Ms Dosch Muller Sieroka

Quantum Field Theory,
its Concepts Viewed from a Semiotic Perspective
Hans Günter Dosch1 , Volkhard F. Müller2 ,

and Norman Sieroka3
Abstract
Examining relativistic quantum field theory we claim that its description
of subnuclear phenomena can be understood most adequately from a semiotic
point of view.
The paper starts off with a concise and non-technical outline of the firmly
based aspects of relativistic quantum field theories. The particular methods,
by which these different aspects have to be accessed, can be described as
distinct facets of quantum field theory. They differ with respect to the rela-
tion between quantum fields and associated particles, and, as we shall argue,
should be interpreted as complementary (semiotic) codes.
Viewing physical theories as symbolic constructions already came to the
fore in the middle of the nineteenth century with the emancipation of the
classical theory of the electromagnetic field from mechanics; most notably, as
we will point out, with the work of Helmholtz, Hertz, Poincaré, and later on
Weyl. Since the epistemological questions posed there are heightened with
regard to quantum field theory, we considerably widen their approach and
relate it to recent discussions in the philosophy of science, like structural
realism and quasi-autonomous domains.
1
Institut für Theoretische Physik, Universität Heidelberg, Philosophenweg16, D-69120
Heidelberg, [email protected]
2
Fachbereich Physik der Technischen Universität Kaiserslautern, Postfach 3049, D-
67653 Kaiserslautern, [email protected]
3
Institut für Theoretische Physik, Universität Heidelberg, Philosophenweg16, D-69120
Heidelberg, [email protected]
1 Introduction
In this paper we discuss epistemological implications of relativistic quantum
field theory. The empirical domain of such a theory is formed by phenomena
ascribed to subnuclear particles, sometimes still called elementary particles.
This latter more traditional designation reflects the lasting desire of physicists
to eventually find and isolate irreducible constituents of matter. Going down
to the atomic level, electrons appear to play such a rôle, whereas the nuclei
of atoms can be considered as compound systems of protons and neutrons,
i.e. of two species of particles. This view makes sense, since the respective
number of these two types of constituents essentially identifies an atomic
nucleus. Extracted from a nucleus, however, the ‘free’ neutron is an unstable
particle: it decays spontaneously into a proton, an electron and an anti-
neutrino. In the past fifty years or so basically the bombardment of matter by
protons or by electrons in specially devised experiments has revealed a large
variety of further subnuclear objects. Successive generations of accelerators
and refined collision devices provided higher and higher collision energies.
All these subnuclear objects are termed ‘particles’ in the physics community,
nearly all of these objects are unstable and decay spontaneously into other
ones. The respective lifetimes of the distinct types, however, differ widely,
ranging from relatively long (103 sec) to extremely short (10−25 sec). Because
of this huge disparity in lifetime the notion of a particle deserves particular
attention, a point laid stress on in our consideration. The study of the
physical behaviour of these subnuclear particles led to distinguish three types
of interactions: the strong, the electromagnetic and the weak interaction.
As the names suggest these interactions differ in their respective strength.
Furthermore, each type shows characteristic conservation laws obeyed in the
observed reactions of the subnuclear particles. On the theoretical side the
Standard Model of particle physics has emerged in the course of time. This
striking achievement is supposed to account for the full hierarchy of the
strong, the electromagnetic and the weak interaction.4
Since we are solely interested in firmly based conclusions, we confine our-
selves to mathematically coherent and experimentally very well corroborated
aspects of quantum field theory. Therefore, we focus on various aspects re-
lated to the Standard Model of particle physics and leave out all speculations
4
A very lucid, non-technical overview of the subnuclear world and of basic elements of
its theoretical representation can be found in Veltman (2003).
1
presently in vogue, as e.g. string theory — interesting as they might be. The
Standard Model is considered to essentially describe the realm of subnu-
clear particles up to the current experimental limit energy, probing distances
down to 10−16 cm. We are aware of some indications — both experimental
and theoretical — that the Standard Model should be modified. We be-
lieve, however, that future developments which might crystalise from today’s
more speculative investigations, will fit neatly in the epistemic scheme we
propose, too. This holds also for the variety of partial models, motivated
by the Standard Model but augmented by crucial additional assumptions or
approximations; these models are not considered here either.
Our paper is organised as follows: section 2 presents a concise general
description of relativistic quantum field theories viewed as physical theories,
avoiding technical formulations as far as possible. Various facets are eluci-
dated and distinguished according to differing aims pursued. In section 3
we return in greater detail to the central question: in which way do par-
ticles, i.e. the objects observed experimentally, emerge in the theory from
the quantised fields, i.e. from the theoretical building blocks? The result-
ing rather complex answer manifests, to which extend the theory copes with
the basic phenomenon of relativistic physics, that mass can be transmuted
into energy and vice versa. It is important to notice that our considerations
envisage from the very beginning quantum theories with their probabilistic
physical interpretation, determined by expectation values; hence, we do not
discuss the so-called particle-wave dualism. In section 4 we trace back the
epistemic discussion to the time, when the notion of the classical electro-
magnetic field gained an autonomous status, freed from the attempts of a
mechanical foundation. There, we encounter the birth of the symbolic in-
terpretation of physical theories — notably in the work of Heinrich Hertz.
In section 5 we look at the various facets of quantum field theory from a
semiotic perspective. In section 6 we give a short résumé.
2 Relativistic Quantum Field Theories

Viewed as Physical Theories
2.1 The Empirical Domain
Phenomena ascribed to subnuclear particles and their interaction form the
physical domain of relativistic quantum field theory. Such a subnuclear par-
2
ticle is identified by its mass and spin, which determine its behaviour under
space-time transformations, and by its electric charge and further charge-like
inner quantum numbers. When a stable particle is isolated from external
perturbations it moves freely with constant velocity: its energy E and its
momentum vector p satisfy the relativistic kinematical relation
E 2 = m 2 c4 + c 2 p 2 , (1)
where m is the rest mass of the particle and c the vacuum velocity of light.
Eventually, the particle is recognised when it triggers an appropriate (macro-
scopic) detection device. These recordings constitute the empirical data to
be met by a relativistic quantum field theory.
The salient feature that characterises the interaction of subnuclear par-
ticles is the possible transmutation of matter into energy and vice versa:
particles can be created or annihilated, provided certain conservation laws
are respected — as for instance charge conservation — when a mass m is
converted into its energy equivalent ∆E = mc2 or vice versa. Hence, defi-
nite configurations of particles exist only asymptotically in time before and
after a process of collision, when the particles are well separated and mu-
tually non-interacting, due to the short range of their interaction. Because
of these properties, scattering experiments, where a certain initial state of
particles is carefully prepared and the resulting final states are analysed by
measurements, play such an important rôle in particle physics.
Many of the objects, however, which are conveniently called subnuclear
particles, are not stable but decay spontaneously into lighter particles. Thus,
strictly speaking, they cannot appear in an asymptotic state. Nevertheless,
it is theoretically appealing and proves to be empirically justified to treat an
unstable particle also as forming asymptotic states, provided its lifetime is
large compared with the reaction time in a scattering process. This indicates
already, that the notion of a subnuclear particle is strongly based on the
related theoretical perspective. We shall be confronted with this problem
several times in the paper.
2.2 The Notion of a Relativistic Quantum Field

The basic concept of the theory are quantum fields defined on space-time,
not particles. Space-time is assumed to be a four-dimensional real vector
space with given metrical properties and Einstein causality, such that the
3
Poincaré group (constituted by translations and Lorentz-transformations) is
implied as symmetry group. This space-time structure fixed in advance —
called Minkowski space — forms the register for recording physical events.
The predictions of a relativistic quantum field theory on the outcome of
scattering processes are of probabilistic nature, in this respect similar to those
of (non-relativistic) quantum mechanics. However, a novel feature occurs: in
these processes particles can be created and annihilated. The quantum fields,
in terms of which the theory is constructed, are operators that depend on
space-time and act on the space of physical state vectors. This dependence
on space-time, however, shows the behaviour of a generalised function or
distribution. Phrased technically, a relativistic quantum field is an operator-
valued distribution. Therefore, a well-defined operator cannot be related to a
definite space-time point x, but only to a space-time domain of finite extent:
a proper operator emerges from the quantum field φ(x) by ‘smearing’ with a
corresponding testfunction f (x):5
Z
φ(f ) := d4 x f (x)φ(x). (2)
The very fact, that a quantum field behaves like a distribution, has far-
reaching consequences: in general, pointwise products of distributions are
not defined mathematically. Hence, the nth power φn (x), n = 2, 3, 4, · · · of
a quantum field φ(x) at a given space-time point x is a priori not defined.
In contradistinction, however, such products are well defined in the realm
of classical field theory, since a classical field ϕ(x) is a (generally differen-
tiable) function. Converting indiscriminately such products of classical fields
into the corresponding products of quantum fields produces mathematically
ill-defined objects, from which inevitably infinities emerge when used in the
context of a quantum field theory. Seen historically, quantum field theory
has grown out of classical field theory by the application of heuristic ‘quanti-
sation rules’, extending the profitable rules that converted classical mechan-
ics into quantum mechanics. Moreover, in order to incorporate the causal
space-time structure together with the Poincaré symmetry into the theory, a
Lagrangian formulation of the classical field theory is chosen. The incorpo-
ration is achieved by keeping the functional form of the classical Lagrangian
density — a local polynomial of the field(s) and of its space-time derivatives
— and just replacing the classical field ϕ(x) by the quantum field φ(x). As an
5
The right hand side of the equation is only a formal representation.
4
example we show the Lagrangian density of a model involving only a single
real scalar field:
1 g
L(x) = ∂µ φ(x) ∂ µ φ(x) + m2 φ2 (x) + φ4 (x), (3)
2 4!
where the first part describes a free field and the term involving the cou-
pling constant g the (self)interaction of the field. As explained above, the
ill-defined local products of field operators appearing lead to infinities, if
one attempts to deduce physical consequences from such a ‘theory’. In the
early period of exploring quantum field theories these infinities caused a
considerable amount of bewilderment. Later, a systematic reformulation of
the originally ill-defined approach to relativistic quantum field theory was
achieved — called perturbative renormalisation theory — which provides a
mathematically well-defined quantum field theory as a formal power series in
a renormalised version gren of the coupling constant. This theory produces
strictly finite physical predictions in each order of the expansion. The price
to be paid will be expounded in the sequel.
2.3 Local Gauge Symmetry

In classical electrodynamics, time evolution and the force acting on charged
particles are directly formulated in terms of the electromagnetic field. This
field can be substituted by an electromagnetic potential acting as an ancillary
mathematical object. Two such potentials, however, the difference of which
is a local, i.e. space-time dependent gauge transformation, provide the same
electromagnetic field, and hence generate identical observable consequences.
In other words, the relation between electromagnetic fields and potentials is
not one-to-one: a whole equivalence class of the former corresponds to a given
field, and vice versa. None the less, it is the electromagnetic potential, not
the field, that enters into the Hamiltonian (and the Lagrangian) form of the
classical equations of motion for charged particles. The heuristic correspon-
dence principle — forming a ‘quantisation rule’ — converts this ‘minimal
coupling’ of a particle to the electromagnetic potential into its descendent
appearing in quantum mechanics. By a similar metamorphosis, the minimal
coupling of a charged matter field to the electromagnetic (vector) potential
in classical field theory is finally elevated to the corresponding shape in a
relativistic quantum field theory. The resulting theory is quantum electro-
dynamics (QED). One should notice that such a heuristic theory generating
5
process is neither tautological nor necessarily successful: the arising new the-
ory has to turn out being physically relevant. QED stands this requirement
beyond doubt. On account of its form QED is a quantum gauge field the-
ory, showing a local (Abelian) gauge symmetry. This symmetry signals three
basic properties of such a theory: i) only equivalence classes of the vector
field are physically effective: all observables predicted by the theory, i.e all
quantities having a direct physical meaning, are invariant under local gauge
transformations, ii) the interaction encoded in the theory is local, i.e. there
is no action-at-a-distance, and all physical effects propagate with finite ve-
locity (‘Nahwirkungsprinzip’), iii) performing successively two Abelian gauge
transformations, they commute, i.e. the result does not depend on the order
applied.
Notably, the Standard Model of particle physics is in its entirety a quan-
tum gauge field theory: the respective interaction of the basic matter fields
in each of its three sectors is generated by a corresponding gauge vector
field: the strong interaction by a (non-Abelian) SU (3) -vector field, and the
electroweak interactions by SU (2) × U (1) -vector fields, which, however, be-
cause of spontaneous symmetry breaking only produce an (Abelian) U (1)
-symmetry in the physical state space. This means, that there are three fam-
ilies of space-time dependent inner field transformations, which do no alter
the Lagrangian of the Standard Model.
2.4 Particles versus Quantum Fields

