1–32

10.1093/philmat/nkz002
Philosophia Mathematica

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Philosophy of Mathematical
Practice — Motivations, Themes
and Prospects†
Jessica Carter∗
Department of Mathematics and Computer Science, University of Southern Denmark,
5230 Odense, Denmark

A BS T R AC T
A number of examples of studies from the field ‘The Philosophy of Mathematical
Practice’ (PMP) are given. To characterise this new field, three different strands
are identified: an agent-based, a historical, and an epistemological PMP. These
differ in how they understand ‘practice’ and which assumptions lie at the core
of their investigations. In the last part a general framework, capturing some
overall structure of the field, is proposed.

1. INTRODUCTION
In his introduction to The Philosophy of Mathematical Practice, Paolo Mancosu
presents a new direction in the philosophy of mathematics, writing

The contributions presented in this book are thus joined by the shared
belief that attention to mathematical practice is a necessary condition for
a renewal of the philosophy of mathematics. We are not simply proposing
new topics for investigation but are also making the claim that these
topics cannot effectively be addressed without extending the range of
mathematical practice one needs to look at when engaged in this kind of
philosophical work. Certain philosophical problems become salient only

†
Silvia De Toffoli, José Ferreirós, Valeria Giardino, Marcus Giaquinto, Colin McLarty,
Marco Panza, and Jean Paul Van Bendegem read drafts, parts of drafts, or responded to my
queries: I thank them all for giving valuable advice. Thanks also to the three anonymous
referees for their insightful comments and constructive criticism.
∗
Orcid.org/0000-0001-9957-9961. E-mail: [email protected]

Philosophia Mathematica (III) Vol. 27 No. 1 c The Author [2019]. Published by Oxford University Press.
All rights reserved. For permissions, please e-mail: [email protected]
• 1
2 • Carter

when the appropriate area of mathematics is taken into consideration.

[Mancosu, 2008, p. 2]

This raises two (related) questions. The first is: what is meant by ‘practice’ ?

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

The second is how to characterise this new way of doing philosophy, taking
mathematical practice seriously, captured by the title Philosophy of Mathe-
matical Practice. Both questions will be dealt with in this paper. In addition,
a range of questions and topics studied under the heading of ‘philosophy of
mathematical practice’ will be presented.
When I write about the philosophy of mathematical practice — from now
on abbreviated PMP — note the following. I do not intend to claim that there
is a necessary tension or conflict between ‘philosophy of mathematical practice’
and ‘philosophy of mathematics’ (although some might claim there is). It is
certainly the case that some of the first proponents of a reorientation in the
philosophy of mathematics thought that the previous way of doing philosophy
was wrong. In the end I will propose that philosophy of mathematical practice
could simply be philosophy of mathematics, although of a particular kind, where
(i) mathematics is taken to be mathematics in every shade and not idealisations
of mathematics, and (ii) an extension of methods is allowed, i.e., the possibility
of bringing in results and tools from other disciplines.
The structure of the paper is as follows. The second section recaps some
of the motivations that have been advanced for changing the direction of the
philosophy of mathematics. The trend observed here is that motivations have
moved from being mainly negative towards the previous way of doing philoso-
phy to a realisation and acceptance that different perspectives are possible when
doing philosophy of mathematics. The third section tries to give a picture of the
current PMP. In order to do this I identify three major strands, characterised
in terms of which assumptions are chosen to be at the core of the investigations,
and in what is taken as the field of investigation, i.e., what counts as ‘practice’.
Finally there is a discussion on how to understand ‘practice’ based on these
strands and a conclusion.

2. MOTIVATIONS
Much has been written about the motivations for turning away from traditional
philosophy of mathematics (see [Corfield, 2003; Ferreirós and Gray, 2006; Man-
cosu, 2008] and others mentioned throughout this text). It is not my intention
to restate all these considerations; rather the aim is to give a few illustrative
examples and to direct attention to the more positive points: first the PMP
proponents’ insistence that the study of practice opens up many interesting
questions that can be dealt with besides traditionally asked questions; and
second a development of the studies of philosophy of mathematics towards
reconciliation between different ways of doing philosophy and integration of
different methodologies.
In the mid-twentieth century a growing dissatisfaction with the concerns of
the philosophy of mathematics led to studies of mathematical practice. Motiva-
tions grew out of a perceived limitation both of the topics studied and methods
 Philosophy of Mathematical Practice • 3

used.1 For one thing philosophers were critical towards the one-sided focus
on foundational questions, that is, focus on logic, set theory, and the three
foundational schools, logicism, formalism and intuitionism. See, e.g., [Putnam,
1967]2 and [Tymoczko, 1998].3 Among these critics Imre Lakatos is often men-

