposal as well as for a
somewhat extended class of nonlocal theories and have proved once more their
incompatibility with quantum mechanics. Even experimental tests of Leggett’s
model have been performed, confirming the crash with quantum predictions.
This line of research is, in our opinion, quite interesting and deserves to be
pursued. However, the above facts - i.e. the supposed ‘naturalness’ of the re-
quests and the persisting incompatibility with quantum mechanics - have given
rise to a very peculiar type of argument. In a sketchy way it can be summarized
as follows: the derivation of Bell’s inequality is based on two assumptions: Lo-
cality and Realism. Accordingly, the experimental violation of Bell’s inequality
requires to abandon at least one of these assumptions. Since giving up locality
along the ‘natural’ way suggested by Leggett does not eliminate the crash of
such proposals with quantum theory, the experimentally tested incompatibility
gives a clear indication that one has to give up the assumption of realism. This
argument, combined with the already emerging opinion, coming from the quan-
tum computational community, i.e. that the concept of reality itself is at stake,
has seen the adoption of this point of view by many, even extremely brilliant
physicists [40, 47, 51].
We would like to make some remarks on this point. First of all we will call
attention on the factual irrelevance of the argument. Secondly we will stress,
and we consider this an extremely important point, that the just mentioned
papers take a completely mistaken attitude towards the real meaning of Bell’s
work. Let us proceed to present the just mentioned criticisms.

• The class of models which have been considered is quite restricted, in

particular it embodies a sort of weak locality request, which is in no way
compelling. This point has been already raised in an important paper
[50].
• There is no doubt that there exist theories which can legitimately be qual-
ified as ‘realistic’ and fully agree with the standard theory. We mention
first of all Bohmian Mechanics and, secondly, a quite elementary toy ex-
ample dealing with a system of two entangled spin 1/2 particles in the
singlet state which has been devised in 1964 by Bell in his fundamental
paper [1].

In agreement with these remarks one cannot claim that the program of
building up a ‘fully realistic nonlocal theory’ equivalent to quantum mechanics
is not viable, an important counterexample of what has been suggested in the
above mentioned papers. But this remark is by no way the most relevant one,
actually it turns out to be rather marginal. What really matters is the fact that
the derivation of Bell’s inequality in no way whatsoever needs an assumption of
realism. In spite of this fundamental fact, which everybody can verify by going
carefully through the proof, a large part of the scientific community shares the
completely wrong opinion that realism is among the basic assumption needed
for the derivation of Bell’s result. Bell himself has stressed this aspect and has
remarked [52] that it is extremely difficult to eradicate this prejudice5 :
5 It has to be remarked that deterministic hidden variable theories assume that the com-

plete specification of the state of the system implies that all physical properties are actually
possessed by the systems prior to any measurement process. This is equivalent to the request
of realism discussed by the above mentioned authors

17
 My own first paper (Physics 1, 195 (1965.) on this subject starts
with a summary of the EPR argument from locality to determin-
istic hidden variables. But the commentators have almost universally
reported that it begins with deterministic hidden variables.
This being the situation we must conclude that in no way whatsoever Bell’s
inequality has something to do with realism. It simply identifies in a straightfor-
ward and lucid way that what quantum phenomena impose to us is to accept the
unescapable fact that natural processes involving entangled states of composite
and far-away systems turn out to be unavoidably non-local. If one is keen to
consider the possibility of giving up realism he can very well try his luck, but for
sure such a big price is not necessary, and even not suggested, by the quantum
nonlocal correlations.
The question whether Bell’s theorem involves some request of realism has
been discussed and proven once more to be unjustified by various authors [53,
54, 55].
We would like to conclude this analysis of some of the most relevant issues of
the foundational investigations on quantum mechanics by expressing our worries
concerning the fact that there are some indications that 80 years of lively debate
on the conceptual problems of our best theory by the most brilliant physicists
of the last century are facing a serious risk of being cancelled by those scientists
who Mermin has qualified as the Q-computation crowd or by those who derive
from experimental results inspired by not strictly convincing theoretical models
unjustified conclusions concerning such an important issue as the one of the
reality of the world around us.

References
References
[1] Bell J S 1964 Physics 1 195
[2] Bell J S 1989 Themes in Contemporary Physics II ed S. Deser and R.J.
Finkelstein (Singapore: World Scientific) p 1
[3] Primas H 1991 Symposium on the foundations of modern physics 1990, Dis-
cussion Sections ed K.V. Laurikainen and J. Viri (University of Turku, Fin-
land)
[4] Bassi A and Ghirardi G C 2000 Phys. Lett. A 275 373
[5] Bell J S 1990 Phys. World 3 33
[6] Adler S 2003 Stud. Hist. Phil. Mod. Phys. 34 135
[7] Joos E and Zeh H D 1985 Z. Phys. B 59, 223
[8] Allori V, Goldstein S, Tumulka R and Zanghı́ N 2008 British J. Philos. Sci.
59 353
[9] Bell J S 1987 Schrödinger–Centenary Celebration of a Polymath ed C W
Kilmister (Cambridge: Cambridge University Press)

