Sei sulla pagina 1di 13
Studies in History and Philosophy of Modern Physics xxx (2017) 1 e 13 Contents lists
History and Philosophy of Modern Physics xxx (2017) 1 e 13 Contents lists available at ScienceDirect

Contents lists available at ScienceDirect

Studies in History and Philosophy of Modern Physics

journal homepage: www.elsevier.com/locate/shpsb

Physics journal homepage: www.elsevier.com/locate/shpsb Are fi eld quanta real objects? Some remarks on the

Are eld quanta real objects? Some remarks on the ontology of quantum eld theory

Tomasz Bigaj

Institute of Philosophy, University of Warsaw, ul. Krakowskie Przedmiescie 3, 00-047 Warsaw, Poland

article info

Article history:

Received 16 April 2017 Received in revised form 29 June 2017 Accepted 9 August 2017 Available online xxx

Keywords:

Quantum eld theory Fock space Harmonic oscillators Field quantization Particles Ontology

abstract

One of the key philosophical questions regarding quantum eld theory is whether it should be given a particle or eld interpretation. The particle interpretation of QFT is commonly viewed as being under- mined by the well-known no-go results, such as the Malament, Reeh-Schlieder and Hegerfeldt theorems. These theorems all focus on the localizability problem within the relativistic framework. In this paper I would like to go back to the basics and ask the simple-minded question of how the notion of quanta appears in the standard procedure of eld quantization, starting with the elementary case of the nite numbers of harmonic oscillators, and proceeding to the more realistic scenario of continuous elds with in nitely many degrees of freedom. I will try to argue that the way the standard formalism introduces the talk of eld quanta does not justify treating them as particle-like objects with well-de ned properties.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction: against the particle interpretation of QFT

Quantum eld theory (QFT) is to date the most successful theory of matter and its interactions at the fundamental level. Historically, QFT appeared on the scene as one way of extending non-relativistic quantum mechanics in response to certain shortcomings and lim- itations of the latter. One particular problem that the early quantum physicists grappled with was how to properly describe the pro- cesses of absorption and emissions of electromagnetic energy by atoms. 1 An adequate analysis of these phenomena requires that we have a theory that is both relativistic (since photons travel at the speed of light) and capable of providing a quantum description of classical elds (since absorption and emission involve electro- magnetic elds and their interaction with matter). QFT aimed to satisfy these two demands. Famously, the second postulate was ful lled by adopting the now standard procedure known as eld quantization. The rst undeniable success of the QFT program and its method of quantization was the development of a quantum theory of electromagnetism, called quantum electrodynamics (QED). Subsequent applications of the mathematical formalism of

E-mail address: t.f.bigaj@uw.edu.pl . 1 See ( Kuhlmann, 2010 , pp. 27 e31) for an overview of this historical episode. Other, more comprehensive historical introductions to QFT can be found in ( Cao, 1997 ) and ( Auyang, 1995 ).

QFT yielded theories of weak and strong (nuclear) interactions (Yang-Mills gauge theories, Glashow-Salam-Weinberg theory of electroweak interactions, and quantum chromodynamics). In spite of its mounting conceptual and technical dif culties, of which the problem with in nities and various renormalization techniques is but one example, QFT is still considered the best working theory of fundamental interactions there is. From a philosophical perspective, one of the most pressing questions regarding QFT is the question of what exactly it tells us about the nature of the constituent elements of reality. What is the proper ontological interpretation of QFT? As its name and origins clearly suggest, the theory seems to deal primarily with various elds (electromagnetic, strong, weak), hence an immediate conclusion may be that according to QFT the most fundamental physical entities are elds. And yet the procedure of eld quanti- zation, which is an indispensable part of QFT, results in a partic- ularization or discretization of elds in the form of interaction carriers (photons, gluons, gauge bosons). Thus it may be surmised that the basic lesson from QFT is that ultimately there are only particles and void. For many working physicists, as well as some philosophers, it is an unquestionable truism that QFT deals pri- marily with particles. In support of this claim we can recall the historical development of the concept of quanta from an ad hoc hypothesis postulated by Planck in order to solve the problem of black body radiation to the corpuscular interpretation of electro- magnetic radiation advanced by Einstein and con rmed in

2 T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1

2

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

experiments (photoelectric effect, Compton scattering). Moreover, it may be pointed out that QFT is a theoretical frame in which the Standard Model is formulated, and the Standard Model is cashed out in terms of elementary particles and their interactions. Simplifying things a bit, we may be tempted to say that because QFT applies to objects such as electrons, photons, gluons and the like,

and because these objects are commonly referred to as particles , ontologically speaking QFT is a theory of particles. However, the argument from scienti c practice for the particle interpretation of QFT is very weak. The fact that physicists label certain elements described by their theories particles does not imply that from an ontological point of view these elements indeed deserve to be categorized as such. For all we know, electrons may turn out to be some aspects or manifestations ( epiphenomena ) of an underlying entity, which may be fundamentally non-particle- like. As a matter of fact, even the founders of QFT were not unan- imous in their opinions regarding the proper ontological interpre-

tation of this theory. For instance, Schr odinger maintained that fundamental physical reality consists of waves only, whereas so- called particles should be rejected, since they lack features such as identi ability over time or distinguishability. Other physicists were more sympathetic towards the standard concept of a particle though. When in 1927 Dirac introduced the procedure of eld quantization and applied it to the case of electromagnetic radiation, he interpreted it as showing conclusively that light quanta are in- dependent, fundamental substances, while the wave function is a mere calculational device with no deeper ontological meaning. However, Dirac's argument can be questioned. As Cao (1997, pp. 160 e165) observes, Dirac's reasoning is based on the confusion of second quantization (quantization of the wave function of a system of particles) with eld quantization (quantization of a real, physical eld, such as the electromagnetic eld). The philo- sophical controversy between the particle and eld interpretations of QFT continues to this day, with numerous arguments presented both for and against each of these two rival conceptions. In what follows we will brie y discuss some of the best-known arguments against the particle interpretation of QFT. A prominent group of arguments of this type are based on several no-go results provable in some variants of QFT. These re- sults typically show that certain common-sense conditions that we usually associate with the concept of particles, such as the assumption of localizability, cannot be jointly satis ed within the broad framework of QFT. To begin with, we have Malament's the- orem, which is not associated with any concrete variant of QFT but instead proceeds from some general assumptions that should be satis ed in all relativistic quantum theories of localizable particles, and yet can be shown to lead to a contradiction. 2 These assump- tions are: localizability, microcausality, translation covariance, and the assumption that the energy is bounded from below. Localiz- ability prescribes that the probability of nding an object simul- taneously in two disjoint regions should be zero (perhaps this postulate could be alternatively and more appropriately labeled no-bilocation ). Mathematically, this condition is expressed in the requirement that projectors representing propositions of the form a particle is localized in a given region should be orthogonal if these regions are disjoint subsets of a hyperplane of simultaneity. Microcausality, in turn, encompasses the relativistic intuition that detection measurements performed in one region cannot instan- taneously affect the statistics of detection measurements in a distant region. This can be spelled out in the form of the condition

that for two disjoint regions D and D 0 on a hyperplane of

2 See ( Malament, 1996 ). The brief presentation of Malament's theorem given above is based on ( Halvorson & Clifton, 2002 ).

simultaneity there is a time interval ε such that if we performed a time-like translation on region D 0 by an amount t smaller than ε , then projectors E D and E D 0 þ t corresponding to appropriate regions should commute. Finally, translation covariance means that trans- lations of the underlying spacetime are represented by a contin- uous group of unitary operators in the sense that applying a translation a to a region D is equivalent to applying the corre- sponding unitary operator U ( a ) to the projector E D . A closely related result is known under the name of Hegerfeldt's theorem. 3 In addition to the energy bounded below and the time- translation covariance from Malament's theorem, it formulates two new conditions that replace localizability and microcausality:

monotonicity and no instantaneous wave spreading. Monotonicity asserts that if we have a nested family of subsets of an element of a foliation of spacetime, and the intersection of this family gives re- gion D, then the intersection of corresponding projectors should

give E D . The most signi cant premise of the theorem states that if a

particle is localized at a given moment in region D, then for any region D 0 such that D 4 D 0 (with the assumption that the bound- aries of D and D 0 are separated by a nite distance) there is an in- terval ε such that for all time-like translations t smaller than ε it is certain that the particle can't be found outside the translated region D 0 þ t (the projector E D is included in the projector E D 0 þ t ). The theorem proves that given these assumptions it follows that any arbitrary time-like translation of E D gives back E D , which means that the particle will forever stay in the same region. By trans- position, if any movement of the particle is possible at all, it must be in the form of an instantaneous spreading violating the principles of relativity. Both theorems sketched above show that within QFT there is a serious problem with the concept of a localizable particle obeying the principles of relativity. However, it is open to debate whether this fact shows the untenability of the particle interpretation of QFT, 4 or rather that any quantum theory is ultimately at odds with some fundamental relativistic intuitions (and, therefore, we have yet to come up with a new theory that could successfully incor- porate both quantum and relativistic aspects of reality, for instance quantum gravity). But the list of arguments against the particle interpretation of quanta is not exhausted yet. Another key result is known as the Reeh-Schlieder theorem, provable within so-called algebraic quantum eld theory (AQFT). Algebraic QFT is a mathe- matical approach based on the assumption that each spatiotem- poral region is associated with an algebra of operators representing possible measurements within this region. 5 AQFT is usually pre- sented in the form of several mathematical axioms (and thus the acronym AQFT is alternatively expanded as axiomatic quantum eld theory), of which the axiom of locality is the most recognizable from the physical point of view. Locality, also known as causality, asserts that observables associated with space-like separated re- gions commute (more precisely, that their commutators in the case of bosons and the anti-commutators in the case of fermions are zero). It is also assumed that the algebra of operators associated with a subset of a given region O is included in the algebra corre- sponding to O (isotony). Another important axiom ensures the existence of a unique state that is invariant under all Poincar e transformations. This state is known as the physical vacuum.