The theoretical description of a relativistic particle in the frame of a quantum
theory was introduced long ago by Wigner6 and does not rely on the notion
of a quantum field. It is rather based on the behaviour of a particle under
space-time transformations, attributing to it a particular unitary irreducible
representation of the Poincaré group connected with mass and spin. This
formulation is also used in the physical state space of a quantum field theory.
To gain clarity on the rôle played by quantum fields it is important to realise
that the connection between the number of fields entering into a quantum
field theory and the number of particles described by it is in general very
indirect. A one-to-one correspondence between fields and particles appears
only in a free field theory or when a model is ab initio formulated perturba-
tively as a formal power series — apart from ghost(!) fields to which we come
6
Wigner (1939).
6
back later. In addition to its dependence on space-time and the associated
symmetry (Poincaré transformations), a quantum field in general carries a
characteristic set of charge-like (additive) quantum numbers, connected with
‘inner’ symmetries. If the Lagrangian of a quantum field theory shows such a
symmetry, the related quantum number provides a conservation law imposed
on the reactions between particles, as e.g. the conservation of electric charge.
In the Standard Model the weak and electromagnetic interaction are
treated perturbatively. Hence, as stated before, the corresponding gauge
and matter fields can be directly related to specific particles, which then are
identified via their decay products. The strong interaction, however, is in
general not accessible by perturbation theory. It is supposed that neither
the matter fields (‘quark’ fields) nor the SU (3)-gauge fields (‘gluon’ fields)
entering the strong interaction part of the Lagrangian, correspond to asymp-
totic particles — even in the restricted sense qualified before. Instead, the
asymptotic particles subject to the strong interaction are supposed to be re-
lated to particular ‘products’ of these fields. This theoretical picture is called
confinement and has not yet been deduced rigorously. But perturbation the-
ory can be used under very restricted kinematical conditions, where only
short distances become relevant. The empirical success of these calculations,
together with model calculations, support the picture drawn for the strongly
interacting particles like proton and neutron. In section 3 we shall further
elaborate on this point.
2.5 The Facets of Quantum Field Theory

In the course of its evolution quantum field theory expanded in several direc-
tions according to various aspects focussed on: the general theory of quantum
fields and local quantum physics consider the basic conceptual frame and
the general properties (consequences) implied. The construction of concrete
models is the subject of perturbative renormalisation theory, of constructive
quantum field theory and of lattice gauge theory. We shall now outline these
facets in turn.
2.5.1 General Theory of Quantum Fields

Parallel to the advent of rigorous perturbative renormalisation theory in the
fifties and sixties, and somehow animated by the clarification achieved there,
vigorous efforts have been made to formulate beyond perturbation theory
7
a mathematically sound general framework of a relativistic quantum field
theory and to analyse it rigorously. The concept of a local quantum field is
preserved as the basic mathematical object, however any recourse to classi-
cal field theory is avoided. The framework is formed by few precisely defined
postulates — usually called Wightman axioms — which give full attention
to the quantum field being an operator-valued distribution. Within this
rather general framework a number of physically important structural con-
sequences have been rigorously deduced.7 In these derivations the locality
property of a field operator plays a major rôle. Among these consequences
we mention the TCP-symmetry, i.e. the combined application of time reflec-
tion T, particle-antiparticle conjugation C, and space reflection P, which is
always a symmetry of a quantum field theory. This occurs irrespectively of
whether an individual symmetry of this triple holds or is violated. Due to it
a particle and its antiparticle have equal mass, an experimentally extremely
well tested prediction. Another result is the theorem on the connection of
spin with statistics — this connection, describing a characteristic property
of microsystems, has to be postulated separately within the realm of quan-
tum mechanics. Furthermore, the Wightman axioms have proven sufficient
to form the basis for a collision theory of the particles accounted for by a
quantum field theory. (The reader is reminded that the physical domain of
a quantum field theory is formed by collision processes of subnuclear parti-
cles.) Theoretically, the Wightman axioms have far-reaching mathematical
implications for the correlation functions of the theory: the Wightman func-
tions. These functions depend by their definition on real space-time vari-
ables, however turn out to allow an analytic continuation in the sense of
analytic function theory to complex values of the space-time variables. Dis-
persion relations for scattering amplitudes follow from the domain attained
by this continuation. Moreover, this domain contains the continuation of
real time values to purely imaginary time values, thus converting the original
Minkowski space into a four-dimensional Euclidean space. Thereby, from the
Wightman functions result the Schwinger functions of the Euclidean formu-
lation of the original quantum field theory. Conversely, from this Euclidean
formulation, the (physical) quantum field theory on Minkowski space can
be reconstructed (Osterwalder-Schrader Theorem8 ). Seen as a mathematical
construction, the Euclidean formulation is based on functional integration.
7
Streater and Wightman (1980), Jost (1965).
8
Osterwalder and Schrader (1973), Osterwalder and Schrader (1975).
8
The Schwinger functions corresponding to bosonic quantum fields emerge as
correlation functions of a probability measure defined on a function space
of generalised functions (distributions). A related well-established Gaussian
measure corresponds to a free quantum field theory and thus the correlation
functions are again explicitly given. In the case of a quantum field theory
showing interaction, however, the measure has to be mathematically con-
structed, substantiating heuristic transformation rules with roots in classical
field theory. In contrast, the Euclidean correlation functions corresponding
to a fermionic quantum field theory emerge from ‘Berezin’-integration,9 de-
fined on a Grassmann-algebra. As the Euclidean formulation allows to apply
powerful mathematical techniques, it strongly stimulated the study of con-
crete models of quantum field theory.
It goes without saying that the theory of free quantum fields satisfies the
Wightman axioms. It is a widespread belief that obtaining a concrete quan-
tum field theory beyond perturbation theory would amount to achieve a
(finite) rigorous mathematical construction showing the properties required
by the Wightman axioms10 or by its Euclidean equivalent.
There is a vast variety of nonperturbative phenomenological models, which
involve elements of the Wightman axioms combined with additional specific
assumptions to approximately deal with selected particular phenomena ob-
served in particle physics.
2.5.2 Perturbative Renormalisation Theory

This approach is based on the well-established theory of quantum fields with-
out interaction — the ‘free theory’ — and treats the physically desired inter-
action as a ‘small perturbation’. The classical Lagrangian, acting as precursor
in the construction of a quantum field theory, contains real parameters as co-
efficients of the field products. In the example presented in eq.(3) these are
denoted by m and g. As a first step towards a finite theory, the naive original
substitution of the classical field ϕ by a quantum field φ is corrected by replac-
ing φ with a renormalised quantum field according to φ = Z 1/2 φrmren and,
moreover, regarding m, g, Z as bare parameters of the theory, devoid of direct
physical meaning. Within the realm of quantum field theory, renormalisa-
9
Berezin (1966).
10
More precisely: these have to be slightly extended in the case of a gauge theory,
however, by introducing a Gupta-Bleuler indefinite metric formulation, in order to account
for the local gauge symmetry — Itzykson and Zuber (1980).
9
tion theory is a method to cure the a priori ill-defined local field products by
disposing of these bare parameters by way of a controlled reparametrisation
procedure. The qualification ‘perturbative’ means that this method is for-
mulated within a formal perturbation expansion of the interacting quantum
fields. As a consequence of the reparametrisation, in each order of the formal
power series expansion all observable quantities of the theory are finite and
infinities never appear, even in intermediate steps. The price to be paid is the
fact, that finite (new) parameters, called renormalised parameters, appear in
the renormalised quantum field theory. However, they are undetermined by
the theory and have to be fixed by comparing the theory with experiment.11
In a somewhat metaphoric language we can say that the bare parameters,
banished by the renormalisation process, have carried with them the infini-
ties inherent in the local products of field operators. It is the central result
of perturbative renormalisation theory in four space-time dimensions, that
in the class of renormalisable quantum field theories this number of finite
undetermined parameters equals the number of bare parameters.
In retrospect, the era of a systematic perturbative renormalisation theory
began with Dyson’s pioneer work.12 After a considerable evolution the rigor-
ous BPHZ-version of perturbative renormalisation theory was accomplished
in the late sixties of the last century.13 There are also technically differ-
ent, however physically equivalent formulations. Moreover, an extension to
cover massless fields was also achieved, emphasised by the enlarged acronym
BPHZL.14 Technically, the BPHZL-method and its relatives make essential
use of the invariance of a theory under translations on space-time. In con-
trast, Epstein and Glaser developed an inductive construction on Minkowski
space,15 strictly based on the locality property of a field operator, which is
treated ab initio properly as an operator-valued distribution. A finite pertur-
bation expansion (as formal power series) is established directly by a method
of distribution-splitting. Since this construction does not rely on translation
invariance, it can be applied when on a quantum field also acts an exter-
nal (classical) field,16 or in the case of a quantum field theory on curved
11
In the example considered: mren , gren , Zren .
12
Dyson (1949b), Dyson (1949a).
13
Bogoliubov and Parasiuk (1957), Hepp (1969), Zimmermann (1970). The acronym
above refers to the initials of the authors quoted.
14
Lowenstein (1976).
15
Epstein and Glaser (1973).
16
Dosch and Müller (1975).
10
space-time in place of Minkowski space.17 With regard to both its technical
complexity and its physical importance for the Standard Model of particle
physics, perturbative renormalisation theory culminated in the demonstra-
tion by ’t Hooft and Veltman that also spontaneously broken non-Abelian
gauge theories are renormalisable.18
One notices, that the complete perturbative construction of the quantum
gauge field theories could be achieved only at the expense of introducing
also unphysical degrees of freedom. At least up to the present, these un-
physical degrees of freedom seem to prove technically indispensable in order
to implement the local gauge symmetry and simultaneously maintaining the
full Poincaré space-time symmetry. The recourse to these additional degrees
of freedom leads to an enlarged state space with indefinite metric, which
embraces the space of physical states as subspace showing a positive met-
ric. Hence, no negative probabilities emerge for physical processes; stated
technically: the scattering matrix (‘S-matrix’) remains unitary. Regarding
these unphysical degrees of freedom more closely one observes, that in the
Abelian case (QED) they are precisely formed by the pure gauge degrees
of freedom of the vector (potential) field. The non-Abelian SU (2)-theory,
however, in addition makes use of auxiliary fields: a Faddeev-Popov ‘ghost’-
and an ‘antighost’-field.19 These are veritable scalar quantum fields — albeit
with fermionic statistics, thus violating the connection of spin with statistics
to hold in the case of physical degrees of freedom.
Perturbative renormalisation theory is the approach to quantum field
theory which provides its most distinctive predictions. The very high degree
of agreement between specific experimental results and their theoretical de-
scription due to QED made a strong empirical case for the physicists’ belief
in the method of perturbative renormalisation. In a wider perspective, the
fact that physicists adhere to relativistic quantum fields as a useful physical
concept can be attributed to this outcome. With the electroweak sector of
the Standard Model perturbatively renormalised quantum field theory pro-
vides another striking instance of a physically relevant theory — although at
the price of introducing the yet to be discovered ‘Higgs-particle’ generating
spontaneous symmetry breaking.
Notwithstanding all this conspicuous success, a penetrating evaluation of
17
Brunetti and Fredenhagen (2000).
18
’t Hooft (1971b), ’t Hooft (1971a), ’t Hooft and Veltman (1972).
19
Faddeev and Slavnov (1991).
11
the theory cannot fail to notice, that it relies (heavily) on unmet theoret-
ical issues. The theories are constructed as formal power series only, with
every indication that these series are divergent. In addition, the extent, i.e.
the highest order of the power series worked out, is dictated by the human
limit to perform the actual calculations involved rather than by a theoret-
ical estimate of the order needed to analyze the given experimental data.
There is a widespread opinion that the series have the mathematical status
of an asymptotic expansion. Whether the divergent series proves to be Borel-
summable in concrete cases and thus yields mathematically the exact result
is important and yet unanswered. Nevertheless, even an obstinate sceptic
among the physicists would in full view of these intriguing questions hesitate
to downgrade perturbative quantum field theory to the level of a mere set of
very efficient rules.
2.5.3 Constructive Quantum Field Theory

This approach was prompted by the mathematically unsatisfactory situation
that the formal renormalised perturbation series in general do not appear
to be summable, i.e. do not represent a well-defined theory. Therefore,
the rigorous construction of a quantum field theory showing interaction was
still an open problem. It was via functional integration in the Euclidean
formulation, that these shortcomings could be surmounted to a notable ex-
tent. By the development of elaborate convergent cluster expansion methods
the rigorous construction of various bosonic quantum field theories in two
and three space-time dimensions has been achieved, demonstrating that the
concept of a quantum field theory allows a mathematical realisation outside
free fields — albeit not yet on the physical (four-dimensional) space-time.20
Seen technically, these theories are super-renormalisable, i.e. are less singu-
lar than just renormalisable ones, but still require renormalisation, achieved
nonperturbatively within their rigorous construction.
Moreover, by shaping Wilson’s intuitive ideas of the renormalisation flow, 21 ,
into an inductive mathematical method, a particular fermionic quantum field
theory in two space-time dimensions: the Gross-Neveu model, was rigorously
constructed.22 This model has two distinguished features: it is a genuine
20
See Glimm and Jaffe (1987).
21
Wilson and Kogut (1974).
22
Gawedzki and Kupiainen (1985), Feldman et al. (1986), Disertori and Rivasseau
(2000).
12
renormalisable theory (i.e. not being super-renormalisable) and it is ultra-
violet asymptotically free.23 Due to these properties the model has some
similarity with a non-Abelian gauge theory in four space-time dimensions,
the latter playing a central physical rôle. This similarity probably elevates the
Gross-Neveu model above the status of a mere toy model. It is worth pointing
out that the rigorously constructed correlation functions of the Gross-Neveu
model are also shown to be equal to their Borel-summed renormalised per-
turbation series. This means that the theory is mathematically determined
by its perturbation expansion.
2.5.4 Lattice Gauge Theories