18
[10] Bohm D 1952 Phys. Rev. 85 166, 1952 Phys. Rev. 85 180
[11] Pauli W in Born M 1971 The Born Einstein Letters (New York: Walter
and Co)
[12] Ghirardi G C, Rimini A and Weber T 1986 Phys. Rev. D 34 470
[13] Ghirardi G C and Bassi A 2003 Phys. Rep. 379 257
[14] Pearle P 1999 Phys. Rev. A 59 80
[15] Ghirardi G C, Pearle P and Rimini A 1990 Phys. Rev. A 42 78
[16] Tumulka R 2006 AIP Conference Proceedings 844 ed A. Bassi et al
(Melville, New York: AIP) p 340
[17] Tumulka R 2006 Proc. Roy. Soc. London A 462 1897
[18] Tumulka R 2006 J. Stat. Phys. 125 821
[19] Ghirardi G C, Grassi R and Benatti F 1995 Found. Phys. 25 5
[20] Hagar A 2007 Stud. Hist. Philos. Mod. Phys. 38 906
[21] Bub J 1997 Interpreting the Quantum world (Cambridge: Cambridge Uni-
versity Press)
[22] Penrose R 1996 Gen. Relativ. Gravit. 28 581
[23] Penrose 2000 Mathematical Physics 2000 (London: Imperial College)
[24] Marshall W, Simon C, Penrose R and Bouwmeester D 2003 Phys. Rev.
Lett. 91 130401

[25] Adler S 2005 J. Phys. A 38 2729

[26] Adler S, Bassi A and Ippoliti E 2005 J. Phys. A 38 2715
[27] Bassi A, Adler S and Ippoliti E 2005 Phys. Rev. Lett. 9 030401
[28] Bohm D and Hiley B J 1993 The Undivided Universe: an Ontological In-
terpretation of Quantum Theory (London: Routledge)
[29] Dürr D, Goldstein S, Münch-Berndl K and Zanghı́ N 1999 Phys. Rev. A
60 2729
[30] Salmos T M 1995 J. Stat. Phys. 80 793, 1996 Bohmian Mechanics and
Quantum Theory: an Appraisal ed J.T. Cushing et al. (Dordrecht: Kluwer)
p 191
[31] Berndl K, Dürr D, Goldstein S and Zanghı́ N 1996 Phys. Rev. A 53 2062
[32] Dewdney C and Horton G 2001 J. Phys. A 34 9871
[33] Ghirardi G C and Grassi R 1996 Bohmian Mechanics and Quantum Theory:
an Appraisal ed J.T. Cushing et al. (Dordrecht: Kluwer) p 353

19
[34] Pearle P 1990 Sixty-two Years of Uncertainty ed A. Miller (New York:
Plenum)
[35] Ghirardi G C, Grassi R and Pearle P 1990 Found. Phys. 20 1271
[36] Dove C and Squires E 1995 Found. Phys. 25 1267
[37] Dowker F and Henson J 2004 J. Stat. Phys. 115 1327
[38] Dowker F and Herbauts I 2004 Classical Quant. Grav. 21, 2963
[39] Dowker F and Herbauts I 2005 Found. Phys. Lett. 18, 499
[40] Zeilinger A 2005 Nature 438 743
[41] Daumer M, Dürr D , Goldstein S, Maudlin T and Tumulka R 2006 AIP
Conference Proceedings 844 ed A. Bassi et al (Melville, New York: AIP) p
129
[42] Mermin D 2006 Quantum Inf. Processes 35 239
[43] Ghirardi G C 2008 Found. Phys. 38 1011
[44] Mermin D 2008 arXiv:0808.1582
[45] Mermin D 2002 Quantum [Un]speakables ed R.A. Bertlmann and A.
Zeilinger (Berlin: Springer) p 271
[46] Leggett A J 2003 Found. Phys.33 1496
[47] Gröblacher S, Paterek T, Kaltenbaek R, Bruckner C, Zukowski M, As-
pelmeyer M and Zeilinger A 2007 Nature 446 871
[48] Branciard C, Ling A, Gisin N, Kurtsiefer C, Lamas-Linares A and Scarani
V 2008 Nature Physics 4 681
[49] Paterek T, Fedrizzi A, Gröblacher S, Jennewein T, Zukovski M, Aspelmeyer
M and Zeilinger A 2007 Phys. Rev. Lett. 99 210406
[50] Colbeck R and Renner R 2008 Phys. Rev. Lett. 101 050403
[51] Aspect A 2005 Nature 438 745

[52] Bell J S 1981 Journal di Physique C2 42 41

[53] Norsen T 2006 Found. Phys. Lett, 19 633, 2007 Found. Phys. 37 311
[54] Tausk D V 2008 arXiv:0809.4000
[55] Laudisa 2008 Found. Phys. 38 1110

20