3 See ( Hegerfeldt, 1974, 1998 ). Again, we will be following Halvorson and Clifton (2002) in the exposition of the theorem.

4 In their analysis of the consequences of Malament's theorem, Fleming and Butter eld (1999) argue that the failure of the assumption of microcausality can be reconciled with the principles of relativity. For an extensive response to this strategy of defending the notion of localizable particles in QFT see ( Halvorson,

2001 ).

5 See ( Haag, 1996 ) for an introduction to AQFT.

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1 e

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

3

The Reeh-Schlieder theorem states that for any bounded, open region O (where boundedness is de ned with respect to a given metric by the condition that no two points in the region are separated by a distance greater than a given number, and open regions are de ned in the standard topological way as sets which contain a ball around each of their points), the vacuum state U is cyclic with respect to the algebra of observables ( O ) on O . 6 The property of cyclicity means that any state in the relevant Hilbert space can be approximated by the result of an action of an element from ( O ) on U. An important corollary of the Reeh-Schlieder theorem is that if the result of an action of an operator from ( O ) on U equals zero, then the operator itself is the zero operator. This fact immediately shows that the local algebra ( O ) cannot contain any projectors on subspaces orthogonal to U, as acting with such a projector on U will yield 0, and yet the projector is not zero. But it is easy to observe that all states with a non-zero number of particles are orthogonal to U. In conclusion, we have to admit that the pro- jection operator P N representing the state with N particles does not belong to the local algebra ( O ), and thus it is impossible to perform a local measurement telling us that there are exactly N particles. Again, as with the previous two no-go results, the culprit seems to be the concept of a localizable particle. As it is impossible to use local measurements in order to verify the number of parti- cles, it seems that these particles can never be forcedto occupy a given bounded region of space-time. 7 Let us nally mention an argument against the particle inter- pretation that is formulated within the framework of interacting elds. As is well known, in the case of quantum systems with nite numbers of degrees of freedom all Hilbert space representations of the commutation relations holding in these systems are unitarily equivalent (the Stone e von Neumann theorem). However, this result does not generalize for the cases of in nitely many degrees of freedom, to which the case of quantum elds belongs. In particular,

it can be proven (Haag's theorem) that if the Lagrangian describing

a given eld contains a component representing interactions, then

the representations associated with different values of the strength of these interactions will be inequivalent ( Haag, 1996 , pp. 53 e 55). The reasoning showing that this is the case is in fact rather simple. Let us consider two unequal values of the coupling constant g representing the strength of interactions. For each g there is a unique state U g which is the lowest eigenstate of the Hamiltonian, and thus it is the vacuum state. And quite clearly different values of g produce distinct states U g , since the Hamiltonian depends non- trivially on g . But if the representations of the commutation re- lations between the strength of the eld f( x ) and its canonical

momentum p( x ) were equivalent for different values of g , then this would show that actually the vacuum states for these values are identical. This follows from the fact that the vacua are invariant

6 See ( Reeh & Schlieder, 1961 ). The philosophical implications of the theorem are thoroughly discussed in ( Redhead, 1995 ). See also ( Kuhlmann, 2010 , pp. 99 e106), where he presents and compares the Malament and Reeh-Schlieder theorems.

7 However, one may wonder if the Reeh-Schlieder theorem actually proves too much. We may reluctantly agree with the conclusion that local measurements can never tell us the exact number of all particles in the universe (in Redhead's words it is not a local question to ask are we in an N -particle state? ”” , Redhead, 1995 , p. 127) However, detecting even one particle in a laboratory constitutes a measure- ment that tells us that the universe is in a state orthogonal to the vacuum state (or, more precisely, that the state of the universe lies in the subspace spanned by vectors corresponding to states with N ¼ 1, N ¼ 2, N ¼ 3, . particles). According to the Reeh-Schlieder theorem this measurement cannot be represented by an element of a local algebra of operators. This surprising fact may be interpreted as showing that the abstract notion of a local observable as introduced in AQFT does not fully coincide with the experimental concept of local measurements. After all, detecting particles in physical laboratories is a common occurrence, and it is hard to see why such a procedure does not deserve to be called local.

under spatial translations, and operators representing translations are not directly dependent on g . Thus if operators of translations were parts of the same Hilbert space representation for different values of g , all vacuum states U g would have to be identical. The signi cance of this result for the particle interpretation of QFT is as follows. If the coupling constant g ¼ 0, this means that we are dealing with the non-interacting case. And it is known that in this case there exists an important representation known as a Fock space representation. 8 As we will see later in the text, the standard assumption is that a Fock space representation of a quantum eld consists of states with well-de ned numbers of particles. But now we can observe that Haag's theorem implies that representations for all non-zero values of g will be unitarily inequivalent, and thus no Fock space representation can exist for the case of interacting elds. This means that in this case the number of particles is not a well-de ned parameter. 9 The best we can do is to consider the concept of numbers of particles as an asymptotic notion when the strength of interactions goes to zero (the emergence of particles at long distances when they no longer interact with one another). The arguments considered in this section are formidable, but perhaps not utterly devastating for the particle interpretation of QFT, if we adopt some of the evasive strategies mentioned already in passing. It may for instance be insisted that some vestiges of particularity can be retained in QFT, in the form of an asymptotic notion of particles 10 whose localizability is seriously limited and whose exact numbers may be sometimes dif cult if not impossible to properly (i.e. locally) measure. 11 However, in the next section we

8 We discuss Fock representations in greater detail in section 4 .

9 This is a slightly simpli ed account of Haag's theorem and its philosophical implications. As Earman and Fraser point out, generally speaking Haag's theorem does not exclude the existence of Fock representations in the case of interacting elds; it only implies that if such a Fock representation exists, it has to be unitarily inequivalent to the Fock representation of free elds ( Earman & Fraser, 2006 , pp. 326 e 330). They suggest that the incorrect interpretation of Haag's theorem is a result of the equivocation affecting the notion of a myrioticrepresentation. A myriotic representation is a representation in which there is no well-de ned number operator (the expectation value of such an operator is in nity), and therefore it is a non-Fock representation. Earman and Fraser observe that there are two senses of myriotic representations : absolute and relative. A Fock represen- tation can be myriotic in the relative sense if its number operator gives in nity relative to the vacuum state de ned in another representation. And Haag's theorem implies that representations in the interacting case will be myriotic, but not necessarily in the absolute sense. Still, as Earman and Fraser write, Fock repre- sentations are generally inappropriate for interacting elds ( Earman & Fraser, 2006 , p. 330). It is worth pointing out that, according to Earman and Fraser, while the existence of a Fock space representation is a necessary condition for the particle interpretation to be acceptable, it is by no means suf cient. As they put it, the latter requires a demonstration that the quanta of excitation of the eld can and do, under appropriate circumstances, display particle-like behavior ( Earman & Fraser, 2006 , p. 330). Unfortunately, they are silent as to what type of behavior can be classi ed as particle-like , hence we will have to use our own judgement in subsequent discussions.

10 Asymptotic particle states in scattering processes are formally de ned as limits when time t goes to minus or plus in nity ( in or out states). Unfortunately in the limit t / ± certain important operators (such as the unitary evolution operator) cease to be mathematically well-de ned. One possible remedy is to apply the procedure known as the LSZ approach, which enables us to consider in- and out-states at nite times and apply to them the usual renormalization techniques. However, this formal method does not seem to help the particle interpretation, since the commutation relations between raising and lowering operators no longer hold at nite times, and hence the total number operator cannot be interpreted as counting the number of interacting particles. See a discussion in ( Teller, 1995 , pp. 122 e124) and a response to Teller's objections to the LSZ approach in ( Bain, 2000 ). Interestingly, Bain claims that the existence of a well-de ned occupation number operator is not necessary for a particle interpretation to hold, as long as we have free asymptotic states. In Bain's opinion the LSZ approach can be also used to defuse the negative consequences of the Reeh-Schlieder theorem.

11 ( Bain, 2011 ) defends the view that the concept of localizability underlying the no-go theorems against the particle interpretation is inappropriate in the relativ- istic context, and thus should be abandoned in RQFT.

4 T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1

4

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

will look more closely at the origins of the particle interpretation of QFT, and we will argue that the very concept of quanta developed in the formalism of this theory is far from being unproblematic from a philosophical point of view. The way quanta are introduced within the mathematical formalism of QFT does not support the claim that they are ontologically robust entities resembling particles in any signi cant sense. We will also look with a critical eye at some standard claims regarding the ontological interpretation of the Fock space formalism. Finally, we will brie y discuss the alternative eld interpretation of QFT, noting that ontologically speaking it is no less dubious than the particle interpretation. The inevitable conclusion that follows is that there is still much work to do for a meta- physician who wants to answer the question of what precisely QFT tells us about the furnishing of the physical world.