A distinctive method to construct a quantum gauge field theory is lattice
gauge theory, initiated by Wilson.24 It is based on the Euclidean functional
integral approach to construct a quantum field theory and does not resort to
perturbation theory, i.e. keeps away from a formal power series expansion.
This method starts with replacing the Euclidean space-time continuum by a
four-dimensional lattice of points with spacing a (a hypercubic lattice) and
discretises the theory in a theoretically appealing form. Basically, the matter
fields are attached to the sites, whereas the vector gauge fields are connected
with the links between nearest neighbour sites of the lattice. This set-up is
built out of extended, i.e. not point-like, gauge-invariant objects. Its dis-
tinctive feature is to fully preserve the local gauge symmetry and, moreover,
to avoid the introduction of unphysical degrees of freedom, like ghost fields.
The Euclidean correlation functions of the lattice theory are mathematically
well-defined. The problem thus posed is to achieve a physically sensible con-
tinuum limit of them, decreasing the lattice spacing a to zero. Although
analytically inaccessible, too, the lattice theory can be studied by numerical
simulation methods. These simulations form essentially an exact algorithm
on finite lattices, the envisaged physical continuum values, however, can only
be reached via an extrapolation.
Lattice gauge theory is presently the only approach to explore construc-
tively a physically relevant gauge theory with strong coupling. The strong
interaction sector of the Standard model, called ‘nonperturbative quantum
chromodynamics (QCD)’, is such an instance. There, the foremost problem is
to deduce the variety of observed asymptotic particles (mesons and baryons)
23
See section 2.6.
24
Wilson (1974).
13
from the basic degrees of freedom of the theory, that is from the matter fields
(quarks) and vector gauge fields (gluons) which lack an asymptotic particle
content. Already numerous physical quantities — including some particle
masses — have been computed using the lattice theory.25 The extrapolation
involved to vanishing lattice spacing a is based on the observation, that di-
mensionless quantities computed with sufficiently small a become less and
less dependent on a, suggesting a limit to exist at vanishing a. The results
obtained up to now are rather promising and appear to confirm that quantum
chromodynamics is an adequate theory of strong interaction phenomena.
2.5.5 Local Quantum Physics

This relatively recent name replaces the former ‘algebraic quantum field the-
ory’, being more in accord with the theoretical approach denoted. The goal
of local quantum physics is to develop a coherent conceptual frame and a
corresponding consistent mathematical structure serving as the foundational
basis of quantum field theory.26 The basic concept of local quantum physics
is the net of local observables, i.e. theoretical constructs that are directly
related to measurements. In more detail: to every bounded space-time re-
gion is attributed an algebra of operators describing the physical observables
of the region. The particular physical system manifests itself in the partic-
ular relations to hold between the algebras of different regions. The salient
feature of this approach is the principle of locality: two operators represent-
ing observables of space-like separated regions commute. This principle is
sometimes called ‘Einstein causality’, as it also restricts the propagation of
quantum effects not to exceed the speed of light. Quantum fields have no
basic status in local quantum physics, since they do not directly represent
physically observable quantities. They act only as particular building blocks
of observables. In this sense, their rôle has some similarity with coordinates
in describing a geometrical configuration.
Among the principal achievements of local quantum physics one finds the
theory of superselection structure inherent in relativistic quantum theories
and a general property (KMS-condition) of such theories in the case of fi-
nite temperature. As already expounded, all three sectors of the Standard
Model are invariant under separate respective local gauge transformations of
the quantum fields that compose this model. By definition, observables are
25
See Montvay and Münster (1994).
26
A recent review can be found in Buchholz and Haag (2000).
14
invariant under these transformations. Hence, in a theory strictly based on
the concept of observables such transformations remain off stage. In view of
the physical pertinence of the principle of local gauge invariance, the alge-
braic scheme has probably to be augmented in order to incorporate this vital
principle — a task not yet fully accomplished.
2.6 Renormalisability versus Effective Field Theories

The singular properties of a relativistic quantum field emerge from assuming
space-time to be continuous and requiring Einstein causality to hold. These
properties prohibit to form local products of quantum fields directly, but
require a modified construction. There are various theoretical arguments
that at very small distances of the order of the Planck length λP ≈ 10−33 cm
our mathematical model of space-time as a continuous classical manifold
(Minkowski space) is untenable. It does not seem overly rash to doubt that
such small distances can ever be probed by a direct experiment. Observing
that present day experimental particle physics reaches ≈ 10−16 cm and is fully
compatible with the Minkowski space as theoretical space-time structure, to
maintain the latter on all scales appears to be a useful idealisation. Idealising
extrapolations are a constitutive element of any physical theory: the classical
field theories of elasticity and of a fluid, e.g., are successful physical theories
in their respective domains of validity, both characterisable by a length of
about 10−1 cm and larger. They ignore completely the atomic structure of
matter with its characteristic length of 10−8 cm.
Adhering to local quantum field theories, a particular procedure — called
renormalisation — is employed to define in a mathematically controlled man-
ner powers of the singular quantum fields. Within this realm, renormalisable
quantum field theories are distinguished by the property that this renormal-
isation procedure introduces only a finite number of free parameters into the
theory, which then have to be determined from outside, e.g. by comparison
with measurements. Moreover, these theories are internally consistent from
largest down to smallest distances. Although the less restrictive concept of
an effective quantum field theory could suffice in principle to meet the de-
mands of empirical adequacy, it is a fact that the traditional preference for
the criterion of renormalisability has created the Standard Model, considered
to be the fundamental theory of present particle physics. It is worth pointing
out that in particular the required renormalisability of the Standard Model’s
electroweak sector provides a stringent condition on the field content of this
15
theory. The condition originates in the peculiarity — traditionally called
the anomaly phenomenon — that a symmetry shown by a given classical
precurser theory can be inescapably violated by the quantisation procedure.
To maintain in the above-mentioned theory local gauge symmetry in con-
junction with renormalisability demands the absence of anomalies. This is
achieved, if the number of quark fields matches the number of lepton fields
— a prediction well in accord with present day experiments.
Within the family of renormalisable quantum field theories a highly selec-
tive criterion is the property that such a theory is ultraviolet ‘asymptotically
free’.27 Given this property the renormalised coupling of the theory becomes
weaker and weaker with growing energy. There are strong arguments that
a renormalisable quantum field theory can be well-defined (outside formal
perturbation theory) only, if it is asymptotically free. In four-dimensional
space-time, solely non-Abelian gauge theories show this distinguished prop-
erty.28 Although a strictly nonperturbative construction of a renormalisable
relativistic quantum field theory in four space-time dimensions could not
be achieved up to now, some insight into this challenging problem can be
gained by inspecting the formal power series generated by the perturbation
expansion as a whole. These respective series are not convergent but di-
verge strongly: at large order n their coefficients grow proportional to n!,
indeed. In the case of an ultraviolet asymptotically free theory, however,
these coefficients appear with an alternating sign (−1)n , whereas all coef-
ficients have the same sign, if the theory is not asymptotically free. As a
consequence, the divergent expansion of an asymptotically free theory might
be Borel summable and thereupon would in the end generate the theory as a
well-defined mathematical structure. In contrast, if (ultraviolet) asymptotic
freedom is absent, the mathematical status of the formal power series remains
obscure. It would be fallacious to conclude that quantum field theories not
involving non-Abelian gauge fields — QED is the foremost example — are
physically meaningless. Although doubtful to exist as separate theories on
a nonperturbative level, their perturbative construction might nevertheless
prove physically sensible. As regards QED, its empirical success is incon-
testable.
To broaden the scope of quantum field theory, recently effective quantum
27
Gross and Wilczek (1973), Politzer (1973).
28
Provided the number of fermionic matter fields entering the theory does not exceed a
certain limit.
16
field theories have received growing interest. One may roughly distinguish
two approaches: a systematic one emerging from an internally consistent
quantum field theory, and a pragmatic, exploratory one, conceiving new the-
ories with inherent restrictions. We first describe the former approach. Given
a perturbatively renormalised field theory which describes the interaction of
several particle species having different masses, it is quite intuitive that in
the low energy domain of the ‘light’ particles their interaction with the heavy
ones does not play an important rôle. A quantitative account thereof fur-
nishes a decoupling theorem.29 Employing Wilson’s renormalisation group
approach it could be substantially extended.30 We describe these results in
some detail because of their paradigmatic rôle. They are derived in the realm
of renormalised perturbation theory to any order (of the formal power series).
Let the full theory involve two interacting fields with a light mass (m) and
a heavy one (M ), respectively. One compares the correlation functions of
the light particles, derived from the full theory, with the corresponding cor-
relation functions, derived from a truncated version of the theory, where the
field of mass M has been deleted. In the simple form of the decoupling theo-
rem the difference of two such corresponding correlation functions has at low
energies the magnitude (m/M )2 (log(M/m))ν , where ν is a natural number.
However, an extended renormalisable quantum field theory with the field of
mass m alone can be constructed such that its outcome for the low energy
processes of the light particles only differs from the corresponding results
following from the full theory in magnitude (m/M )2n (log(M/m))ν , where n
is any natural number. To this end one has to take into account not only
the correlation functions of the field of mass m already considered, but also
correlation functions of this field with insertions of ‘irrelevant’ local operators
formed with this field. The correlation functions of the refined theory are
then given as linear combinations (depending on the desired degree n of accu-
racy) of correlation functions with appropriately chosen operator insertions.
In this way, the presence of the field with the heavy mass M in the full theory
is accounted for in the theory involving only the field with the light mass m
alone, up to the accuracy stated. One should notice that the irrelevant local
terms are treated as insertions and are not introduced into the Lagrangian,
thus keeping the refined theory renormalisable. QED can be seen in this
perspective as a subtheory of the Standard Model’s electromagnetic sector.
29
Appelquist and Carrazone (1975).
30
Kim (1995).
17
In the second approach, the pragmatic endeavour, effective quantum field
theories are intentionally devised towards a more or less narrow empirical do-
main. These theories in general cannot be deduced in a strict sense from an
embracing more fundamental theory. They are not intended as fully fledged
quantum field theories, but to account for certain low energy phenomena.
Hence, they can tolerate a persistent ultraviolet cutoff or non-renormalisable
interaction terms. Fermi’s original theory of nuclear β-decay and its later
extension to the V - A theory of weak interactions,31 or Weinberg’s phe-
nomenological approach to chiral dynamics32 are typical examples of such
theories.
3 Particles and Fields

The objects of particle physics are subnuclear particles, their properties and
their interactions. All experimental information we have on these particles
comes from scattering experiments (in the widest sense). As outlined in the
previous section quantum field theory is the adequate theory to describe these
scattering processes. Any epistemic interpretation of quantum field theory is
therefore confronted with the problem of relating the field with the particle
concept.33 In the following case study we want to show how delicate this
relation is, here we shall only use internal, that is physical criteria, but we
shall show in the following sections that semiotic concepts are useful for a
meaningful interpretation, since the definition of a subnuclear particle is by
no means obvious; we first state different possible approaches.34
• If one insists that there is a definite relation between energy and mo-
mentum, as in Wigners general definition of a particle, the invariant
mass of the particle must have a definite value, see eq. (1). This in turn
implies that the particle must be stable. Only in that case there ex-
ists a definite invariant mass. Unstable particles show a certain spread
in the mass, the so called mass width. Under the severe restriction
31
Feynman and Gell-Mann (1958).
32
Weinberg (1979). For more recent developments in a completely different context see,
e.g., Intriligator and Seiberg (1996).
33
As mentioned in the Introduction we do not have in mind here the so called ‘particle-
wave’ dualism encountered in quantum mechanics.
34
For somewhat different approaches see Cao (1996).
18
of stability only the electron, the proton35 together with the respective
antiparticles, and the photon qualify as subnuclear particles, since they
are stable to our knowledge.36
• Since particles are the fundamental objects as far as experiments are
concerned, a practical definition might be based on the requirement
that the lifetime of a particle is long enough to construct beams of these
entities and to experiment with them. In this way besides electrons,
protons and photons also the π-mesons and K-mesons and hyperons
e.g. qualify as particles.
• A less stringent definition only demands a ‘rather sharp distribution
of the rest mass’ or — expressed more formally — a pole appearing
in a corresponding scattering amplitude not too far from the real axis.
This was a point of view much advocated in the sixties and explains the
entry ‘Particle listings’ in the very important annual Review of Parti-
cle Physics.37 This definition has a rather fuzzy frontier as becomes
already clear from the appearance and disappearance of ‘particles’ in
this listing.
• A definition of a particle which is better suited for the theoretical frame-
work is the following: We call a particle an object which were stable
if weak and electromagnetic interactions were absent. In this way also
the intermediate bosons with a width of around 2.5 GeV, corresponding
to lifetimes of around 3 · 10−25 seconds, qualify as particles, whereas
the heavy J/ψ-meson with a width of 0.000084 GeV, corresponding to
the much larger lifetime of 7.5 · 10−21 seconds, does not.
This shows that questions like ‘What is a particle?’ or ‘Does such-and-
such particle exist?’ can only be answered in a given context of theoretical
concepts and experimental feasibility. We will provide further interpretation
of this feature in the paper’s last section.
Moreover, the notions of fundamental and composite particles can be
introduced only in a given theoretical context. Rutherford proposed in 1920
35
A possible finite lifetime longer than the age of the universe does not need to concern
us here.
36
We do not dwell on the so called infrared problems, which are related to the interaction
of charged particles with very low energetic photons, since these problems are solved in a
very compelling model (Bloch-Nordsiek).
37
Hagiwara and others (2002).
19
to search for a tight bound state of a proton and an electron which would be
electrically neutral. The search was successful, the neutron was discovered
1932 by Chadwick, but nobody would nowadays consider the neutron as a
bound state of an electron and a proton, though it decays into a proton, an
electron and a light neutral particle, the antineutrino. Also the Z 0 meson
is not considered as a bound state of an electron and a positron, though
it decays, among others, into these particles. On the other hand, scientists
call the J/ψ-meson a bound state of a quark and an antiquark (the so-
called charmed quarks), though it cannot be separated into these objects.
The question of compositeness is in our present understanding very closely
connected with the relation between the concepts ‘particle’ and ‘quantised
field’.
As mentioned in the preceding section, the theoretical framework of par-
ticle physics is quantum field theory and we have to interpret the relation
of the basic theoretical concepts, the quantum fields, and the entities with
which experiments are done, that is particles. This relation is only explicit in
the model of a free field theory, that is a field theory without any interactions,
see section 2.4. In a free field theory the field operator is directly constructed
in terms of creation and annihilation operators of a given particle-antiparticle
species. In view of the rather simple renormalisation rules for the product of
two field operators in a free theory this means that in practice the particle
properties can be read off from the classical Lagrangian. Free theories are
unrealistic, however, and treating interaction by perturbation theory is by far
the most developed part of relativistic quantum field theory. Here again we
can first consider the so called tree approximation, where the simple renor-
malisation procedures of the free theory are applied. It corresponds to the
lowest order in the perturbative expansion. In this approximation a unique
identification of particles with fields is also possible. For the leptonic part
of the Standard Model of electroweak interactions the quantum fields occur-
ring in the free part of the Lagrangian correspond to the ‘observed’ particles.
Whereas for the gauge bosons Z 0 and W ± the extended definition allowing
for a certain energy width has to be applied in the particle definition. The
fact that already the tree approximation yields good quantitative results for
the width and that the higher order corrections using renormalised perturba-
tion theory give excellent agreement with experiment shows the fruitfulness
of that approach.
The situation is much less clear for the strong interaction. Progress in the
last 30 years resulted from abandoning the concept that the observed parti-
20
cles are directly related to quantum fields occurring in the theory. Rather it
is hoped that the properties of the observed particles (hadrons) can be calcu-
lated within a theory the basic fields of which do not correspond to particles.
These basic fields are the quark and gluon fields. This new concept has roots
in four different facets of quantum chromodynamics and can lead to quite
different concepts of ‘elementary particles’:
• 1) Perturbation theory: In certain reactions and under certain kine-