2. Simple harmonic oscillator

The quantum-mechanical analysis of the simple harmonic oscillator (SHO) has played an important heuristic and conceptual role in the development of a full theory of quantum elds. The simple and elegant mathematical framework in which the standard description of SHO systems is usually cashed out serves as an effective tool that can be generalized and then applied to the fundamental procedure of eld quantization. While it is rather unrealistic to expect that all ontological commitments and pre- suppositions of a full-blown quantum eld theory can be traced back to those of the elementary theory of SHO, some connections between the conceptual underpinnings of the two theoretical frameworks should be possible to identify. In particular, the concept of a quantum , so fundamental in any serious ontological interpretation of QFT, is already present in its rudimentary form in the standard quantum-mechanical approach to SHO. In what fol- lows we will try to track the conceptual development 12 of the notion of quanta from its role of a mere bookkeeping device in the case of SHO to the ontologically-committing context of the particle representation in the formalism of (non-interacting) QFT. A simple harmonic oscillator is an object of a non-zero mass m which experiences the action of a force that is directly proportional (and opposite) to the amount of spatial displacement: F ¼ e u 2 x ( u is the characteristic frequency of the oscillator). In classical me- chanics the total energy of an oscillator is given by the well-known

expression E ¼

1 2 m u 2 x 2 , where p is the momentum of the

particle, and x its position. The form of this expression remains essentially the same in quantum mechanics, except that the

measurable quantities E , p and x are replaced by linear operators

system is

2 m u 2 b 2 . Solving the non-relativistic Schr odinger equa-

H ¼

tion we obtain the following formula describing the eigenstates of

the above-formulated Hamiltonian: E n ¼ Z u n þ 2 , where n is a

non-negative integer. Thus a quantum-mechanical harmonic oscillator can occupy only discrete energy levels: 2 Z u; 3 2 Z u; 2 Z u; etc. 14

2

b . 13

x

x

p

2 m þ

b

H ; p ;

b

2

b

b

and

Thus the

Hamiltonian of the

1

1

p

2 m þ 1

5

Corresponding to the eigenvalues E n there are eigenvectors which can be conveniently presented as integer-indexed kets j n . Now, it is possible to formally introduce a pair of new operators b a

and b y with the help of the following formulas:

a

b

a ¼

r

ffiffiffiffiffiffiffiffi

m

u

2

Z

b x þ i

m

u

b

p

(2.1)

b

a y ¼

r

ffiffiffiffiffiffiffiffi

m

u

2

Z

b x

i

m

u

b

p

(2.2)

From these de nitions we can easily derive the commutation

a ; b y ¼ 1, as well as reformulate the Hamiltonian in terms

relation ½ b a

of the new operators:

b

H

a

¼ Z u b

y b a þ 2

1

(2.3)

b

Given that the action of the Hamiltonian H on the eigenvector

j n produces vector Z uðn þ 2 Þj n , it is a simple exercise to verify that

1

the operators b a and b y act on the energy state j n by, respectively,

lowering and raising it by one unit:

a

b

a

j n ¼

p

ffiffiffi n jn 1

(2.4)

b

a

y j n ¼

p

ffiffiffiffiffiffiffiffiffiffiffiffi

n þ 1 j n þ 1

(2.5)

with the additional proviso that the action of b a on the zero-point energy state j 0 gives the zero vector (which is of course distinct

y

from j 0 ). These facts justify the standard terminology in which b a is called the raising operator, while b a the lowering operator. In addition to that, it can be checked that the action of the operator

b a y b a that gures in formula (2.3) on vector j n produces n j n . Thus

the operator b y b a measures the number of the energy level of the oscillator, and hence is usually referred to as the number operator. So much for the (rather elementary) formalism. However, the standard gloss that is offered in connection with the above- introduced mathematical model of SHO is that the energy levels of a harmonic oscillator can be interpreted as corresponding to new entities that may be called quanta of energy . 15 More precisely, we

are invited to accept the following paraphrase: instead of saying that the oscillator occupies the n -th energy level, we can express the same fact by saying that the oscillator produced n quanta of energy, each carrying the unit of energy equal to - u. But what can justify a transition from the language in which we talk about the

physical oscillator and its properties (i.e. its energy levels) to a

language that introduces reference to a new category of entities over and above the bearer of the energy properties? The only

reason that is supplied by physicists is that the energy levels of the oscillator are discrete and equally spaced, hence they admit a formal talk in terms of the number of something . But it is de -

nitely too hasty to jump to the conclusion that we have suf cient

a

12 We have to stress that the conceptual development of a theory is usually different from the historical one. In subsequent discussions we will largely ignore the latter one.

13 The mathematical details of the quantum-mechanical analysis of SHO can be found in any elementary textbook, e.g. ( Lancaster & Blundell, 2014 ; chap. 2).

14 The fact that the lowest energy level n ¼ 0 is actually non-zero is usually dis- missed as insigni cant on the basis of the assumption that only differences in energy really matter physically. What is signi cant is that the differences between the energy levels of the quantum oscillator come in multiples of - u.

15 This interpretation is accepted almost universally and with no hesitation in both physical textbooks and philosophical literature. Redhead, commenting on the procedure of quantizing a collection of independent harmonic oscillators (which we will discuss in the next section) writes for example the spectrum for E can be interpreted as arising from n k particles with energies - u k . ( ) So we have here the simple equivalence: the excitation number n k of the k-mode is identi ed with the number of particles with momentum - k associated with that state of the eld. Thus we get a particle interpretation of the quantized eld ( Redhead, 1988 , pp. 13 e14).

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1 e

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

5

grounds here to posit a new category of entities. 16 From a meta- physical point of view the quanta of excitation, as they are some- times referred to, lack some of the fundamental characteristics of genuine entities. I take it that the minimal requirements that genuine physical entities should satisfy are as follows. First of all, they should be countable, i.e. it should make sense (at least in some physical situations) to ask the question of how many objects there are. On the other hand, I am not assuming that they must be capable of being ordered (and thus I do not presuppose that a non- identical permutation of a set of objects must create a new physical situation). 17 More importantly, genuine entities should be capable of possessing a suf ciently broad range of attributes. In the case of physical entities, an obvious requirement seems to be that they possess measurable properties such as (relative) position, mo- mentum, energy and the like. Possessing different (but often imprecise) values of these properties should be a contingent mat- ter, i.e. a given object should be able to change its velocity, position, etc. without losing its own identity. Taking into account the lessons from the relativistic variants of quantum mechanics we do not want to assume in addition that physical entities must be localizable in any strong sense of the word. So it is admissible that a given entity can never be characterized by a de nite location. However, we insist that some changes of the vague or fuzzy attributions of properties should be possible. As it turns out, only the rst of the above characteristics (i.e. countability) can be uncontroversially attributed to the quanta of excitation. To begin with, quanta do not seem to be capable of being sub- jects of property attributions. On the face of it, they seem to possess at the very least the energy properties, as they are quanta of energy, after all. But this is problematic. Quanta of oscillation do not possess energy; if anything, they are energies (or energy portions, to be precise). The energy is still attributed to the underlying physical oscillator. Moreover, even if we agreed that in some sense an in- dividual quantum possesses the amount of energy that equals - u, still it is curious that quanta associated with a given oscillator are incapable of receiving any other amount of energy. The energy level of, let's say, 102 units of - u, is described as consisting of 102 quanta each carrying the equal amount of energy, rather than one quantum whose energy is 102 - u. But it seems that genuine phys- ical entities should be capable of possessing some possible prop- erties that they don't possess in actuality. 18 The table in front of me is painted brown, but it could equally well be black. If quanta cannot be said to possess (or not to possess) some properties accidentally, it is doubtful that they are legitimate physical entities. Furthermore, the quanta of oscillation in the case of individual SHOs lack some other attributes that are usually associated with physical objects, such as position. It doesn't make much sense to ask where in space a particular quantum of oscillation is located. Admittedly, it may be noticed that we could attribute a particular value of momentum to the quanta of oscillation, since momentum is directly related to energy. But again, a given quantum could not possess an amount of momentum different from the one that it actually possesses.

16 A similar opinion is expressed by Kuhlmann, who writes all that is established so far is a certain discreteness of physical quantities which is one feature of par- ticles. However, this is not yet conclusive evidence for a particle interpretation of QFT ( Kuhlmann, 2010 , p. 30).

17 Several authors insist that quantum objects do not possess individual identities that would enable us to order them in different ways, and thus are non-individuals that can be only counted (see French & Krause, 2006 for an extensive analysis of this position). I do not want to discuss here this claim, but see sec. 4.2 for some critical comments.

18 We stress that this characteristic applies to physical objects only. Abstract en- tities, such as mathematical ones, may not be able to possess any properties that they don't possess actually. I am grateful to James Ladyman for pointing this out to me.

The main feature of quanta that supposedly justi es their objectual character is their countability. That is, it makes sense to ask how many quanta associated with this and this energy level are there?. But the possibility of being counted is too imsy a feature to justify treating quanta as metaphysically viable entities. By analogy, we can equally well count money in a bank account, but this does not make it legitimate to treat an individual currency unit as an independent entity. A particular amount of money is a determinate of a certain measurable property of this account (or its owner) e namely the capability to buy some xed quantity of goods or services. But it would be premature to insist that on top of that such a quantitative characteristic also measures the number of some entities (pounds, dollars, euros) associated with this account. This would be similar to the objecti cation of meters on the basis of the fact that the length of a particular object is exactly 4 m (and thus it must contain 4 entities of some kind). Needless to say, the commonly used terminology that insists on renaming the lowering and raising operators as annihilation and creation operators is rather misleading. So far we were unable to produce a compelling argument for the claim that when we move the oscillator from the energy state j n to the next energy level j n þ 1 , we create a new entity in the process. 19 This skepticism can be supported by citing examples of the use of raising and lowering operators in other quantum-mechanical contexts which never- theless are not interpreted in terms of creating or annihilating any entities. The well-known quantum-mechanical formalization of angular momentum provides such an example, in which we can move between discrete values (ranging from e m to m ) of the angular momentum in a given direction z by applying appropriate raising and lowering operators L þ and L e de ned in terms of the components of angular momentum: L þ ¼ L x þ i L y and L e ¼ L x e i L y . Despite the formal analogy between this case and the case of SHO, it is not customary to talk about quanta of angular momentum as if they were independent entities which could be identi ed with familiar objects such as photons or electrons down the road. We can thus conclude that at this stage the paraphrases in terms of quanta of energy can be at best seen as a metaphorical way of speaking, a façon de parler with no deeper metaphysical meaning. 20 But this assessment may obviously change when we move to more realistic physical scenarios.

3. Systems of N oscillators

Before we consider examples that make closer contact with the quantum theory of elds, let us brie y discuss a seemingly trivial generalization of the case of a simple harmonic oscillator. This

19 Teller calls the interpretation of raising and lowering operators as representing physical processes of creation and annihilation misleading in the extreme ( Teller, 1995 , p. 137). He adds: Raising and lowering operators constitute no more than a convenient formal tool for working with this formalism, characterizing a relation between states that differ in occupation number by one. It is a huge, and as far as I can see unwarranted, step from these facts to the claim that the operators somehow describe a process of the actual creation and annihilation of quanta ( Teller, 1995 , pp. 137 e138). The Nobel prize laureate Richard Feynman once expressed, in his own unique style, a very similar sentiment, when he wrote I remember that when someone had started to teach me about creation and annihilation operators, that this operator creates an electron, I said, how do you create an electron? It disagrees with the conservation of charge , and in that way, I blocked my mind from learning a very practical scheme of calculation. ( Feynman, 1965 ).

20 It should be mentioned here that an additional physical motivation for calling the excitation - u a quantum can come from the well-known hypothesis of the corpuscular character of electromagnetic radiation as embodied in the notion of a photon. However, the corpuscular character of electromagnetic waves can be supported by experimentally veri ed phenomena, such as the Compton scattering and photoelectric effect. To my best knowledge, no similar arguments exist in the case of a single harmonic oscillator.

6 T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1

6

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

generalization relates to the number of oscillators only: instead of considering one oscillator we can describe a set of N independent, non-interacting oscillators, each with its own characteristic fre-

quency u k . From a physical point of view this case is easily reducible

b

to the previous one: in place of one Hamiltonian H we now have N

b

Hamiltonians H k , each with its own energy levels. But it may be argued that this example provides a slightly stronger case for the quanta of excitation. The reason for this is that now it is at least formally possible to speak about a particular quantum having different energies. Suppose that the state of the entire system of N oscillators is such that the oscillators numbered i and j are in the rst level of excitation, while the remaining oscillators occupy the lowest energy levels. One way of describing this situation in terms of quanta is that two quanta have been produced: one with energy Z u i and the other with different energy Z u j . This may prompt us to conclude that quanta can be discerned by the amount of energy they carry. This does not necessarily imply that the quantum with energy Z u i could possess a different amount of energy equal to Z u j ,

since the particular portions of energy may be considered an essential property of the given quantum. But it is feasible to insist that quanta are bareentities with no internal nature (essence) that can take on any amount of energy admissible in a given physical setting. Adopting this perspective, we may be tempted to interpret the following process (encoded in the combined action of one annihilation and one creation operator):

b b a i j 0 ; ; 1 ; ; 0 ; 0 ; ¼ j 0 ; ; 0 ; ; 1 ; 0 ;

a j

y

(3.1)

as consisting in moving one quantum from the state of energy Z u i to the state of energy Z u j . Of course we must be careful not to read too much into our formalism. In the case when the number of appropriate quanta is greater than 1, we may face some annoying interpretational ques- tions. Suppose, for instance, that we consider the following process (for simplicity we will limit ourselves to the case of four indepen- dent oscillators):

y

y

b a 4 b a 3 b a 2 b a 1 j 1 ; 1 ; 0 ; 0 ¼ j0 ; 0 ; 1 ; 1

(3.2)

Now, it doesn't make sense to ask whether the quantum with energy - u 3 in the nal state is the same as the quantum with en- ergy - u 1 or - u 2 in the initial state (the order in which we applied the creation and annihilation operators does not make a difference, since operators associated with different values of u do commute e see section 4 for details). But problems with the reidenti cation of objects over time are well known in standard quantum mechanics, for instance in scattering experiments involving particles of the same type. If we are willing to accept that quantum particles may lose their identity in physical processes, we should not refrain from admitting that the same applies to quanta of excitation. However, the majority of the conceptual problems that we described in connection with the case of a single oscillator remain unsolved. In addition to that we may point out that the rede- scription of the case of N independent oscillators in terms of the quanta of energy does not seem to offer any immediate bene ts that could justify the cost of expanding our ontology. Everything that can be said in terms of quanta, can be equivalently said in terms of oscillators and their energy levels with little or no loss in simplicity, elegance or other related virtues. But this may change when we consider a slightly different case of coupled oscillators. Coupled oscillators do not behave independently from each other; instead they physically interact with one another. One simple model of N coupled oscillators describes them as a chain of equally spaced masses connected by springs. Due to the fact that adjoining

masses interact with one another, the Hamiltonian of the entire system will contain components that depend on the differences in the displacement for neighboring masses. This means that the equations of motion cannot be decoupled into the set of N inde-

pendent equations, each describing one oscillator. And for obvious reasons the set of N coupled equations is much harder to solve than N independent equations. But it turns out that by switching to the formalism describing quanta of excitation (which, in the case of mechanical oscillators, are often referred to as phonons ) instead of oscillating masses we can effectively decouple the entire problem. We will not present all the mathematical details of this procedure, noting only its key steps (see Lancaster & Blundell, 2014 , pp. 25 e 27 for details). The rst step is to write down the position and momentum operators for each mass in the form of their Fourier expansions. By imposing appro- priate periodic boundary conditions we can make sure that the Fourier expansions contain exactly N distinct modes (waves, or pure tones ). Now, moving to the reciprocal space we consider the Fourier transforms of the original positions x j and momenta p j of the masses, and we interpret these transforms as if they were po- sitions and momenta of a new type of entities, e.g. the modes of excitation. As this is the crucial step of the entire argument, it may be perhaps worth looking into in a slightly greater detail. The Fourier decompositions of the positions and momenta of the masses into N independent modes of oscillation can be given in the usual form as follows ( a is a constant representing the spatial dis- tance between oscillating masses):

x j ¼ p 1 ffiffiffiffi

N

N

= 2

X

k ¼ N = 2

x k e i kja

~

(3.3)

p

p j ¼ 1

ffiffiffiffi

N

N

= 2

X

k ¼ N = 2

p k e i kja

~

(3.4)

The summation indices k represent wave numbers that identify

distinct modes of oscillation. Now, the coef cients ~ x k and p k of the

expansions (which themselves can be expanded in terms of x j and p j in reciprocal formulas precisely analogous to (3.3) and (3.4) ) are treated formally as if they were parameters ( position and mo- mentum ) characterizing the corresponding wave modes. Once this interpretation is imposed, it is a matter of some calculations to

prove that the total Hamiltonian expressed in terms of ~ x k and

will be the sum of N independent Hamiltonians, each of the form exactly analogous to the formula (2.3) that described an individual harmonic oscillator. Thus the problem becomes decoupled and is now open to the usual treatment that we applied to the system of N non-interacting oscillators. This is a remarkable result. It amounts to the observation that by switching to the description in terms of distinct modes of oscilla- tion rather than oscillating masses we achieve a substantial gain in the form of a much more treatable formalization of the problem. If we treat seriously the role of factors such as simplicity, elegance, and economy in the process of selecting scienti c hypotheses and theories, and if in addition we are willing to accept the popular philosophical view regarding how the con rmation of an entire theory trickles down to the con rmation of its existential com- mitments, we may conclude that the hypothesis of the existence of quanta of energy receives a substantial boost from the aforemen- tioned example of coupled oscillators. However, we should notice that we have already moved well beyond the original character- ization of the quanta of oscillation as presented in the context of SHO. In the process of decoupling the initial problem we attrib- uted positions and momenta (in the form of the Fourier

~

~

p k

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1 e

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

7

transforms of the positions and momenta of the oscillating parti- cles) to the modes of oscillation, and this step was conspicuously absent in the earlier characterizations of quanta. Moreover, the formalization adopted in the current case forces us to admit that there are exactly N quanta of oscillation associated with the considered system. This stands in stark contrast to the case of SHO, where the number of the quanta of energy was unbounded and depended only on the state of excitation of the oscillator. Hence one may even wonder if we are talking now about entities of the same type as in the previously considered cases.

4. The Fock space formalism

Leaving the problem of the relationship between phonons and quanta of oscillation aside, we can now explore yet another formal analogy that appears to give support to the interpretation of the quanta of energy as independent, particle-like entities. As we have seen in previous sections, the states of the system of harmonic oscillators can be written within a formalism that involves repeated actions of the raising operator on the ground-level energy state. This formalism has been actually developed in the context of the description of systems of N particles of the same type, and it carries the name of the Fock space formalism. In the standard quantum- mechanical formalism the joint state of N particles is represented by a vector in the N -fold tensor product of single-particle Hilbert spaces: H 5 H 5 5 H . However, this representation suffers from the problem of redundancy, since states that differ by permuting identical particles should not be distinguished. The Fock space representation is not vulnerable to the redundancy problem, and moreover is exible enough to consider cases involving inde- terminate numbers of particles (for instance in the form of super- positions of states with different particle numbers). Formally, a Fock space F is a direct sum of the type: F¼F 0 4 F 1 4 F 2 4 , where each summand F k is a space of states for k identical particles. 21 Obviously, F 0 consists of just one ray repre- senting the state with no particles, and F 1 is the usual one-particle Hilbert space. When k 2, F k is assumed to be isomorphic with the symmetric (or antisymmetric) sector of the full k-fold tensor product of one-particle Hilbert spaces. States belonging to F can be presented in a convenient notation as follows. We start with selecting a maximal observable O in the one-particle Hilbert space H , and we consider its eigenstates, which constitute an ortho- normal basis for H (for simplicity we will assume that the set of eigenvectors for O is countable, thus they can be written down as j o 1 , j o 2 , ). Now we can construct a (potentially) in nite sequence of non-negative integers in the form of a ket j n 1 , n 2 , n 3 , , with an obvious interpretation that the occupation number n j represents n j particles in the state j o j . Note that this ket can be always written as a vector v in the subspace F N where N ¼ P n i . The exact form of

i

the appropriate vector v is given by symmetrizing (anti-

symmetrizing)

5 5 j o 2 i n 1 þ n 2 5 The apparent advantage of the new notation,

however, is that it supposedly eliminates the need of using labels attached to individual particles, and thus dispenses with the concept of the primitive identity of particles. 22

the

product

j o 1 i 1 5 5 j o 1 i n 1 5 j o 2 i n 1 þ 1

21 See ( Teller, 1995 ; chap. 3), ( Lancaster & Blundell, 2014 ; chap. 3).

22 However, French and Krause (2006 , p. 390) point out that the standard occu- pation number notation only masks the use of labels, which are in a sense still indispensable, as it is mathematically impossible to build a vector space over the complex eld out of the sequences of the form j n 1 , n 2 , n 3 , . So, in order to take full advantage of the mathematical machinery of Hilbert spaces, we have to treat kets j n 1 , n 2 , n 3 , as shorthand for the labeled tensor products of the form written above.