matical conditions a hadronic amplitude can be separated (factorised)
into a long range part which is not accessible to perturbation theory
and a short range part which is. In that way, for instance, relations
between scattering amplitudes of particles at different kinematical con-
ditions can be calculated and compared with experiments. This has
been done very successfully in deep inelastic scattering experiments.
Highly energetic electrons or positrons are scattered off protons and
the calculated dependence of the cross sections on (high) momentum
transfer is compared with experiment. Another example are the so
called jet events, where the topology of the produced observed particles
reflects that of the quarks and gluons; again a quantitative comparison
with experiment is possible here. The generally accepted ‘proof for the
existence of a gluon’ by observing three jets in the annihilation of an
electron and a positron with very high energies is based on such cal-
culations. It should be emphasised that the sophisticated framework
of renormalised perturbation theory is essential for such a simple as-
sertion that the structure of three particle jets reflects the creation of
a ‘quark, an antiquark and a gluon’. Such three jet events formed by
hadrons were observed for some time, but could not safely be related
to an elementary process involving three fundamental fields. In pertur-
bative quantum chromodynamics all amplitudes are calculated on the
basis as if the quarks and gluons were particles. It is, however, noted
that, in contradistinction to the electroweak theory, the perturbatively
calculated strong scattering amplitude is the more affected by uncon-
trollable nonperturbative terms, for instance by power corrections, the
nearer one comes to the singularities induced by the quark propaga-
tors. It is nevertheless possible — but by no means compelling — to
extend the concept of a particle. In perturbation theory there is a well
defined reciprocal correspondence between particles and fields as men-
tioned above. If perturbation theory can be applied to certain reactions
21
one could say that the existence of the particles corresponding to the
fields manifests itself there. We should, however, emphasise that this
definition has no consequence whatsoever. The statement quarks and
gluons are ‘particles’ is synonymous with the following: the perturba-
tively renormalised QCD Lagrangian containing quark and gluon fields
leads to certain consequences which can be compared with experiment.
It should be noted that, according to this definition, also the so called
ghost fields would be promoted to describe ghost particles, since their
occurrence in covariant renormalised quantum field theory is necessary.
• 2) Lattice gauge theory: There is up to now no analytic way to treat

a realistic quantum field theory like the Standard Model in a nonpertur-
bative way. The only attempt in this direction is lattice gauge theory,
see section 2.5.4. Here, observed hadron masses can be obtained indeed
from numerical calculations based directly on the QCD Lagrangian, but
only in the case of finite lattice spacings. The limit to the continuum,
that is to full Lorentz invariance, has not yet been established math-
ematically. The numerical results are very encouraging, however, and
many conjectures made in nonperturbative quantum chromodynamics
can be checked quantitatively. In this approach the quarks and gluons
enter essentially as fields but have no particle properties whatsoever,38
since the propagators of single quarks and gluons vanish due to gauge
invariance.
• 3) Non-relativistic quantum field theory: After the discovery of

heavy mesons like the J/ψ and Υ it became clear that the hypothesis
that these mesons are composed of two quarks leads to similar precise
predictions as in atomic physics. One treats the quarks within the non-
relativistic Schrödinger equation like particles bound by a potential and
can calculate corrections to the nonrelativistic results as power series
Λ
in QCDM
where ΛQCD is some typical hadronic scale and M the mass
of the heavy quark. It should be noted that simple model calculations
are of typical back-of-an-envelope type, but that for reliable predictions
the full machinery of effective renormalised perturbation theory must
be applied.
38
Even after analytic continuation due to Minkowski metric.
22
• 4) General theory of quantum fields: From the Wightman pos-
tulates there follows no direct relation between the quantum fields en-
tering a theory and its particle content. The latter is supposed to
be a consequence of the inherent interaction and to manifest itself in
particular properties of the Wightman functions.
4 Theories of Signs and Symbols, and Struc-

tural Realism
The antagonism in physics between a particle as typical ingredient of a mech-
anistic interpretation of nature and the more formal concept of a field dates
back to the middle of the 19th century. The ongoing success of Newto-
nian and Euler-Lagrangian mechanics led to a widely accepted mechanistic
Weltbild during the 19th century. This Weltbild, however, came to a crisis
through the advance of electrodynamics which later culminated in the es-
tablishment of Maxwell’s equations. Ironically, the essential new ingredient
and cornerstone of these equations, the displacement current, was derived
in a mechanistic model39 of the æther, the assumed carrier of electromag-
netic phenomena. Therefore electrodynamics was formally embedded in the
mechanistic Euler-Lagrangian field theory. It turned out, however, that a
mechanistic interpretation of the æther became less and less tenable and so
the concept of a field was emerging as a map of space-time points to measur-
able quantities without assuming a special carrier of the field. This concept
existed besides the old concepts of mechanics and allowed a perfectly ade-
quate description of electrodynamic phenomena. Since many of the epistemic
problems related to quantum field theory started with the raise of electrody-
namical field theory in the middle of the 19th century it is not astonishing
that also many of the philosophical concepts still important to tackle mod-
ern problems came up at the same time.40 This is particularly true for the
39
Here and in the following we mean ‘model’ to refer to a set of assumptions adapted to
phenomenological description rather than something strictly derived from a set of math-
ematical postulates. The latter we would call a theory. For some examples of theories,
models (and effective field theories) in the context of quantum field theory see Hartmann
(2001).
40
Presumably this is due to the fact that along with this transition the fundamental
entities involved changed quite radically (going from particles to fields) so that even basic
Newtonian principles like ‘action equals reaction’ got lost, see e.g. Poincaré (1905), VIII,
23
‘structural realism’ which in analytic philosophy is generally associated with
the work of Henri Poincaré.41 Thus, before talking about the history in more
detail, let us briefly introduce the modern origin of structural realism; in par-
ticular since it is — in different ways though — invoked by several writers in
the interpretation of quantum field theory at the moment.42
The modern origin of ‘structural realism’ lies in the tension between what
is called the miracle argument and the pessimistic meta-induction. The mir-
acle argument expresses the idea that scientific realism is the only philosophy
of science that can explain why certain theories are so successful in their pre-
dictions whereas this fact remains a miracle for all those accounts that view
theories and the entities they deal with as constructions of our mind in some
sense. For why should our mental constructions depict natural phenomena
so well? The pessimistic meta-induction on the other hand is the inference
that, since all former scientific theories have turned out to be wrong if ap-
plied generally, our current theories will also turn out to be wrong . This,
of course, is particularly worrying for the realist. Thus his concern becomes
this, in the words of John Worrall : ‘is it possible to have the best of both
worlds, to account (no matter how tentatively) for the empirical success of
theoretical science without running fool of the historical facts about theory
change?’43
Structural realism about science is the view that most physical theories
are structurally correct. This is not to say — as entity realism does — that
physical theories mirror the objects of a mind-independent world directly but
merely that the nature of the linkages they introduce is true. The easiest
way to understand what is meant here by ‘linkage’ or ‘structure’ is to have a
look at the mathematical equations that are involved. Often old theories are
mathematically limiting cases of new theories — the notorious example in
the literature being the transition from Fresnel to Maxwell in the description
of optical phenomena. Thus, structural realism seems to be more cumulative
Le principe de Newton.
41
Although the term ‘structural realism’ itself was presumably coined in papers by
Grover Maxwell in 1970. However, Maxwell developed his approach by considering later
works of Russell and by using Ramsey-sentences (i.e. an entity is ‘whatever it is’ that
satisfies the relations held by predicate variables in the Ramsey-sentence).
42
Quite straightforwardly, for instance, in Cao (2001), and more formally elaborated
(along the lines of Hermann Weyl to which we will also come back later), for instance, in
Auyang (1995).
43
Worral (1996), p. 151.
24
and more promising in coping with the pessimistic meta-induction.
A further reason why talking about structures or relations seems funda-
mental for physics is that our experiments find out about the way in which
the things under investigation act, i.e. the way they impinge on our measur-
ing devices. Hence, one has to acknowledge the possibility that we might be
unable to distinguish between two quite different intrinsic properties, namely
if they are exactly the same with respect to the way they impinge on our
instruments. By the same token we must then accept the possibility that we
know next to nothing about the intrinsic nature of physical objects. Also the
fact that we normally observe coherence in properties, that certain structural
properties seem to be tied together, is of little help here.44 For the fact that
some relational properties come in bundles is no guarantee for that bundle
to describe the intrinsic nature of a physical entity.
In the following we give a short historical sketch of concepts which we
think are not only the roots of structural realism45 but are also the basis
for a semiotic approach going further than structural realism. We start with
the theory of signs (symbols) of Hermann von Helmholtz and Heinrich Hertz
and then find some further development in the works of Henri Poincaré and
Hermann Weyl. Obviously we do not give this historical reconstruction only
for its own sake, but because we think that in comparison to structural
realism it will provide us with a more appropriate perspective to evaluate
quantum field theory as a physical theory. Thus, this paper’s implicit critique
on modern analytic structural realism is indeed a systematic rather than
a historical one. Of course, we think it to be important to ‘get history
right’, but as a critique against Worrall etc. this would grasp at thin air to
some extend. For although Worrall and others mainly discuss Poincaré they
admittedly did not attempt to give a historical reconstruction of the origin
of structural realism.
There was a geographical distribution of the prevalence of either the mech-
anistic or the more formal epistemological concept of physics. The former one
was especially flourishing on the British islands whereas the latter one was
more popular on the continent, notably in France and Germany. This dif-
ference between the British and continental epistemolological concept is ex-
tensively discussed by Duhem in his book La Théorie Physique.46 Helmholtz
44
Such an argument has been brought forward, e.g., in Chakravartty (2002).
45
See also Cao (1996).
46
Duhem (1914 1957), I,IV Les théories abstraites et les modèles mécaniques.
25
writes that ‘English physicists like Lord Kelvin and Maxwell’47 were more
satisfied by mechanical explanations than by the ‘plain most general expla-
nation of the facts and their laws as they are given in physics by the systems
of differential equations.’48 He himself, however, adhered to this formal rep-
resentation and felt most assured by that.
We have mentioned the argument brought forward in favour of structural
realism that we know next to nothing on the intrinsic nature of physical
objects, but only how they act on our measuring devices. Helmholtz antic-
ipated this argument by applying it already to our sensual receptions;49 in
that he was largely influenced by the Sinnesphysiologie of Johannes Müller.50
Helmholtz writes:
Our sensations are actions which are evoked by external causes
in our organs, and how such an action manifests itself depends of
course essentially on the nature of the apparatus which is acted
on. In so far as our sensation gives us a message of the peculiarity
of the evoking external influence it can be accepted as a sign of
it but not as a copy.51
The essential features of his theory of signs can be found in his ‘Handbook
of Physiological Optiks’ the most elaborate version is in his lecture ‘Die
Thatsachen in der Wahrnehmung’ where the above quotation is taken from.
Though he insists that our sensual impressions are only signs they are not
to be disposed of as empty ‘Schein’ but they are signs of something, be it
existing or happening, and what is the most important, they map the law of
what is happening. For Helmholtz the relevant feature of science is not the
particular set of signs, but that what he calls ‘law’ and which he defines as
the unchanging relation between changing variables (signs). In this respect
his above quoted assessment of the differential equations as the more assuring
feature of a theory becomes clear.
47
Of course, both are Scottish.
48
Foreword to Hertz (1894).
49
Recently the same point has been made in Psillos (2001), p. S14.
50
Helmholtz (1892), Helmholtz (1921).
51
‘Unsere Empfindungen sind Wirkungen, welche durch äussere Ursachen in unseren
Organen hervorgebracht werden, und wie eine solche Wirkung sich äussert, hängt natürlich
ganz wesentlich von der Art des Apparates ab, auf den gewirkt wird. Insofern die Qualität
unserer Empfindung uns von der Eigentümlichkeit der äusseren Einwirkung, durch welche
sie erregt ist, eine Nachricht gibt, kann sie als ein Zeichen derselben gelten, aber nicht als
Abbild.’ Helmholtz (1892), vol II, p. 226.
26
Helmholtz did not try do develop an elaborate epistemological scheme,
his ideas were rather intended to give a basis to his proper work which em-
braced physiology, physics and mathematics. His main concern is a warning
against drawing metaphysical conclusions from results of natural science and
attaining a dogmatic stance on them. He concedes that hypotheses are nec-
essary for scientific research and even more for human action. In this sense he
regards also metaphysical hypotheses as possibly legitimate. But he warns:
‘But is unworthy of a thinker who wants to be scientific if he forgets the
hypothetical [. . . ] origin.’52
Heinrich Hertz was a student of Helmholtz. He contributed largely to the
development of electrodynamical field theory, both by his experimental and
theoretical investigations. Einstein, for instance, referred to the ‘Maxwell-
Hertz theory’ of electrodynamics in his writings. Like Helmholtz, Hertz was
more satisfied by mathematical equations than by mechanistic pictures. On
Maxwell’s equations he writes admiringly: ‘One cannot read this beautiful
theory without sometimes feeling as if those mathematical formulæ‘ had their
own life and intelligence, as if those were more clever than we, even more
clever than their inventor, as if they would yield more than was put into
them at the time.’53
Hertz considers as the principal aim of conscious natural science (be-
wusster Naturerkenntnis) to foresee future experiences. In order to reach
that he proposes a ‘sign theory’ which is more explicit than Helmholtz’s and
he gives a set of rules, both formative and descriptive.
We form for us inner phantoms (Scheinbilder) or symbols of the