In the next step we can introduce raising and lowering operators

b

and b a j with the intention to represent the process of adding or

subtracting one particle in the state j o j , in which case it is perfectly legitimate to use the creation and annihilation terminology. However, the details of the action of these operators on available states depend on the statistics of the considered particles. The fundamental distinction between bosons and fermions can be made by noting rst that the application of two creation operators in different orders should lead to physically indistinguishable states, since considered particles are of the same type. Ignoring the possibility of paraparticles, this leaves us with two options: either

a j

y

y

b a b a

j

y

k

a

¼ b k b

y

y

j

a (bosons) or b a k ¼ b

a

y

a

j

b

y

k b a (fermions). Using the letter c

j

y

y

for the creation/annihilation operators in the case of fermions, we

can sum up the distinction between bosons and fermions in the form of the well-known commutation/anticommutation relations:

h

h

b a ; b a

j

a

y

a

k i ¼ b

y

y

j

b a j ; b a k ¼ 0

b a j ; b

y

k i ¼ d jk

b

a

y

k

a

b k b

y

y

a ¼ 0

j

n

y

j

b c

; b

c

c

k o ¼ b

y

y

j

b

c

b c j ; b c k ¼ 0

n

c

b c j ; b

y k o ¼ d jk

k þ b c k b c ¼ 0

j

y

y

y

(4.1)

(4.2)

The bosonic commutation relations (4.1) imply that the actions of the creation and annihilation operators on the states written in the appropriate basis will be exactly as in the SHO case (formulas (2.4) and (2.5) ). On the other hand, the fermionic anticommutators (4.2) ensure that the repeated action of a creation/annihilation

operator results in the zero vector: b c ¼ b c j b c j ¼ 0. This implies that

c

j

b

j

y

y

the occupation numbers n j for fermions can be only 0 or 1, which

re ects the Pauli exclusion principle.

It has to be admitted that the mathematical congruity between the Fock space formalism as applied to the case of N particles, and

the Fock space formalism describing the case of energy levels for harmonic oscillators, may give us reason to interpret the latter in terms of some particle-like entities. However, formal analogies have their limitations, as we have already noted in section 2 . The above-described formalism has been based on the assumption that

a basis for an appropriate Fock space can be constructed with the

help of the spectral decomposition of a selected single-particle observable O , so the premise that we talk about particles and their

properties has already been built into the procedure. On the other hand, distinct slots in the occupation number formalism applied to the case of N harmonic oscillators differ only with respect to which oscillator (and therefore which characteristic frequency) they represent. In order for the analogy between the two cases to be complete, we need to make the crucial assumption that the space built formally out of the set of orthogonal vectors, each of which corresponds to a distinct oscillating frequency u k , represents possible states of a single object e a quantum. But this is precisely

what we wanted to argue for in the rst place, so the argument from analogy turns out to be circular. It seems that the existence of

a Fock space representation does not by itself guarantee that the

theory in question describes particles rather than other entities,

such as elds. We ought to look for some further arguments in support of the ontological interpretation of QFT in terms of particles.

8 T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1

8

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

4.1. An argument that a Fock representation supports a quanta interpretation

In this subsection I will brie y review an extended argument given in ( Fraser, 2008 ) for the claim that a Fock representation in the case of a free eld admits (indeed necessitates) an ontological interpretation in terms of particle-like quanta. Fraser notes rst that any Fock representation possesses the following formal fea- tures: (1) it contains well-de ned creation and annihilation oper- ators, as well as the number operator, (2) there exists a unique ground state j 0 , (3) the entire Fock space can be spanned by applying creation operators to the ground state. She correctly ob- serves that in spite of these formal features (and the suggestive nomenclature), a Fock representation is not automatically guaran- teed to admit a quanta interpretation. After all, the number oper- ator can count values of any discrete property (for instance energy or angular momentum levels) instead of entities. In order to argue for the quanta interpretation of the number operator, Fraser uses an expression for the Hamiltonian written in terms of the number operator (see formula 5.2 later in this paper with d(0) assumed to be zero). Then she observes that the ground state j 0 is an eigen-

b

state for H with the corresponding eigenvalue equal to 0. This clearly indicates that the ground state corresponds to zero energy. This, plus the fact that the ground state is invariant under the ac-

tions of the Poincar e group, constitutes for her an argument that the ket j 0 represents the situation in which there is no particle. However, this argument evidently falls short of establishing that the Fock space formalism necessitates the no-particle interpretation of the ground state. Rather, it shows that such an interpretation is compatible with the formalism. Fraser writes: it is reasonable to interpret j 0 as a state in which there is no particle, because it is a unique state with the desired properties: it is the lowest energy state and looks the same to all inertial observers . And it is certainly true that a state describing a situation with zero particles should be unique and should possess the above-mentioned properties. But this only establishes that if we knew independently that a Fock representation contains states corresponding to some numbers of particles, we would have to conclude that the ground state is the only representation of the situation with no particles. But alterna- tive interpretations, for instance in terms of discrete energy levels of some underlying physical system (be it elds or systems of harmonic oscillators) are still possible. A eld with zero energy (and thus identical to zero everywhere) would also look the same for all inertial observers, precisely as is the case with vacuum. 23 Fraser's argument does not stop here but continues to consider one- and many-particle states. Essentially, her reasoning is as fol- lows: we observe rst that states obtained by applying a sequence of creation operators to the ground state correspond to energies that can be written down as a sum (integral) in which each summand has the form n p - u p , where n p is the number of creation operators cor- responding to a given value of momentum p . Given the assumption that we are considering a eld obeying the Klein-Gordon equation (see section 5 for details), the characteristic frequency - u p turns out

to be equal to

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p 2 þ m 2 , which e as Fraser observes e is the correct

p

relativistic formula for the energy of a particle with momentum p and mass m . According to her, this single observation provides us with the de nitive argument for the quanta interpretation . But this can be doubted. First of all, to interpret the parameter m guring

23 In a sense, it may be observed that an interpretation of the vacuum state is necessarily given in terms of the lack of something , but what this something is depends on how we interpret other states orthogonal to the vacuum. Thus it seems impossible to give a conclusive argument in favor of the particle interpretation merely on the basis of an analysis of the ground state.

in the Klein-Gordon equation as the mass of an object already begs the question, since it assumes from the outset that we are dealing here with particle-like objects (see sec. 5 for a more extensive analysis of this problem). In the absence of a conclusive argument that m characterizes some objects, we are left with essentially the same mathematical fact as in the case of SHO's, namely the discreteness and equal-spacing of admissible energy levels, as the only ground for the quanta interpretation. And, as we have seen earlier, this is a very shaky ground to base an entire ontological interpretation of QFT. It seems that Fraser's argument does not advance the particle interpretation of free elds far beyond what we have already considered and dismissed.

4.2. The problem of the discernibility of particles

Let us digress for a moment from the main subject of the paper, which is an evaluation of the ontological status of quanta, in order to make a more general observation regarding the intuitive interpre- tation of the Fock space formalism vis- a-vis the standard formalism of symmetric/antisymmetric sectors of the tensor products of Hil- bert spaces. Apart from the fact that the satisfaction of permutation- invariance is automatically guaranteed by the fact that the Fock formalism does not even make it possible to meaningfully apply a permutation to a state (the state registers only occupation numbers without saying which particle belongs where), we may notice that on the face of it the formalism seems to imply that particles can be discerned by their state-dependent properties. If we consider for instance the state of the system of two indistinguishable particles given by the ket j 0, 1,1, 0, , its standard reading should be that one particle occupies state j o 2 while the other is in the orthogonal state j o 3 , and hence the particles are qualitatively discernible by two distinct values o 2 and o 3 of the observable O . In the case of bosons it is possible to have a different state j 0, 2, 0, , which obviously implies that the two particle occupy the exact same state. However, two fermions can never do that, so as long as their state is not an intricate superposition, their discernibility is guaranteed. But this conclusion ies in the face of the well-entrenched view that quantum particles of the same type (whether bosons or fermions) are never distin- guishable by their properties. This view, which goes back to the seminal works by French and Redhead (1988), Redhead and Teller (1992) , and Butter eld (1993) , is based on an analysis of the sym- metrized and antisymmetrized states of two particles in the tensor- product formalism. It is perhaps best to use the standard case of two spin-1/2 par- ticles as an example. Let us consider the joint state of two electrons which we initially classify as including one electron with the z-spin up, and one with the z-spin down. When we write down this state as the direct product j [ z 1 5 j Y z 2 , we can immediately notice that this vector does not satisfy the requirement of antisymmetricity. In order to correct this problem, we antisymmetrize the product and thereby obtain the following state with the right type symmetry:

1

p ffiffiffi j [ z 1 j Y z 2 j Y z 1 j [ z 2

2

(4.3)

Now, in order to talk about properties of the individual com- ponents of the system described in (4.3) we may resort to the standard procedure of calculating partial traces of the above-state (or, equivalently, we can compute expectation values of the oper- ators of the form O (1) 5 I (2) and I (1) 5 O (2) in state 4.3). Either way, the conclusion is clearly that whatever quantum-mechanical property is possessed by particle number 1, it will be possessed by particle number 2 (this should be obvious, given the fact that state 4.3 was designed to remain invariant under the permutation of labels 1 and 2). But how can this observation be reconciled with

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1 e

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

9

the fact, stated earlier, that in the occupation number notation the state of the two electrons is given as j 1, 1 z which under the stan- dard reading implies that we have two particles, one of which is in the state spin- z up , and the other in the state spin- z down ? An inescapable consequence of the above is that the labels used in formula (4.3) can't possibly refer to the same objects as the distinct slots containing two ones in the corresponding occupa- tion number ket. But which method of reference picks out the true electrons, and which is a mere diversion? Those who want to retain the Indiscernibility Thesis will insist that we should not read the occupation number kets literally, but instead interpret them as a mere notational variant of the more fundamental, label-bearing states of the form (4.3) . 24 But an alternative solution is to treat the occupation number formalism as revealing the true ontological commitments of the underlying theory, while rejecting the use of labels in the standard tensor-product formalism as part of the formalism with no physical meaning. Following this suggestion we arrive at an unorthodox interpretation of the standard formalism, according to which the only admissible way to individuate particles that jointly occupy the state given in (4.3) is with the help of the qualitative properties of the entire system which can be encoded in symmetric projection operators. The details of this individuation procedure can be found elsewhere ( Bigaj, 2015; Caulton, 2014 ). For our purposes it is suf cient to observe that the unorthodox approach to the individuation of composite systems of indistin- guishable particles 25 is perfectly congruous with the intuitive interpretation of the Fock space formalism. Two electrons occu- pying the state (4.3) can be said to possess two precise values of z - spin, exactly as implied by the direct reading of the ket j 1, 1 z . However, we should brie y mention one thorny issue that may be taken as undermining the non-standard interpretation of joint states of indistinguishable particles. As is well known, the fermionic state (4.3) has the property of being spherically symmetric, which means that it can be written equivalently in any selected basis of the spin space. The Fock space formalism expresses this fact in the observation that all fermionic kets of the form j 1, 1 n , where n is any direction of spin, are identical. But how can two electrons possess de nite values of incompatible properties (i.e. different components of spin)? This is a serious interpretational issue that deserves a separate treatment (see Caulton, 2015; Bigaj, 2016 for some attempts along this line). For now we can only observe that the unorthodox approach to individuation is in no worse position than any inter- pretation of quantum mechanics that relaxes one of its standard interpretational rules (usually the eigenstate-eigenvalue link). Two main strategies of presenting and justifying such a relaxation can be applied here. One is based on the assumption that the alternative presentations of the state (4.3) in different spin-bases correspond to distinct methods of carving up the entire composite system. Thus

24 For instance, Bas van Fraassen states emphatically that we must de nitely not think of the above vector [i.e. the ket of the form j n 1 , n 2 , n 3 , ] as the state of a simple collection of particles with n 1 in pure state x 1 and so forth. ( van Fraassen, 1991, p. 441). He goes on to explain that particles whose joint state is described by an occupation number ket are in fact in the same mixed state obtained by reducing the original symmetric (antisymmetric) state. The argument by French and Krause mentioned in ft. 21 seems to point in the same general direction. However, they want to avoid reverting to the label-based formalism, and therefore insist that the correct interpretation of occupation number kets should be given in terms of a non-standard logic of non-individuals (e.g. quasi-set theory). For them an occupation number ket describes a collection of N particles that are countable but not numerable .

25 We have to keep in mind that the adjective indistinguishable used to refer to particles of the same type (electrons, photons, neutrinos) means only that the particles are not distinguishable by their state-independent properties. This does not exclude the possibility, accepted in the currently considered approach, that indistinguishable particles may be distinguished by their state-dependent properties.

we may insist that in a sense what we have here is an in nity of alternative pairs of electrons with values of spin in different di- rections being ascribed to each pair. Another solution is to seek a preferred basis in which the decomposition of the state (4.3) should be performed. This can be done in most cases of interest if we introduce the spatial degrees of freedom and consider position as an ontologically privileged parameter (for details see Bigaj, 2016 ).

5. Field quantization

Let us now return to the task of interpreting the concept of quanta, moving on to a more realistic case of eld quantization. The procedure of building a quantum version of a classical eld is well known from a mathematical point of view. However, its precise ontological meaning remains somewhat controversial. We assume at the outset the existence of a classical physical eld f( x ) whose quantum theory we are trying to develop (a typical example used in this context is a massive scalar eld satisfying the Klein-Gordon equation). This eld assigns particular values (intensities) to spatiotemporal points (to make matters simpler we will consider a non-relativistic case in which we can discuss the eld at a particular moment of time as assigning properties to spatial points). Formally, the procedure of the quantization of f( x ) consists of nding oper- ators representing the eld f and its conjugate momentum p in such a way that the quantum-mechanical commutation relation

½ f b ð x Þ ; pbð y Þ ¼ i dð x y Þ is satis ed. This can be achieved using the following, standard method. The starting point is to rewrite the Fourier expansion (3.3) that we gave in the case of N oscillator so that it can cover the case of continuous systems. Heuristically, this is done by assuming that the constant a representing the distance between adjacent oscillators approaches zero, and consequently the number of oscillators increases to in nity. In the limit the discrete number of parameters x j representing positions of indi- vidual oscillators is replaced by a continuous function f( x ), and instead of the summation we substitute integration. The quanti- zation of the Fourier expansion is achieved when we replace real- valued functions by appropriate linear operators. Using the de - nitions (2.1) and (2.2) which enable us to express coef cients in appropriate Fourier expansions in terms of raising and lowering operators, we arrive at a new, quantum-mechanical expression of the considered eld ( Peskin & Schroeder, 1995 , p. 19ff; Lancaster & Blundell, 2014 , p. 98ff):

b

f ð x Þ ¼ Z

a

K b

p e ipx þ b

a p e ipx d p

y

(5.1)

where K ¼

p 2 þ m 2 (the positive relativistic

solution for energy). The next step in the formal quantization procedure is to sub-

stitute f b ð x Þ for the ordinary eld f( x ) in the classical expression of the Hamiltonian governing the eld. Depending on the form of the Hamiltonian, it may be possible to diagonalize it, i.e. nd its eigenstates and eigenvalues. In particular, in the case of a massive scalar eld, the Hamiltonian can be shown to have an expression in terms of creation and annihilation operators which is exactly analogous to the Hamiltonian (2.3) for the case of SHO:

1 2 , and E p ¼ 1 ð 2 pÞ 2 3 ð 2
1
2 , and E p ¼
1
ð 2 pÞ 2 3 ð 2 E p Þ

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

p

b

H

a

¼ Z E p b

y

p b a p þ

1

2 dð 0 Þ d p

(5.2)

The in nite value of d(0) is a matter of concern, which is usually dealt with by attaching physical importance to energy differences rather than to absolute values of energy. An alternative method of taming the in nity in (5.2) is to deploy a formal procedure known

10 T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1

10

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

as normal ordering. After applying normal ordering to the Hamiltonian in (5.2) we effectively eliminate the second summand in the formula. The conclusion of the quantization procedure in the case of non-interacting massive scalar elds is that the eigenstates of the Hamiltonian can be built out of the vacuum state j 0 by

applying to it any number of creation operators b p . The ontolog-

ical interpretation of this formal procedure is perfectly analogous to the interpretation we have already encountered in the case of quantum-mechanical harmonic oscillators. Each application of the

a

y

operator b p is said to create a quantum with momentum p , thus any

state of excitation of a given eld can be thought of as involving a number n p of quanta for each available momentum p . Formally speaking, we have arrived at a Fock representation of

the commutation relation ½ b ð x Þ ; pbð y Þ ¼ i dð x y Þ . This is usually

taken as ensuring that the particle interpretation of the eld is available. However, we have to keep in mind Earman and Fraser's word of caution, as well as the argument presented in section 4 : the existence of a Fock representation does not by itself guarantee that the operators of creation and annihilation will involve anything remotely resembling particles. We have to make sure that the behavior of the quanta of the eld will be appropriate for a particle-like object. And, as we recall, quanta of excitation in the case of harmonic oscillators lack certain fundamental features of particles, such as the ability to possess various properties other than the xed amounts of energy and associated momentum. So it seems that the case of continuous elds suffers from conceptual problems similar to those affecting the previously considered case of N harmonic oscillators. In response to that, it may be argued that the above-sketched procedure of quantizing a classical eld opens a road to assigning to the quanta of excitation properties, such as mass, that were not readily available in previous cases (this argument was already mentioned in Section 4.1 ). The way to do this is rather simple: in the original Klein-Gordon equation there is a parameter m that enters into the formula resembling the proper relativistic relation be- tween energy and momentum: E 2 ¼ p 2 þ m 2 . It is thus natural to

interpret this parameter as representing the mass of the quanta. As a bonus, we receive an explanation of why all quanta of excitation possess the same mass e simply because there is only one parameter m and it does not depend on momentum. 26 In spite of its naturalness, one can have legitimate reservations as to the appro- priateness of the proposed interpretation of parameter m . The

crucial step of the method of nding the energy spectrum in the case of the Klein-Gordon eld f is the transformation of the original equation into the set of equations that are formally identical to the equation of a simple harmonic oscillator. Each of these equations corresponds to one particular mode of the original eld, given by its momentum p , or the corresponding characteristic frequency

a

y

f

u p ¼

1

Z

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

p

p 2 þ m 2 However, in the original case of SHO the param-

eter m referred to the mass of the oscillator, and not to the mass of the quantum of excitation! In fact it doesn't even make sense to attribute m to the quanta, since they do not perform the oscillations (they are modes of oscillation). In the case of elds there are no massive particles that do the oscillations (instead, it is the eld it- self that oscillates), however it is unclear whether this fact can in any way legitimize the proposed shift in the interpretation of the parameter m which is now transferred for no apparent reason from the characteristic of the underlying oscillator ( eld) to its quanta of excitation.