things and in such a way that the logical (denknotwendig) con-
sequences of the symbols are always pictures of the physically
necessary (naturnotwendig54 ) consequences of the depicted ob-
52
‘Unwürdig eines wissenschaftlich sein wollenden Denkers aber ist es, wenn er den
hypothetischen [. . . ] Ursprung vergisst.’ Helmholtz (1892), p. 243.
53
‘Man kann diese wunderbare Theorie nicht studieren, ohne bisweilen die Empfindung
zu haben, als wohne den mathematischen Formeln selbständiges Leben und eigener Ver-
stand inne als seien dieselben klüger als wir, klüger sogar als ihr Erfinder, als gäben sie
uns mehr heraus, als seinerzeit in sie hineingelegt wurde.’ Hertz (1889), p. 11
54
Hertz himself uses the terms ‘denknotwendig’ and ‘naturnotwendig’ instead of ‘logisch’
and ‘gesetzlich’. However, given the context it is obvious that he does not use ‘necessity’
in the way we do it today when talking about laws of nature. Something similar holds for
his use of ‘essential’ which does not mean ‘metaphysically prior’ in Hertz but something
like ‘experimentally settled’.
27
jects.55
He does not take it for granted that such a procedure is possible, but notes
that experience tells us, that it is. Hertz gives several selection criteria for
the symbols. Although those criteria are more important for the gathering
of models than for the development of a theory (which is what we are mainly
interested in), we think that they are worth mentioning. Hertz writes:
The images we speak of are our imaginations of the things; they

have with the things the one essential correpondence which lies
in the fulfillment of the above mentioned postulates.56
He also points out that by this congruence requirement the construction

of the pictures is by no means unique, but that different pictures can be
distinguished according to their ‘admissibility’ (Zulässigkeit), ‘correctness’
(Richtigkeit) and ‘appropriateness’ (Zweckmäßigkeit). Pictures are admis-
sible if they do not contradict logic (laws of thought), and admissible pic-
tures are correct if ‘their essential relations do not contradict the relations
between external objects’. However, admissible and correct pictures can
(and normally will) differ in their appropriateness which comes in two guises;
namely as ‘distinctness’ (Deutlichkeit) and ‘simplicity’ (Einfachheit). A pic-
ture which mirrors more of an object’s essential relations is called a clearer
one, and amongst equally clear pictures the one with the less vacuous rela-
tions (i.e. relations which occur in the picture but not in the external world)
is the simpler one.57 Such competing pictures normally occur if one is sim-
plifying certain aspects of a more complete theory in models. These criteria
as destinctness and simplicity can be important also in more fundamental
theoretical questions, for instance the relation between conventional quan-
tum field theory and the approach by the algebra of local observables which
concentrates exclusively on the local observables, see section 2.5.5.
55
‘Wir machen uns innere Scheinbilder oder Symbole der äusseren Gegenstände, und
zwar machen wir sie von folgender Art, dass die denknotwendigen Folgen der Bilder
stets wieder Bilder seien von den naturnotwendigen Folgen der abgebildeten Gegenstände.’
Hertz (1894), pp. 1f.
56
‘Die Bilder, von welchen wir reden, sind unsere Vorstellungen von den Dingen; sie
haben mit den Dingen die eine wesentliche Übereinstimmung, welche in der Erfüllung der
genannten Forderung liegt.’ Hertz (1894), p. 2.
57
Note that Hertz’s definition of ‘simplicity’ is not like the common blurry one, that one
wouldn’t like as a functional relation a polynomial of 27th order or something alike.
28
There is a crucial difference between the sign theory of Hertz and that of
Helmholtz. Helmholtz’s signs are related to the sensual impressions whereas
those of Hertz can be free creations of the mind. Here a semiotic distinc-
tion drawn by Charles Sanders Peirce is helpful to contrast between these
two types of ‘signs’; namely the distinction between indexical signs and con-
ventional signs (symbols).58 The former show some existential relation to
their objects of reference; i.e. their meaning is based on a cause and effect
relationship. Thus, for instance, a weather cock is an indexical sign of the
wind since it is turned by the wind into its direction. A symbol on the other
hand lacks such a cause and effect relationship. Its meaning is based on a
conventional regularity as, for instance, is our use of the word ‘strawberries’
to refer to strawberries. For obviously there is neither any causal relation
nor any resemblance between the string of letters in ‘strawberries’ and those
red and eatable objects. Also the relation between a quantised field, the es-
sential ingredient of the theory and a particle, on which the interpretation of
experiments is based on, can be very indirect, as discussed in the preceding
section.
Thus, one can take Helmholtz’s approach, which started from sense per-
ception, to be a theory of indexical signs; whereas Hertz’s theory should
already count as one of symbols.
The historical development went further away from a theory of indexical
signs and more and more towards a theory of symbols (conventional signs).
Both Helmholtz and Hertz were certainly motivated by the un-reconciled
differences of classical mechanics and of Maxwell’s theory but they had no
reason to doubt the unconditioned physical validity of classical mechanics and
Maxwell’s theory. This was no longer the case for the next main contributor
in this tradition, the great mathematician Henri Poincaré who lived long
enough to see the beginning erosion of classical physics at the turn to the
20th century. He expounded his views on the epistemological foundation
of science in three works: Science et l’hypothèse (1903), La valeur de la
science (1905), and Science et méthode (1908). Very much like Helmholtz
and Hertz he takes the success of science as a starting point and concludes
— in accordance with today’s miracle argument — that this success would
not be possible if science would not give us knowledge of something of reality
58
Peirce also introduces ‘icons’ (iconic signs) which have some relation of resemblance
to their denoted object. However, since straightforward resemblances are already ruled
out in Helmholtz’s approach, we do not need to take icons into account here.
29
(quelque chose de la réalité). However, he continues in line with what is today
called the pessimistic meta-induction, that what science can attain are not
the things themselves, as the naive entity realists think, but only the relations
between the things (les rapports entre les choses). Outside these relations
there is no recognizable (connaisable) reality.59 Thus, this strain in Poincaré’s
work is the starting point of nowadays structural realism as introduced above.
However, this is mainly based on a single passage in which Poincaré discusses
the transition from Fresnel’s to Maxwell’s description of optical phenomena;
it is closely related to Hertz’s introduction of signs mentioned above:
[. . . ] the aim of Fresnel was not to find out whether there

really is an ether, whether it is or it is not formed of atoms,
whether these atoms really move in this or that sense; his object
was to foresee optical phenomena.
Now, Fresnel’s theory always permits of this, to-day as well as
before Maxwell. The differential equations are always true [. . . ]
They teach us, now as then, that there is such and such a relation
between some thing and some other thing; only this something
formerly we called motion; we now call it electric current. But
these appellations were only images substituted for the real ob-
jects which nature will eternally hide from us. The true relations
between these real objects are the only reality we can attain to,
and the only condition is that the same relations exist between
these objects as between the images by which we are forced to
replace them. If these relations are known to us, what matter if
we deem it convenient to replace one image by another.60
59
‘Cela ne pourrait être si elle ne nous faisait connaı̂tre quelque chose de réalité; mais
ce qu’elle peut atteindre, ce ne sont pas les choses elles-mêmes, comme les pensent les
dogmatistes naı̈fs, ce sont seulement les rapports entre les choses; en dehors de ces rapports,
il n’y a pas de réalité connaisable.’ Poincaré (1927), Introduction, p. 4.
60
‘Nulle théorie semblait plus solide que celle de Fresnel qui attribuait la lumière aux
mouvements de l‘éther. Cependant, on lui préfère maintenant celle de Maxwell. Cela
veut-il dire que l’oeuvre de Fresnel a été vaine? Non, car le but de Fresnel n‘était pas
de savoir s’il y a réellement un éther, s’il est ou non formeé d’atomes, si ces atomes se
meuvent réellement dans tel ou tel sense; c‘était de prévoir les phénomènes optiques. Or,
cela, la théorie de Fresnel le permet toujours, aujourd’hui aussi bien qu’avant Maxwell. Les
équations differentielles sont toujours vraies; [. . . ] Elles nous apprennent, après comme
avant, qu’il y a tel rapport entre quelque chose et quelque autre chose; seulement, ce
quelque chose nous l’appelions autrefois mouvement, nous l’appelons maintenant courant
30
Obviously one can read this quote in two ways, namely as making purely
epistemic claims or as making metaphysical claims (this will mainly depend
on the translation and reading of ‘The true relations between these real ob-
jects are the only reality we can attain to [. . . ]’). Thus, taking only this
passage into account, Poincaré’s ‘structural realism’ could come in two ver-
sions: according to modern terminology one is called epistemic (or restrictive)
structural realism and the other ontic (or eliminative) structural realism. 61
The two views differ about the question as to whether there is something
else in the world than structure. While ontic structural realism says no, epis-
temic structural realism allows for entities that stand in the relations we find
out about. However, according to the latter we will never know about those
entities; i.e. nature will eternally hide from us the things in themselves (cf.
what we said about the detection of intrinsic properties earlier on). Thus
one might call epistemic structural realism a ‘Kantian physicalism’.62
The main critique that has been brought forward against ontic structural
realism — and with which the authors concur — is that one cannot have
relations without relata. Thus, we are more sympathetic to its epistemic
version which is also much more in line with Poincaré. For Poincaré stresses
the importance of conventions in science — indeed he is notoriously credited
as the very founding father of conventionalism (and surely not of realism!).
He illustrates the rôle of conventions in science by the following thought
experiment. Assume a fictitious earth covered with thick clouds such that
the stars were never visible. Though scientists living on that fictitious earth
would finally find out that many facts can be explained by assuming that
this earth turns itself around an axis, this statement is a pure convention.
The statements ‘the earth turns itself’ and ‘it is more convenient to assume
that the earth turns itself’ have accordingly one and the same meaning.63 He
électrique. Mais ces appelations n‘étaient que des images substituées aux objets réels que
la nature nous cachera éternellement. Les rapports véritables entre ces objets réels sont
la seule réalité que nous puissions atteindre, et la seule condition, c’est qu’il y ait les
mêmes rapports entre ces objets qu’entre les images que nous sommes forcés de mettre
à leur place. Si ces rapports nous sont connus, qu’importe si nous jugeons commode de
remplacer une image par une autre.’ Poincaré (1927), X, p. 190.
61
A distinction drawn by Ladyman (1998) and by Psillos (2001). The most prominent
advocates of epistemic structural realism are Worrall and Zahar, while ontic structural
realism is argued for notably in the works of French and Ladyman. See, e.g., Worral
(1996), Zahar (1996), and Ladyman (1998).
62
See Jackson (1998), pp. 23–24.
63
Poincaré (1927), VIII p. 139, Poincaré (1905), XI, sect. 7, pp. 271ff.
31
also stresses that the choice of geometry is purely conventional. He explicitly
says that the axioms of Euclid are neither synthetic judgments a priori nor
experimental results; they are conventions.64 Hence, this conventional and
not the realistic aspect was the dominant one in his philosophy.65 This can
also be seen from another passage in Science et l’hypothèse where Poincaré
again refers to Fresnel’s theory (this time to his laws of refraction) but now
stresses their approximative character.66
The delicacy of the balance between conventionalism and structural re-
alism becomes clear from the following résumé he draws near the end of La
valeur de la science:
In summary, the only objective reality are the relations between

the things (choses) from which the universal harmony starts.
Doubtless these relations, this harmony, would not be conceived
outside a mind who conceives or feels them. But they are never-
theless objective [intersubjective] since they are, become, or stay
common to all thinking beings.67
Poincaré lived until 1912 and saw the big changes occurring at the turn
of the century. He seems not to have been impressed strongly by the birth of
quantum physics, but the effects of electrodynamics on mechanics shook him
considerably, as can be read from the last chapter of Science et l’hypothèse,
entitled ‘La fin de la matière’. He also foresaw the consequences on the best
established physical theory, Newton’s theory on gravity: ‘If there is no more
mass, what becomes of the law of Newton?’, he asks.
Therefore Poincaré is aware that ‘theories seem fragile’ but he asserts that
they do not die fully and from each of them there rests something. This some-
thing has to be looked for and sorted out, but it is this, and only this, which
is the ‘true reality’. This process of sorting out was particularly necessary in
the further development towards the theory of (special) relativity, quantum
64
Poincaré (1927), III p. 66.
65
The realistic aspects are mainly brought in to prevent conventionalism to drift into
what he calls ‘nominalism’ — cf. Poincaré (1905), X pp. 233ff in particular.
66
Poincaré (1927), X, p. 211.
67
‘En résumé, la seule réalité objective, ce sont les rapports des choses d’où résulte
l’harmomie universelle. Sans doutes ces rapports, cette harmonie ne sauraient être conçus
en dehors d’un esprit qui les conçoit ou qui les sent. Mais ils sont néanmois objectifs parce
qu’ils sont, deviendront, ou resteront communs à tous les êtres pensants.’ Poincaré (1905),
XI, p. 271.
32
mechanics and quantum field theory. In this transition also mathematical re-
lations which were considered to be absolutely valid physical relations turned
out to be obeyed only approximately and could be violated vehemently under
certain conditions. This was of course a shock for those believing in an on-
tological significance of the concepts of science, but even the conventionalist
(and ‘structural realist’) Poincaré writes somewhat bewildered on relativistic
mechanics:
The essential attribute of matter is its mass, its inertia. The mass
is that which always at any place stays constant [. . . ] If therefore
one has proven that the mass, the inertia does not truly belong
to it in reality [. . . ] , that this mass, the constance par excellence
is itself susceptible to changes, then one could say that matter
does not exist. Now that is precisely what one announces.68
The fourth eminent figure whose interpretation of science we shall discuss
briefly is the mathematician Hermann Weyl. He did not only live during the
period of great changes in the first decades of the 20th century but also
contributed essentially to them and was fully aware of the epistemological
impact. In the preface to his articles ‘What is matter’ (1923) he notes: ‘It is
perhaps especially now to early, to speak about the essence of matter, where
quantum mechanics [. . . ] keeps her secrets still in a firmly closed shell.’69
As is also evident from the formulation ‘essence of matter’ he uses, at least
hypothetically, metaphysical concepts. As a consequence, in his discussion of
Hertz’s ‘Mechanik’ he emphasises the actual treatment of matter in the main
part of the work, rather than the symbolic programme in the introduction.
He sees Hertz as a follower of Huygens in the line of the ‘substance theory’
of matter but he remarks that with Hertz ‘the concept of substance is by
means of mathematical generalisation formalised to an abstract scheme.’70
He admits that the final systematic form of science might be of a similar kind
68
‘L’attribut essentiel de la matière, c’est sa masse, son inertie. La masse est ce qui
partout et toujours demeure constant [. . . ]. Si donc on venait à démontrer que la masse,
l’inertie de la matière ne lui appartient pas en réalité [. . . ] que cette masse, la constante
par excellence, est elle-même susceptible d’altération, on pourrait bien dire que la matière
n’existe pas. Or, c’est là précisément ce qu’on annonce.’ Poincaré (1927), XIV, p. 282.
69
‘Es ist vielleicht gerade heute besonders verfrüht, über das Wesen der Materie zu reden,
wo die Quantentheorie [. . . ] ihr Geheimnis noch immer in fest verschlossener Schale hält.’
Weyl (1924).
70
Weyl (1924), p. 17. This transition from a metaphysical to a mathematical formal
point of view seems to have a long tradition. Leibniz, the opponent of Huygens in matters
33
but he doubts ‘that through the cancellation of metaphysical intuition [. . . ]
to which the concept of substance belongs, theoretical interpretation looses
everything stringent.’
Weyl had generalised Riemannian geometry introducing the gauge princi-
ple by which the length of a measuring device could depend on its history in
space-time. In that way he could obtain a stringent common interpretation of
the two great branches of classical field theory: Einstein’s theory of gravita-
tion and Maxwell’s electrodynamics. But the application of Weyl’s principle
to geometry contradicted experiment, as especially Einstein realised imme-
diately. The general principle of ‘gauge invariance’, however, turned out to
be one of the most fruitful ones in quantum field theory. London, who was
the first to recognise the fruitfulness of Weyl’s postulates in the then newly
developed modern form of quantum mechanics (1927), commented on the
long adherence of Weyl to his theory:71 ‘An uncommon clear metaphysical
conviction was necessary’ not to accept the experimental evidence.72 Weyl,
however, immediately took up the idea that quantum mechanics was the
right application of his gauge theory, in particular after Dirac had given a
relativistic description of it.73 But now it was not the length of a measuring
device which was gauged, but the very abstract phase of a wave function,
in quantum field theory the wave of the charged field. As we have already
mentioned in section 2 all three sectors of the Standard Model of particle
physics are gauge theories in the sense of Weyl.
In contrast to his metaphysical inclinations of his earlier writings, Weyl
refers in the philosophical publications of his later years to the symbolic form
as the adequate approach to mathematics and physics. This becomes clear
already from the titles of his articles; ‘Science as a Symbolic Construction of
Man’74 and ‘On the Symbolism of Mathematics and Mathematical Physics’.75
In the former one he shows in case studies, very much like the philosopher
Ernst Cassirer, the development from substantial to symbolic forms. Here
of atomism, writes 1694 in a letter to l’Hospital that ‘his metaphysics is all mathematics,
so-to-speak, or could become it’, Leibniz (1849), vol II, p. 258. We shall see a similar
development in the arguments of Weyl.
71
London (1927).
72
The discussion between Einstein and Weyl has been documented by Straumann Strau-
mann (1987), see also O’Raifeartaigh and Straumann (1998).
73
Weyl (1929).
74
Wissenschaft als symbolische Konstruktion des Menschen. Weyl (1949).
75
Über den Symbolismus der Mathematik und mathematischen Physik. Weyl (1953).
34
he also quotes approvingly the symbolic programme of Hertz and he comes
to such definite statements as:
it is the free in symbols acting spirit which constructs himself in
physics a frame to which he refers the manifold of phenomena.
He does not need for that imported means like space and time
and particles of substance; he takes everything from himself.76
He refers to the attempts of Alfred North Whitehead and Bertrand Russell
to base reality on events and considers them as ‘Loves Labour Lost’ since
we are left with nothing than symbolic construction and that the recent
developments in modern physics by general relativity and quantum theory
confirmed that.77
One aim of our paper is to corroborate this point of view analysing es-
pecially the developments of quantum field theory in the 80ies of the 20th
century. However, since we have to talk about several symbolic representa-
tions, it will be helpful to briefly consider one further idea brought forward
by Weyl to evaluate the ‘objective core’ of different representations.78 The
basic idea is that there are always (trivial) differences between different rep-
resentations, but that there are also certain features which are kept stable if
a transformation from one representation to another occurs.79 If the origin of
some geometric object represented in a Cartesian coordinate system is shifted
or the axes are rotated the absolute coordinates of the object will change,
too. However, the shape of the figure (the length of its sides etc.) will stay
the same. Thus, the length of the sides rather than a certain abscissa value
is an objective property of the figure. In Weyl’s more technical terms:
What we learn from our whole discussion and what has indeed
become a guiding principle in modern mathematics is this les-
son: Whenever you have to do with a structure-endowed entity Σ
76
‘[. . . ] dass es der freie, in Symbolen schaffende Geist ist, der sich in der Physik ein
objektives Gerüst baut, auf dass er die Mannigfaltigkeit der Phänomene ordnend bezieht.
Er bedarf dazu keiner solchen von aussen gelieferten Mittel wie Raum, Zeit und Substanz-
partikel; er nimmt alles aus sich selbst’. Weyl (1949).
77
At least with respect to Whitehead this is no adequate critique. For his philosophy is
exactly one of (the transformation of) symbols.
78
Indeed this approach can be traced back to Helmholtz who maintained that the special
nature of a melody is it’s invariance under transpositions, which is also an essential feature
of space Helmholtz (1877) XIX, p. 596 f. It was also taken over, e.g., by Cassirer in his
paper ‘The Concept of Group and the Theory of Perception’.
79
Cf. Weyl (1952) and Weyl (2000), Ch. 13.
35
try to determine its group of automorphisms, the group of those
element-wise transformations which leave all structural relations
undisturbed. You can expect to gain a deep insight into the con-
stitution of Σ in this way.80
Weyl also emphasises that transformation invariance and symmetry be-

long together and that they are important not only for mathematics and the
exact sciences but also for arts etc.81 Indeed we regard it to be important for
any symbolic system and in particular that any comparison between sym-
bolic representations can make use of this type of transformation-invariance-
approach.82
5 A Theory of Symbols for

Quantum Field Theory
5.1 Symbols and Relations
The aim of this section is to widen the symbolic interpretation of science ini-
tiated by Helmholtz, Hertz and Poincaré. We introduce some concepts from
modern semiotics and apply them to quantum field theory. This will lead
to a discussion on two levels of symbolic representation; namely the level of
single symbols and the level of relations between symbols. For as we have
seen, Helmholtz and Hertz claim that objectivity in science is reflected in
relations between symbols. This view — the discussion of which will lead
us back to structural realism — is also held by the mathematician Poincaré
who emphasises the conventional character more strongly than the physicists
Helmholtz and Hertz. However, the fact that we make symbols rather than
80
Weyl (1952), p. 144.
81
Cf. Weyl (1952), p. 145, e.g.
82
Recently Weyl’s transformation-invariance-approach was put into spotlight in the con-
text of quantum field theory by Auyang (1995). His version roughly goes like this: While
intrinsic properties or ‘things in themselves’ — which we may denote by x — are necessar-
ily hidden, we can encounter their functional relations. Thus, even without any knowledge
of x, we can know about f (x) = xf , g(x) = xg etc. If on top of that we also know the rules
of transformation to get from xf to xg (i.e. we know the function g ◦ f −1 ), we will know
the invariant properties under these transformations and hence the most likely candidates
for objective properties (i.e. for the ‘real’ or intrinsic properties of x).
36
somehow have them, has already been stressed by Hertz. Indeed this pro-
cess of representing outside phenomena by symbols is crucial for science, for
it allows to establish mathematical or logical relations; i.e. plays an ever
increasing rôle especially in modern physics.
Before coming back to quantum field theory, it seems important to say
something more about the level of symbolic (mathematical) relations and to
introduce the main ideas by a familiar example, namely time. The change of
the mathematical relations which hold between symbols is much less radical
than the change of the conceptual basis underlying these symbols, especially
if it is based on notions of everyday life. In accordance with semiotics, the
latter — these rather individual everyday associations — might be called
‘connotations’. Whereas the strictly rule-governed inner-mathematical mean-
ings — i.e. the ‘mathematical valence’, the relationship which the symbol
contracts with other symbols of the theory’s ‘mathematical vocabulary’ —
would be the ‘denotation’.83 The absolute time in Newtonian physics, for in-
stance, is conceptually very different from that of relativistic mechanics. Our
commonsense notions of simultaneity, of the flow of time etc. are in excel-
lent agreement with Newtonian physics but get confused as soon as we start
talking about relativistic mechanics. However, things look completely differ-
ent on the denotative level of the symbols, i.e. on the formal mathematical
level where time occurs as a real variable t. Here the kinematical relations
of Newtonian mechanics are simply obtained from the relativistic ones by an
expansion with respect to the ratio v/c, where v is the velocity connected
with a mass m and c the vacuum velocity of light. The relativistic relation (1)
expressed through the velocity vector v rather than the momentum vector p
of a particle is:
mc2
E=q . (4)
v2
1− c2
Performing the above mentioned expansion yields:
v2
!!
2 m 2
E = mc + v 1 + O 2 . (5)
2 c
The first term is the rest energy, the second term thekinetic energy of a
2
particle in non-relativistic mechanics and the term O vc2 indicates that the
83
Compare the definition of denotation as the ‘valence in the semantic field’ and of
connotation as ‘emotional or cognitive meaning’ in Eco (1968).
37
2

relativistic correction is of the order vc2 and thus small if the modulus of the
velocity |v| is small compared to the velocity of light c. The corresponding
relativistic relation between the time coordinate t0 at a point with position
vector x0 in one (inertial) system K0 and the time and space coordinates t, x
in a different (inertial) system K moving relative to K 0 with velocity (vector)
v, is given by
1 v·x

t0 = q t − . (6)
2
1 − v2 c2
c
We see that in the limit where |v|/c tends to zero one obtains t0 = t, the
time becomes ‘absolute’ in the sense that it does not depend on the inertial
system, and hence on the kinematical state of the observer.
We consider it not only remarkable but characteristic, too, that the pro-
tagonists of a symbolic interpretation of science, Helmholtz, Hertz, Poincaré
and Weyl, are also eminent physicists and mathematicians. For it is very
plausible that such a view is first advocated by (mathematical) physicist,
who were in a position to experience the permanence in mathematical de-
scriptions; i.e. in the symbolic relations. As opposed to them people judging
from outside tend to interpret transitions between theories much more dra-
matically. Rather than the continuity in mathematical description, i.e. the
symbols’ denotation, they stress the discontinuities in their connotations. So
Thomas S. Kuhn states in the case study of the derivation of non-relativistic
physics from the relativistic one that ‘those laws have to be reinterpreted
in a way that would be impossible until after Einstein’s work’,84 whereas in
the denotative framework of the theory of symbols the limiting laws, that is
the relation between the symbols, are indeed the ones of classical mechanics.
Thus, the change really concerns only the connotational background of the
symbols and not their denotations; i.e. not the ‘mathematical valence’ of t.
Since objectivity in science is reflected not in the symbols themselves but
in their relations, Kuhn’s arguments on scientific revolutions — or what we
would call a ‘connotation shift’ — might thus be historically correct but do
seem epistemologically irrelevant from a physicist’s perspective.85
As just seen, in some cases it is possible to perform the transition between
two theories in a mathematically rigorous way. In that case we can call the
limit an effective theory (cf. section 2.6). In other cases there are many
84
Kuhn (1970), p. 101.
85
Similar arguments can be brought forward against Feyerabend’s criticism of consis-
tency conditions. Feyerabend (1975), sect. 3.
38
structural similarities but nevertheless there exists (at least up to now) no
derivation in a strict mathematical sense. Coming back to the physical main
topic of this paper, the transition from quantum to classical field theory is
such a case. We speak of the classical level of a quantum field theory if we
make an expansion of amplitudes in powers of h̄, the Planck action quan-
tum, and consider only the leading order. However, the intentions of and
questions asked by the two theories are so different that we think it more
appropriate to speak of different symbolic representations. The symbols of
the two theories are completely different but there are close relations between
the structures, e.g. internal or external symmetries. Normally the ‘higher’
(more fundamental) theory does not make obsolete the ‘lower’ one. This is
obvious for dealing with ‘macroscopic’ problems. The unsolved difficulties of
quantum field theory e.g. do not prevent to solve most complicated problems
concerning the propagation of radio waves. It might even be necessary to use
the lower theory in the higher one in an essential way, as e.g. in the analytic
description of bound states in quantum field theory. In order to calculate
the Lamb-shift, one of the triumphs of relativistic quantum field theory, one
starts with a quantum field theory in a given ‘external’ classical field, though
this field should in principle also be described by quantum field theory. Thus,
to calculate the Lamb-shift one has to make use of different symbolic rep-
resentations at the same time; namely the ‘higher’ quantum and the ‘lower’
classical field theory. To elaborate on how and why classical and quantum
physics should be viewed as different symbolic representations and why even
within quantum field theory there are several symbolic representations, it is
helpful to briefly introduce some more semiotic terminology.86
5.2 A Semiotic View of Quantum Field Theory