26 Huggett sees this explanation as a remarkable case of uni cation. See ( Huggett, 2000 , p. 624).

Another important characteristic that the quanta of excitations lacked in the SHO case was position. Again, it may be argued that the eld case fares better in this respect, since there is a well- established procedure that can enable us to talk about creating a quantum of eld excitation at a precise location x in space. In fact, it

f ð x Þ y

is precisely the eld operator f

b ð x Þ and its complex conjugate b

that play the role of the annihilation and creation operators in the position space. This interpretation follows from the straightforward

computation of the result of an action of the operator b ð x Þ y on the

ground level state j 0 . Using the Fourier expansion of b ð x Þ y in the

momentum basis it can be veried that the result of applying this operator to the vacuum state has the same mathematical form as the well-known representation of the exact position state j x in the momentum basis ( Peskin & Schroeder, 1995 , p. 24). Moreover, the

calculation of the expression y j b ð x Þ y j 0 yields d( x e y ), which gives

an additional reason to believe that b ð x Þ y j 0 represents the state of

a particle located at point x . The last statement can be further

con rmed by applying the total number operator (represented by

the sum, or more appropriately, the integral of all operators b a

over all available momenta p ) to the state b ð x Þ y j 0 . The result of this

application is the original state b ð x Þ y j 0 , which shows that the state

f b ð x Þ y j 0 is an eigenstate of the number operator with the corre-

sponding eigenvalue equal to 1, and therefore there is indeed one particle created at x ( Lancaster & Blundell, 2014 , p. 38). Undeniably this is a powerful argument which e if accepted e

brushes away one of the central objections against the concept of quanta presented earlier. However, it may be worth pointing out that there are certain gaps in the argument which slightly diminish its philosophical strength. The crucial step of the argument is the formal analogy with the case of the quantum-mechanical description of one particle. In this case we know that the state j x representing one particle with well-de ned position can be alternatively presented in the momentum basis as the Fourier transform of j x . Thus the momentum-state of the particle in question is given in the form of a plain wave spread in the momentum space (an innite superposition

of the states of the particle with well-de ned momenta). Now, in the

case of QFT we have proven that the state b ð x Þ y j 0 can be presented as

a mathematically analogous superposition of states, each of which

represents a situation with one particle created at a precise mo- mentum value p . Strictly speaking, this is not the same as a super- position of the states in which one and the same particle has different momenta. Rather, what we have here is a superposition of states, each of which represents some particle with momentum p. (Admit- tedly, this remark is likely to be dismissed by a physicist as nitpicking, but that is precisely what distinguishes physics from philosophy.) Thus, the analogy with the case of one particle is not complete,

and the conclusion regarding the interpretation of the state f b ð x Þ y j 0 stands in need of some further supporting arguments. Such an argument may be sought in the fact, mentioned above, that the

total number operator gives the value 1 in the state b ð x Þ y j 0 , thus

presumably closing the gap. Apparently, if there is exactly one particle in the situation described by the superposition of states

b

a p j 0 , this seems to con rm that each state b p j 0 represents a case

in which this very same particle has different momenta p . And yet

this step seems to be slightly too quick. To begin with, we can write

f

f

f

f

y p b a p

f

f

f

f

y

a

y

down numerous combinations of the basic one-particle states b p j 0

for which the total number operator will give a different or even not

a

y

well-de ned value. So it is not generally true that the states b p j 0

can be interpreted as describing the same particle with different momenta, since in that case the total number of particles should always be one. Why, then, should we conveniently adopt this

a

y

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1 e

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

11

special interpretation just in the case of a state that has the form of a Fourier transform of j x ? And, more generally, it is by no means a

foregone conclusion that the sum of all the operators b p b a p indeed

measures the total number of particles actually existing in a given state. In states with well-de ned numbers of particles for each value of p this operator gives the results we would normally expect,

a

y

but in intricate superpositions of states b p j 0 it may just tell us

about the propensities of the system to reveal certain numbers in experiments rather than informing us about the objective existence of particles in the considered pre-measurement state. Finally, there is a problem that is directly linked to the contin- uous character of elds. As we have already explained, the Fock representation underlying the particle interpretation is given in the basis formed by the action of a potentially in nite sequence of creation operators on the vacuum state j 0 . Thus the number of the quanta of excitation associated with a particular state has no upper bound. This fact is acknowledged by the commentators, and yet its ontological and physical consequences are rarely mentioned. A situation in which an accessible state of a eld can contain an in nite number of particles is hardly physical. Moreover, it may be pointed out that the particle interpretation of QFT admits even more troublesome scenarios. If we accept that each mode of the eld associated with a particular value of momentum p is inter- preted as a separate harmonic oscillator with its own number of quanta of excitement, then mathematically it is possible that there can be uncountably many quanta corresponding to a particular state of the eld. For this to happen it suf ces that the state of the eld is non-zero within a continuous region of the momentum space, for instance in an interval [ p 1 , p 2 ], which means, according to the particle interpretation, that all harmonic oscillators de ned by the values of p within this interval have at least one quantum of excitation associated with them. But the existence of an uncount- able number of physical objects, each possessing a non-zero mass, is indeed hard to swallow. 27

a

y

6. Quanta or elds?

In the light of the above remarks it can be asked if an inter- pretation of quantum elds in terms of discrete, countable entities is a viable one. Or, perhaps, should we accept that ultimately quantum eld theory is about elds, and quanta play only a heu- ristic role which should be treated purely instrumentally without affecting our fundamental ontology? It has to be admitted that the situation is far from being clear. First of all, we should note that in the process of quantizing a classical eld we are radically changing not only the mathematical description of the considered eld, but also its intuitive physical (or even metaphysical) interpretation. The classical picture of a eld is that it is a kind of ethereal medium that is spread in spacetime. It is open to debate whether we should follow the substantive reading of a classical eld (gravitational, electromagnetic, etc.), or admit simply that a eld is just an assignment of various values of a particular measurable quantity to

27 The continuity of elds may prompt us to reconsider the standard interpreta-

tion of the operators b p ; b a p and n b

p b

correspond to numbers of particles, perhaps it is more accurate to say that they

represent densities of particles. For instance, if we look at formula (5.2) expressing the total energy of the eld, we can notice that for the integral to make sense the

a y

p ¼ b

a a p . Instead of saying that these operators

y

number operator n p ¼ b p b a p should be interpreted as giving us the number of

particles per unit of momentum, and not its absolute value. Thus, for instance, in the case when we have only one mode of the eld excited at value p , we should use Dirac's delta function in order to get a meaningful answer for the total energy value. But the jump from discrete, countable objects to smeared densities of objects is a

step away from the ordinary particle interpretation and towards an alternative eld interpretation.

a

y

each individual spatiotemporal point with no underlying sub- stance . However, a crucial observation to be made here is that

different physical states of a given eld (different con gurations of

a eld) correspond to different assignments of these values to

appropriate points. Quantities assigned to points are typically represented by mathematical objects of a certain kind (real

numbers, vectors, or tensors), and the kind of mathematical objects

associated with a given eld determines its character (whether it is

a scalar, vector, or tensor eld). To illustrate this observation, let us consider a standard problem

in classical electrodynamics: how to calculate an electric eld in a

given closed region R of space. In order to do that, we need two pieces

of information: the distribution of electrical charge within R , and the

boundary conditions in the form of the values of the electric potential

at

the boundaries of R . Once this is supplied, there is a unique solution

of

the Maxwell equations that describes the exact form of the electric

eld in that case. Given alternative boundary conditions and the charge distribution, we can arrive at a different eld assignment. These different physical realizations can be thought of as different eld congurations, or different physical states of the same under-

lying entity, which is an electromagnetic eld. On the face of it, the transition from the classical to the quantum does not change this general picture, only its mathematical details. As we have seen, instead of ascribing real numbers or vectors or even tensors to individual points, we associate with them linear operators on an appropriate Hilbert space. But this alteration has deeper consequences for the meaning of the applied mathematical

machinery. In order to talk about alternative physical realizations of

a given eld, we no longer consider alternative assignments of values to points. Instead, the assignment involves immutable operators that are xed for a given eld (in Teller's terminology they are determinables rather than determinates ). 28 When we

look at formula (5.1) , we note that symbols b p and b a p have the same

referents regardless of the physical state of the eld: they refer to

de nite mathematical objects (operators) selected by their action

on particular states in the Fock space. Different physical states of a

eld correspond to the exact same formula (5.1) expressing the

eld f b ð x Þ . How, then, can we distinguish between distinct physical realizations (con gurations) of a eld? The answer is given in the form of quantum-mechanical states. A particular eld can be in the ground (vacuum) state j 0 , or in any state of excitation produced by

a

y

applying a sequence of creation operators b p to j 0 .

Thus (5.1) no longer describes individual physical con gurations (assignments of values) of a eld. The proper interpretation of the

eld operator f b ð x Þ is somewhat tricky. We may provisionally see it as picking out a physical object whose different energy states are

a

y

given in the form of combinations b p 1 b a p 2 j 0 . In other words, a

eld is the bearer of the energy properties described by the appropriate kets. But once we realize that concrete realizations of a eld are no longer considered as determined by various alternative assignments of properties to points, we may start wondering what the true nature of elds is in the current approach. And one possible approach is to simply read off the metaphysical interpretation of a

a

y

y

28 The immutable character of a given quantum eld is no longer assumed when

we move from the Schr odinger to the Heisenberg picture, in which operators become time-dependent while states remain xed. However, as Teller (1995 , p. 97ff) explains, we should not interpret the Heisenberg picture as implying that time-dependent operators represent values of a quantum eld evolving in time. Rather, the Heisenberg operators encode patterns of evolution of a given value for any system possessing such a value. In other words, the Heisenberg operators are

just part of an algorithm that enables us to calculate the expectation values of measurable quantities at any time given that we have the initial state.