In semiotics a system of symbols with given rules that allow transfer of in-
formation is called a ‘code’. In this sense not only language is a code, but
also, for instance, human perception involves codes (think of the distinction
between foreground and background in Gestalt psychology, e.g.); within the
arts classicism and romanticism can count as codes and in particular any
physical theory put in mathematical terms is a code. For all computational
manipulations have to follow certain rules and obviously transport informa-
tion about external phenomena.
86
The following terminology is taken from Eco (1968).
39
A code has a ‘vocabulary’ of basic units which — using the code’s (syn-
tactic) rules — can be combined to generate larger meaningful combinations.
Some codes can be analysed with regard to two structural levels called ‘first’
and ‘second articulation’. The first articulation deals with the smallest mean-
ingful units of the code, the so called ‘signs’. In case of spoken language this
would be morphemes or words. At the level of the second articulation a code
divides into minimal functional units which have purely differential charac-
ter; i.e. do not have meaning in themselves. They are sometimes called
‘figures’. Taking the example of spoken language again, its second articu-
lation consists of phonemes. Thus, the crucial difference between the two
levels is that the basic units of the first articulation denote while those of the
second articulation lack denotative power.
This semiotic terminology can now be applied to what we think are differ-
ent symbolic representations in modern physics. Classical mechanics, classi-
cal field theory, non-relativistic quantum mechanics and quantum field theory
should all be taken to be different codes (indeed within quantum field the-
ory further distinctions have to be made as we will see shortly). All these
codes more or less share their secondR articulation, namely the mathematical
language expressed in ‘figures’ like ‘ ’, ‘+’, ‘∂’ and further structures. How-
ever, they disagree in their first articulation, i.e. in their denoting terms and
thus also in the entities they describe. Take, as an illustration, electrons,
heavy and light quarks. The symbolic systems in which electrons occur in-
clude classical mechanics, non-relativistic quantum mechanics and quantum
field theory. Heavy quarks only make sense in the context of non-relativistic
quantum mechanics and quantum field theory, whereas light quarks only in
that of quantum field theory. This is just another (more elegant) phrasing
for what has already been stated in our case study on ‘Particle and Fields’
in section 3, namely that questions like ‘What is a particle?’ or ‘Does such-
and-such particle exist?’ depend on the theoretical background. Given our
semiotic approach this simply amounts to say that the denoting terms of dif-
ferent codes might (and indeed sometimes do) vary. Moreover, we think that
whether some ‘particles’ appear closer to our intuition than others depends
on the number of codes in which those ‘particles’ do occur.
We now argue that the various facets of quantum field theory elucidated
in section 2 do form different codes. The reader is reminded that up to
now no quantum field theory describing interaction between particles in
four space-time dimensions has been constructed on a sound mathemati-
cal footing. Wightman’s postulates (axioms) are considered to form the core
40
structure of a theory based on local field operators; but they circumvent
to formulate the dynamical evolution in terms of field operators. It is this
dynamical evolution, however, that entails the specific physical outcome of
the theory. Nevertheless, the postulates of Wightman allow to derive some
stringent structural consequences as e.g. the CPT-theorem, the connection
of spin with statistics, and dispersion relations, cf. 2.5.1. These predictions
are in fact experimentally very well satisfied and any violation of them would
seriously call in question the concept of a local quantum field. Within the
Wightman postulates different concrete quantum field theories appear con-
ceivable. Actually, in fictitious worlds having one or two spatial dimensions
only, various quantum field theories with internal dynamics have been rigor-
ously constructed and shown to realise the Wightman postulates, cf. section
2.5.3.
A concrete quantum field theory has to be generated by ‘quantising’ a
heuristic classical precursor theory, the Lagrangian density of which is a
function of the fields involved and determines the interaction. The Standard
Model, in particular, is a non-Abelian gauge theory. As already mentioned
above there exists (at present) no fully fledged quantum field theory of this
model. Instead, there are two complementary codes to perform a ‘quan-
tisation’ of the classical precursor: perturbative renormalisation theory, cf.
section 2.5.2, and lattice gauge theory, cf. section 2.5.4. The first one, ab
initio only formulated as a formal power series based on free quantum fields,
cannot create hadrons. Moreover, it employs in addition ghost fields to im-
plement (covariantly) the local gauge symmetry. The second one avoids these
unphysical degrees of freedom and starts with a well-defined nonperturbative
set-up, which involves only the fundamental degrees of freedom, however re-
stricted to a Euclidean space-time lattice. The respective predictive power
of these two codes points to complementary physical domains. The pertur-
bative theory accounts for electroweak processes and to the short-distance
behaviour of quantum chromodynamics. In contrast, the lattice gauge theory
aims at the long-distance properties of quantum chromodynamics, i.e. the
spectrum of hadrons dynamically generated from the fundamental degrees of
freedom of the theory.
The respective denotations used in the general theory of quantised fields,
in perturbative renormalisation theory, and in lattice gauge theory differ
markedly from each other, although they have common elements, too. These
differences do not allow to arrange the codes in a hierarchy. The physically
distinguished non-Abelian gauge symmetry is not addressed in the Wight-
41
man postulates, e.g. In spite of the fact that they could not be realised
constructively87 as yet, they remain a theoretically appealing vision. Both,
the perturbative gauge theory and the lattice gauge theory encompass ele-
ments of these postulates.
We may look at this state of affairs from a structural realist’s perspective.
Basing structural realism on the general theory of quantised fields amounts to
restrict the theoretical claim to the very core of local quantum field theory.
The price to be paid for circumnavigating thereby the formulation of an
explicit dynamics is that only few, albeit fundamental and experimentally
very well satisfied physical relations result. Besides, the persistent threat to
any realistic interpretation by the pessimistic meta-induction may materialise
here e.g. if at very small distances (Planck length) the notion of Minkowski
space-time becomes obsolete, and hence the Wightman postulates would be
subject to revision. A different objection aims at the hidden dynamics in
the general theory of quantised fields: as a mere framework it might be
compatible with various types of concrete quantum field theories. Or, to
express it more emphatically: there might be very different worlds, which
all exhibit the structural features resulting from the Wightman postulates.
In order to arrive at a wealth of physical relations aiming to cope with the
immense body of detailed measurements performed in the subnuclear domain,
an adequate concrete quantum field theory has to be created, i.e. in terms
of specific fields and their interaction, thus fixing the dynamical content of
the theory. The Standard Model is supposed to play this rôle. As already
stated, the ‘quantisation’ of this model could not yet be achieved fully in the
form of a proper mathematical construction. However, partial approaches
to this goal are attempted, which appear as facets of quantum field theory
with respective codes. To a large extent, the choice between these codes is
dictated by the question one has in mind.
This, however, puts a further and much more pressing problem onto struc-
tural realism; namely what we might christen the ‘problem of synchronic
existence’. Remember that structural realism has been introduced to cope
with the pessimistic meta-induction. The latter, however, is a problem of
‘diachronic existence’ of different theories. Structural realism is adapted to
explain why the purported entities of different theories which occur one after
the other might change; like in the transition from Fresnel to Maxwell, to use
the notorious example. However, it is not adapted to explain why there are
87
In four-dimensional space-time.
42
different theories that exist at one and the same time — as it is the case with
the different facets of quantum field theory. For structural realism would
have to maintain that one of the facets is the ‘real’ one. By the same token,
structural realism would not be able to cope with the miracle argument any
longer. For if it took perturbation theory to be the right structure then the
success of lattice gauge theory would become a miracle and vice versa. These
approaches, however, can coexist, since they aim at different domains of the
full theory.
Indeed it seems that the synchronic existence of different facets is of-
ten neglected within the philosophy of quantum field theory and has led to
some rather misleading ideas. One is the debate about ‘quasi-autonomous
domains’, brought forward in the context of effective field theories, see sec-
tion 2.6.88 The basic idea here is that due to the decoupling theorem —
i.e. due to the fact that higher energy scales do have negligible relevance for
an investigation at a lower energy scale — the objects of the physical world
appear to be layered into separate domains which are more or less indepen-
dent from each other. At first glance this idea looks like it would provide us
with a convenient (stratified) inventory of the world; at one level there are
perhaps quarks and electrons, the next level might be inhibited by protons
and neutrons etc. However, this is not the case. For the context in which
these domains are argued for is purely perturbative. This means that, for in-
stance, strictly speaking not even hadrons do occur in any of those domains.
Even talking about deep inelastic scattering is of no help since the involved
structure functions are nonperturbative entities. Thus, the different facets
of quantum field theory have to be taken into account if one was willing to
extract a ‘full ontology of our world’. However, the idea of quasi-autonomous
domains would get really messed up then. For should we assume a separate
quasi-autonomous domain for the objects of lattice gauge theory then etc.?
The reason why — at first sight — this notion about quasi-autonomous
domains seems to have commonsense on its side can easily be explained by our
semiotic approach. For if we have accurate theories on different levels (energy
scales), and hence different codes with different denoted entities for each of
those levels, it seems quite straightforward to take those entities ‘for real’ and
assume the same hierarchy in their existence as in the theories. However, this
step from semiotics to reality (from denotation to existence) is not valid; it
88
These domains have been introduced by Cao and Schweber (1993) and are criticised
amongst others by Robinson (1992) and Hartmann (2001).
43
is what Whitehead would call a ‘fallacy of misplaced concreteness’. First,
just because something occurs in a symbolic representation we should not
naively take it for ‘real’. Second, also a given symbolic system as a whole
should not be conflated with reality. The notion about quasi-autonomous
domains as derived from perturbation theory violates both. Thus, it seems
more promising to stay with a semiotic approach and not to take symbols
to refer one-to-one to external objects. This has practical consequences.
Consider a sequence of effective theories, pertaining each to a different scale.
The cutoff, which is necessary in a non-renormalisable effective theory, would
then gain a realistic meaning. This, however, would allow to calculate for
instance the energy density of the vacuum, which would turn out to be many,
many orders of magnitude larger than the measured one.
In contrast to the structural realism of most of the contemporary protag-
onists, our semiotic approach — which we developed along the thoughts of
Helmholtz, Hertz, Poincaré, and Weyl — does not commit a ‘fallacy of mis-
placed concreteness’. However, coming back to Worrall’s initial question89 ,
alluded to at the beginning of section 4, it seems that our approach cannot
provide us with ‘the best of both worlds’. It abandons a straightforward re-
alist interpretation and thus fails to counter the miracle argument. However,
as we have seen, even structural realism fails to account for the miracle argu-
ment given that there are different successful facets at one and the same time.
A semiotic approach on the other hand is not bothered by such a synchronic
existence. Additionally, the pessimistic meta-induction poses no problem on
it as well. Symbolic systems have always changed and presumably will always
be changing.
Admittedly, we did not say much about the process of sign creation and
only made some negative claims about their relation to external objects.
However, even if we left open some important questions, we have shown the
elaboration of a semiotic approach to be necessary for a full account of con-
temporary physics. The semiotic approach allows to account for the different
synchronic facets of quantum field theory; as indeed any serious interpreta-
tion of contemporary physics should. Moreover, whereas both structural
realism and a semiotic approach run fool of the miracle argument with re-
spect to quantum field theory, the latter at least copes with the pessimistic
meta-induction and commits no ‘fallacy of misplaced concreteness’.
89
Worral (1996).
44
6 Summary
Examining relativistic quantum field theory we have claimed that its sym-
bolic descriptions of subnuclear phenomena can be understood most ade-
quately from a semiotic point of view. We have emphasised that such an
interpretation of a physical theory is not restricted to particle physics, but
already came to the fore in the middle of the nineteenth century with the
emancipation of the theory of the classical electromagnetic field from me-
chanics. Notably the semiotic approach goes back to the work of Helmholtz,
Hertz and Poincaré, who played a major role in the development of classi-
cal field theory. Looking at a theory of quantised fields, the epistemological
questions posed with regard to classical field theory are heightened, even if
one strictly keeps from the beginning to a probabilistic interpretation of a
quantum theory.
Relativistic quantum field theory is widely recognised as forming the ad-
equate theoretical frame to the phenomena of particle physics observed up
to the present. Most notably, the ‘Standard Model’ achieved yields an excel-
lent description of many subnuclear phenomena and made some spectacular
predictions later experimentally confirmed. However, a mathematically con-
sistent full-scale construction of a relativistic quantum field theory in four-
dimensional space-time, let alone of the Standard Model, is still missing.
Therefore, different aspects have to be accessed by particular methods. We
have described these approaches as distinct facets of quantum field theory
and have subsequently argued to interpret them as different codes which com-
plement each other. As regards the Standard Model, perturbative renormali-
sation theory accounts for short-distance effects, whereas lattice gauge theory
aims at ‘large distances’, i.e. at those characterising the ‘size’ of hadrons.
It has been pointed out, that the respective relation between quantum fields
and associated particles differs within these codes. Moreover, we have con-
sidered the concept of an effective field theory and a structural realist’s view,
which have received particular attention in particle physics and philosophy
of science. Evaluating their virtues and deficiencies, we have again stressed
the advantages of the semiotic perspective.
There are strong indications that the semiotic point of view also proves
appropriate to assess the analysis of experiments, the host of phenomenolog-
ical models and theoretical speculations in the realm of particle physics.
The extension of the semiotic approach beyond the scientific enterprise
of particle physics, however, is even more fascinating. It is possible in so far
45
as all human and cultural activities can be understood as (transformations
of) symbols. Though they differ widely in their respective details, famously
such philosophies of symbols have been brought forward by Charles S. Peirce,
Ernst Cassirer, Alfred N. Whitehead, Nelson Goodman or Susanne Langer.
46
References
T. Appelquist and J. Carrazone. Infrared singularities and massive fields.
Phys. Rev., D11:2856–2861, 1975.
S.Y. Auyang. How Is Quantum Field Theory Possible? Oxford University
Press, Oxford, 1995.
F.A. Berezin. The Method of Second Quantization. Academic Press, New
York, 1966.
N.N. Bogoliubov and O.S. Parasiuk. Über die Multiplikation der Kausal-
funktionen in der Quantentheorie der Felder. Acta Math., 97:227–266, 1957.
R. Brunetti and K. Fredenhagen. Microlocal Analysis of Interacting Quantum
Field Theories: Renormalization on Physical Backgrounds. Commun. Math.
Phys., 208:623–661, 2000.
D. Buchholz and R. Haag. The Quest For Understanding in Relativistic
Quantum Physics. J. Math. Phys., 41:3674–3697, 2000.
T.Y. Cao and S.S. Schweber. The Conceptual Foundations and the Philo-
sophical Aspects of Renormalization Theory. Synthese, 97(1):33–108, 1993.
Tin Yu Cao. Conceptual developments of 20th century field theories. Cam-
bridge University Press, Cambridge, UK, 1996.
T.Y. Cao. Representation or Construction? An Interpretation of Quantum
Field Theory. In T.Y. Cao, editor, The Proceedings of the Twentieth World
Congress of Philosophy, volume 10 (Philosophy of Science), pages 115–123,
Bowling Green, 2001. Philosophy Documentation Center.
A. Chakravartty. The Structuralist Conception of Objects. Manuscript,
2002.
M. Disertori and V. Rivasseau. Continous Constructive Fermionic Renor-
malization. Ann. Henri Poincaré, 1:1–57, 2000.
H.G. Dosch and V.F. Müller. Renormalization of quantum electrodynamics in
an arbitrary strong time independent external field. Fortschr. Phys., 23:661–
689, 1975.
Pierre Duhem. La Théorie Physique. Vrin, Paris, 2 edition, 1914 (1957).
47
F.J. Dyson. The S Matrix in Quantum Electrodynamics. Phys. Rev.,
75:1736–1755, 1949.
F.J. Dyson. The Radiation Theories of Tomonaga, Schwinger, and Feynman.
Phys. Rev., 75:486–502, 1949.
U. Eco. La struttura assente. Bompiani, Milano, 1968.
H. Epstein and V. Glaser. The role of locality in perturbation theory. Ann.
Inst. H. Poincaré, 19:211–2955, 1973.
L.D. Faddeev and A.A. Slavnov. Gauge Fields: An Introduction to Quantum
Theory. Addison-Wesley, 2 edition, 1991.
J. Feldman, J. Magnen, V. Rivasseau, and R. Sénéor. A renormalizable field
theory: the massive Gross-Neveu model in two dimensions. Comm. Math.
Phys., 103:67–103, 1986.
Paul Feyerabend. Against Method. Outline of an anachistic theory of knowl-
edge. Verso, London, 1975.
R.P. Feynman and M. Gell-Mann. Theory of the Fermi Interaction. Phys.
Rev., 109:193, 1958.
K. Gawedzki and A. Kupiainen. Gross-Neveu Model Through Convergent
Perturbation Expansion. Comm. Math. Phys., 102:1–30, 1985.
J. Glimm and A. Jaffe. Quantum Physics. Springer, New York, 2 edition,
1987.
D.J. Gross and D. Wilczek. Ultraviolet behavior of non-Abelian gauge the-
ories. Phys. Rev. Lett., 30:1343–1346, 1973.
K. Hagiwara and others. Review of Particle Properties. Physical Review D,
66:010001–1 –974, 2002.
S. Hartmann. Effective Field Theories, Reductionism and Scientific Expla-
nation. Studies in History and Philosophy of Modern Physics, 32(2):267–304,
2001.
Hermann Helmholtz. Die Lehre von den Tonempfindungen. Vieweg, Braun-
schweig, 4 edition, 1877. Engl. Translation by A. J. Ellis, Dover 1954.
Hermann von Helmholtz. Handbuch der physiologischen Optik. 2 edition,
1892.
48
Hermann von Helmholtz. Die Tatsachen in der Wahrnehmung. In Paul
Hertz and Moritz Schlick, editors, Hermann von Helmholtz, Schriften zur
Erkenntnistheorie, page 109ff. Springer, Berlin, 1921.
K. Hepp. Théorie de la renormalisation. Springer, Berlin, 1969.
Heinrich Hertz. Über die Beziehung von Licht und Elektricität. Strauss,
Bonn, 1889. Also in Gesammelte Werke, Bd I.
Heinrich Hertz. Die Prinzipien der Mechanik. Barth, Leipzig, 1 edition, 1894.
Mit einem Vorworte von H. v. Helmholtz.
K.A. Intriligator and N. Seiberg. Lectures on supersymmetric gauge theo-
ries and electric-magnetic duality. Nucl. Phys. Proc. Suppl., 45BC:1, 1996.
arXiv:hep-th 9509066.
C. Itzykson and J.-B. Zuber. Quantum Field Theory. McGraw-Hill, New
York, 1 edition, 1980.
F. Jackson. From Metaphysics to Ethics. Clarendon Press, Oxford, 1998.
R. Jost. The General Theory of Quantized Fields. American Mathematical
Society, Providence, RI, 1965.
C. Kim. A Renormalization Group Flow Approach to Decoupling and Irrel-
evant Operators. Ann. Phys., 243:117–143, 1995.
Thomas S. Kuhn. The Structure of Scientific Revolutions. Chicago, 2 edition,
1970.
J. Ladyman. What is Structural Realism? Studies in History and Philosophy
of Science, 29(3):409–424, 1998.
G.W. Leibniz. Mathematische Schriften. Berlin, 1849. Her. Gerhardt.
F. London. Quantenmechanische Bedeutung der Theorie von Weyl. Z.Phys.,
42:375, 1927.
J.H. Lowenstein. Convergence Theorems for Renormalized Feynman Inte-
grals with Zero-Mass Propagators. Comm. Math. Phys., 47:53–68, 1976.
I. Montvay and G. Münster. Quantum Fields on a Lattice. Cambridge Univ.
Press, Cambridge, 1 edition, 1994.
49
Lochlain O’Raifeartaigh and Norbert Straumann. Early History of gauge
Theories and Kaluza-Klein Theories, with a Glance at recent developments.
hep-ph/9810524, 1998.
K. Osterwalder and R. Schrader. Axioms for Euclidean Green’s functions.I.
Commun. Math. Phys., 31:83– 112, 1973.
K. Osterwalder and R. Schrader. Axioms for Euclidean Green’s functions.II.
Commun. Math. Phys., 42:281– 305, 1975.
Henri Poincaré. La Valeur de la Science. Flammarion, Paris, 1905.
Henri Poincaré. La Science et l’Hypothèse. Flammarion, Paris, 1927.
H.D. Politzer. Reliable perturbative results for strong interactions? Phys.
Rev. Lett., 30:1346–1349, 1973.
S. Psillos. Is Structural Realism Possible? Philosophy of Science,
68(Proc.):S13–S24, 2001.
D. Robinson. Renormalization and the Effective Field Theory Programme.
In D. Hull, M. Forbes, and K. Okruhlik, editors, Proceedings of the 1992
Biennial Meeting of the Philosophy of Science Association, volume 1, pages
393–403, East Lansing, 1992. Philosophy of Science Association.
Norbert Straumann. Zum Ursprung der Eichtheorien bei Hermann Weyl.
Phys. Bl., 43:414, 1987.
R.F. Streater and A.S. Wightman. PCT, Spin and Statistics, and all That.
Benjamin, Reading, MA, 3 edition, 1980.
G. ’t Hooft and M. Veltman. Combinatorics of Gauge Fields. Nucl. Phys.,
B50:318–353, 1972.
G. ’t Hooft. Renormalizable Lagrangians for Massive Yang-Mills Fields. Nucl.
Phys., B35:167–188, 1971.
G. ’t Hooft. Renormalization of Massless Yang-Mills Fields. Nucl. Phys.,
B33:173–199, 1971.
M. Veltman. Facts and Mysteries in Elementary Particle Physics. World
Scientific, New Jersey, 2003.
S. Weinberg. Phenomenological Lagrangians. Physica, A 96:327, 1979.
50
Hermann Weyl. Was ist Materie. Berlin, 1924. reprinted Darmstadt 1977.
Hermann Weyl. Electron and Gravitation. I. Z. Phys., 56:330, 1929.
Hermann Weyl. Wissenschaft als symbolische Konstruktion des Menschen.
Eranos-Jahrbuch, page 375, 1949. Reprinted in Gesammelte Abhandlungen,
K. Chandraskharan ed., Berlin 1968.
H. Weyl. Symmetry. Princeton University Press, Princeton (N.J.), 1952.
Hermann Weyl. Über den Symbolismus der Mathematik und mathematis-
chen Physik. Studium generale, 6:219, 1953. Reprinted in Gesammelte Ab-
handlungen, K. Chandraskharan ed., Berlin 1968.
H. Weyl. Philosophie der Mathematik und Naturwissenschaft. Oldenbourg
Verlag, Munich, 2000.
E.P. Wigner. On unitary representations of the inhomogeneous Lorentz
group. Ann. Math., 40:149–204, 1939.
K.G. Wilson and J. Kogut. The Renormalization Group and the Epsilon -
Expansion. Phys. Rep., 12C:75–199, 1974.
K.G. Wilson. Confinement of quarks. Phys. Rev., D10:2445–2459, 1974.
J. Worral. Structural Realism: The Best of Both Worlds? In D. Papineau,
editor, The Philosophy of Science, pages 139–165. Oxford University Press,
Oxford, 1996.
E. G. Zahar. Poincaré’s Structural Realism and his Logic of Discovery. In
J.-L. Greffe, G. Heinzmann, and K. Lorenz, editors, Henri Poincaré: Science
and Philosophy, pages 45–68. Akademie Verlag, Berlin, 1996.
W. Zimmermann. Local Operator Products and Renormalization in Quan-
tum Field Theory. In Lectures on Elementary Particles and Quantum Field
Theory, volume 1, Cambridge, MA, 1970. M.I.T Press.
51

Ms Dosch Muller Sieroka

Uploaded by

Copyright:

Available Formats

Ms Dosch Muller Sieroka

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ms Dosch Muller Sieroka

Uploaded by

Copyright:

Available Formats

Quantum Field Theory,

its Concepts Viewed from a Semiotic Perspective

Hans Günter Dosch1 , Volkhard F. Müller2 ,

2 Relativistic Quantum Field Theories

2.2 The Notion of a Relativistic Quantum Field

2.3 Local Gauge Symmetry

2.4 Particles versus Quantum Fields

2.5 The Facets of Quantum Field Theory

2.5.1 General Theory of Quantum Fields

2.5.2 Perturbative Renormalisation Theory

2.5.3 Constructive Quantum Field Theory

2.5.4 Lattice Gauge Theories

2.5.5 Local Quantum Physics

2.6 Renormalisability versus Effective Field Theories

3 Particles and Fields

• 1) Perturbation theory: In certain reactions and under certain kine-

• 2) Lattice gauge theory: There is up to now no analytic way to treat

• 3) Non-relativistic quantum field theory: After the discovery of

4 Theories of Signs and Symbols, and Struc-

We form for us inner phantoms (Scheinbilder) or symbols of the

The images we speak of are our imaginations of the things; they

He also points out that by this congruence requirement the construction

[. . . ] the aim of Fresnel was not to find out whether there

In summary, the only objective reality are the relations between

Weyl also emphasises that transformation invariance and symmetry be-

5 A Theory of Symbols for

5.2 A Semiotic View of Quantum Field Theory

You might also like