12 T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1

12

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

quantum eld from the most natural interpretation of its physical properties, as described in energy states. This most natural inter- pretation, as we have already seen, is the interpretation in terms of particle-like quanta. Thus a suggestion can be made that the mathematical object given in (5.1) refers, in a roundabout way, to a system of particles which physicists consider as associated with a given eld. 29 The connection between quantum elds as given in (5.1) and particles can be observed to be even stronger when we

recall that f

b ð x Þ plays an additional role of an operator that increases

the number of particles at a given location x by one. 30 In order not to revert to the old particle interpretation of QFT we have to propose an alternative reading of the states of a given quantum eld. Such a eld interpretation is available in the form of the wavefunctional approach (cf. Huggett, 2000 , pp. 620 e 622; Baker, 2009 , pp. 588 e 592). In this approach the space of states for a particular eld is created, roughly speaking, as the set of square- integrable wave functions de ned on the space of classical eld con gurations. That is, if f( x ) represents a classical eld con gu- ration assigning particular values to individual points, then the quantum states of the considered eld are given in the form of

functionals j[ f( x )] which encode the information about the prob- ability that the eld will adopt a classical con guration in the form of the function f( x ) when measured. To put it differently, according to the wavefunctional interpretation a state in QFT represents propensities for the manifestation of certain (classical) elds in the event of measurement ( Baker, 2009 , p. 590). Apparently, this interpretation of QFT brings the concept of a quantum eld closer to the classical notion, since it is now possible to reinterpret a

con guration of a quantum eld f b ð x Þ in terms of an assignment of the expectation value of the eld f in the state j to each spatio- temporal point x . Thus the key difference between the classical and quantum cases could be spelled out as follows: while a con gura- tion of a classical eld is a particular assignment of the precise values of the eld to points, a con guration of a quantum eld assigns the expectation (mean) values of this eld. Still, the close analogy between the classical and quantum cases mentioned above should not obscure the fact that fundamentally we are dealing here with two different underlying ontological concep- tions of a eld. To highlight these differences let us compare again the ways classical and quantum elds are built (see Teller, 1995 , pp. 98 e 103). In the classical case we start with a particular eld equa- tion (such as the Klein-Gordon equation), and we write a general solution to this equation, usually in the form of a linear combination of speci c solutions. The coef cients present in this combination are arbitrary functions that can be replaced by speci c numbers when appropriate initial and boundary conditions are imposed (thus the general solution looks more like a schema that can be lled in). This leads to a particular con guration of the eld: an assignment of numbers (vectors, tensors) to spatiotemporal points. In the quantum case, on the other hand, this way of proceeding is blocked from the outset by replacing arbitrary functions in the general solution by appropriate xed operators. The resulting eld operator is a new entity which is no longer a general schema in which we can sub- stitute speci c numbers for the arbitrary parameters, depending on the initial/boundary conditions. Rather, this operator can produce

29 Of course this system of particles is not xed in terms of their exact number, since the numbers of particles associated with particular modes (momenta) is precisely what distinguishes separate states. But this is nothing new e we have already seen that the Fock space formalism in the case of ordinary quantum me- chanics allows for systems with changing, or even inde nite number of particles.

30 Again, in the light of earlier remarks we should note that a more appropriate way of speaking is that f b ð x Þ represents the density of particles added to a given state, so that the actual number of particles added is calculated by integrating it over a given area of space.

expectation values of the eld when coupled with some quantum- mechanical states. However, physical states can't oat in the air; they have to be the states of something. What are they the states of, then? Are the functionals j[ f( x )] the states of the classical eld? But a classical eld is just an assignment of numbers (vectors, tensors) to points, and the quantum-mechanical states in the wavefunctional approach are supposed to be superpositions of such assignments. But it just doesn't make sense to claim that a superposition of as- signments is a property of one of these assignments! An alternative answer may be that the states characterize the quantum eld un-

derstood as the eld operator f b ð x Þ. However, this is inconsistent with the standard role operators play in the quantum-mechanical formalism. Operators are supposed to represent measurable prop- erties (observables) of physical systems, not systems themselves. Thus it looks like we have uncovered an important hole in the eld interpretation of QFT; namely the lack of an entity that can be properly viewed as the physical systemwhose states are given in the form of wavefunctionals, and whose properties are represented by the eld operators. When considering an ontologicalinterpretation of Q FTin terms of elds rather than particles, we should not forget that the ontological status and nature of elds is far from being obvious even in the classical case. The main problem that an ontologist encounters when talking about various elds is how to individuate them. We have already seen that the distinction between elds and their con gu- ration may give rise to legitimate questions. Are different con gu- rations of a classical eld (e.g. different intensities of an electric eld in a particular spatiotemporal region) distinct entities, or just different states of the same underlying object? How about two electromagnetic elds in distant regions brought about by different electric charges? Similarly, we may wonder whether the (classical) superposition of two elds is one object or two objects that merely coincide spatiotemporally. Before a comprehensive eld interpreta- tion of QFT (or any other theory for that matter) can be proposed, questions like these have to be addressed and properly dealt with. As if this wasn't enough, there are arguments showing that the wavefunctional interpretation of QFT inherits some of the dif - culties that affected the particle interpretation in the rst place. As Baker (2009) points out, the existence of inequivalent Hilbert space representations is equally damning for the eld interpretation as for the particle interpretation. The existence of the so-called Rin- dler representation associated with an accelerated observer creates a serious challenge to the particle interpretation of QFT. Because inertial and non-inertial observers will apply different splittings of the solutions into positive and negative frequencies (these split- tings depend on the de nition of time translation, different for accelerating and non-accelerating observers), the resulting Fock space representations will be inequivalent. Consequently, the number operators applied in each representation will produce different expectation values, which leads to the disastrous conse- quence that the number of observed particles depends on the state of motion of the observer. Baker notes that a very similar problem affects the eld interpretation: from the existence of unitarily inequivalent representations it follows that the accelerating and non-accelerating observers will use inequivalent de nitions of a eld con guration. He admits that this does not necessarily imply that these observers will ascribe different eld content to the same state (e.g. the vacuum), because their de nitions of the eld oper- ators are the same ( Baker, 2009 , p. 596). Still, as we recall, the con guration of a eld is given not by the eld operators but by the quantum-mechanical state in which the eld is. And if these states in two inequivalent representations are necessarily distinct, this seems to imply that the con gurations of the corresponding eld are different too.

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1 e

T. Bigaj / Studies in History and Philosophy of Modern Physics xxx (2017) 1e13

13

Baker formulates further challenges to the wavefunctional analysis of QFT based on spontaneous symmetry breaking, as well on the case of interacting elds. We need not go into the details of these arguments, noting only that the conclusion from the above cases is that the eld interpretation is not obviously superior to the particle interpretation. Thus the critical remarks made in this paper with respect to the cogency of the concept of quanta in QFT should not be construed as an indirect argument in favor of the alternative eld interpretation. Perhaps what we need at this stage is a third approach that would break the particle- eld dichotomy. There is no shortage of alternative proposals of ontological interpretations of QFT (e.g. interpretations based on the ontology of events, processes,

or tropes) 31 ; however a detailed analysis of these attempts has to be

saved for another occasion.

7. Conclusion

The main focus of this paper has been on the ontological analysis

of the concept of quanta that arises from the mathematical process

known as eld quantization. I have argued that a wide gap sepa- rates the formal notion of a eld quantum from the ontologically committing concept of a particle (interaction carrier). The idea of introducing quanta of excitation in the case of simple harmonic oscillators, which is the precursor of the procedure of eld quan- tization, is based entirely on the discreteness of the available en-

ergy levels. This formal basis is too weak to support an ontologically robust concept of particle-like quanta equipped with a complete set

of physical characteristics and capable of changing their properties.

Moreover, the introduction of a new type of entities describing excitations of harmonic oscillators seems arbitrary and unmoti- vated at this stage. This situation changes slightly in the case of

coupled oscillators, where the reformulation of the problem in terms of quanta of oscillation living in the reciprocal space offers

a theoretical advantage in the form of decoupled equations of

motion. However, the ontological status of quanta as independent entities remains highly dubious even in this scenario. Moving on to the case of the quantization of classical elds, such as the Klein- Gordon scalar eld, we encounter some formal arguments enabling us to equip the quanta of elds with physical properties

such as mass and position. And yet under closer scrutiny it turns out that these arguments are incomplete at best and plainly misguided

at worst. In addition to this problem, the continuous character of

elds forces us to rethink the standard interpretation of creation and annihilation operators in terms of adding/subtracting indi- vidual particles in favor of something akin to a matter-density interpretation. I have also analyzed the Fock space formalism used in QFT, arguing that the existence of a Fock space represen- tation for a given theory does not automatically guarantee that the theory speaks about particle-like objects. All these arguments suggest, quite independently from the well-known no-go theo- rems, that the particle interpretation of QFT is in serious trouble. Given that the alternative eld interpretation of QFT is in no better position due to its own interpretive problems, it seems that the issue of the proper ontology of QFT will remain an open question for quite a while.

Acknowledgments

I am grateful to James Ladyman and two anonymous referees for their extensive critical remarks to earlier versions of this paper. The

work on this paper was supported by the Marie Curie International Outgoing Grant FP7-PEOPLE-2012-IOF-328285.

References