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

tioned as the first to engage with the philosophy of mathematical practice. In a
famous passage he expressed the following: ‘paraphrasing Kant: the history of
mathematics, lacking the guidance of philosophy, has become blind, while the
philosophy of mathematics, turning its back on the most intriguing phenom-
ena in the history of mathematics, has become empty’ [1976a, p. 2; emphasis
original].4 Lakatos’s Proofs and Refutations [1976a] illustrates how these faults
can be remedied. It is based on an extensive case study on the historical devel-
opment of Euler’s formula for polyhedra. If V denotes the number of vertices,
E, the number of edges and F , the number of faces of a given polyhedron, the
formula V − E + F = 2 holds. According to the account given by Lakatos,
the first step consists of giving a proof of the formula so that it becomes a
theorem. In the next step the theorem is challenged by the discovery of a num-
ber of counterexamples, referred to as ‘monsters’. The monsters are handled
by, for example, modifying the definition of the concept of a polyhedron. In
a later step it is proposed that only specific types of polyhedra fulfil the for-
mula.5 Towards the end, Lakatos introduces the notion of a ‘proof generated
theorem’. This means essentially that one considers the constructions made
in the lemmas of the proof and determines which types of polyhedra support
these constructions. The theorem is thus valid for such polyhedra. A lesson
Lakatos learns from this case study is that the formulations of theorems and
their proofs are interrelated in a forever ongoing process. In an earlier paper
‘A renaissance of empiricism’ [Lakatos, 1976b] further argues that the ideal

1
A number of authors also mention the ‘Kuhnean turn’ in the philosophy of science
as inspiration for a similar development in the philosophy of mathematics, at the same
time regretting that the turn in mathematics has been very slow to come about (see e.g.,
[Giardino et al., 2012]). It should be noted that even before the mid-twentieth century
there were studies taking the ‘practice’ of mathematics into account. In French philosophy,
for example, one could in particular mention the work of Cavaillès and Lautman taking
(the history of) mathematics as an important part of philosophy; see [Benis Sinaceur,
2006] and [Lautman, 2011].
2
In ‘Mathematics without foundations’ Hilary Putnam objects that the ‘famous “isms”
in the philosophy of mathematics’ (p. 7) do not contribute to a clarification of our con-
ceptions of mathematical truth, objects, or necessity. On the foundations of mathematics
he further writes ‘I don’t think mathematics has a crisis in its foundations; indeed, I do
not believe mathematics either has or needs “foundations” ’ [1967, p. 5].
3
The volume New Directions in the Philosophy of Mathematics (1986) contains
a number of contributions criticising the dominant ‘foundational perspectives’ in the
philosophy of mathematics. The editor, Thomas Tymoczko, writes about the contribu-
tors’ motivations that they ‘were frustrated by the inability of traditional philosophical
formulations to articulate the actual experience of mathematicians’ [Tymoczko, 1998,
p. ix].
4
The reference to Kant is to his Critique of Pure Reason, A51/B51.
5
Proposals include, e.g., that simple or convex polyhedra are ‘Eulerian’.
4 • Carter

of ‘Euclidean’ mathematics should be given up. This is the picture of mathe-

matics (or science) where one picks out true axioms and places them at ‘the
top’. Propositions are then derived from these so that ‘truth flows downward
in the system’ [Lakatos, 1976b, p. 205]. Instead Lakatos claims that method-

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

ologically mathematics is more like the natural sciences and so discusses which
type of statements could be considered as falsifiers of an axiomatic theory. The
monsters found in his Proofs and Refutations can be seen as examples of such
falsifiers. Lakatos introduces the label quasi-empiricism for views of this sort.
Another route to PMP comes from a perceived lack of connection between
mathematics and the debate on the ontology of mathematics.6 Two influential
articles by Paul Benacerraf [1965; 1973] have set the stage for work in this
debate.7 Arguments often start out from assumptions about mathematics with-
out testing whether these correspond to what mathematicians do or think. One
such assumption is that mathematical theorems are necessarily true. From this
assumption it has been argued (e.g., [Katz, 1998]) that mathematical objects
exist by necessity. Another assumption employed is that some mathematical
statements are true (accompanied by the argument that truth implies exis-
tence). Indispensability arguments consist of (at least) two components. The
first notes that mathematics is indispensable to natural science. The second,
referred to as ‘confirmational holism’, claims that mathematical statements are
themselves confirmed whenever the scientific theories in which they take part
are confirmed.8 All these assumptions face similar reactions as above — as do
most counter assumptions in the debates on realism and anti-realism (e.g., the
fictionalist claim that mathematical statements are not true). PMP may enter
the picture in trying to investigate these assumptions in terms of mathematical
practice.
As an example of addressing questions from ontology by drawing on mathe-
matical practice one could mention the contributions of Penelope Maddy. Her

6
Aldo Antonelli [2001] points to the fact that mathematics often is used as a test
case of metaphysicians’ positions on abstract objects and that mathematical facts rarely
enter such investigations. Antonelli instead urges philosophers to engage in ‘Mathematical
Philosophy’ (so coined by Russell) about which he says:

on the other hand, is respectful of, but not subsidiary to, current mathematical prac-
tice. It engages the issues, points out conceptual tensions, and highlights unexpected
consequences. Mathematical philosophy positions itself neither above nor below
mathematics, but rather on a par with it, taking the role of an equal interlocutor
[2001, p. 1].

7
Benacerraf’s ‘What numbers could not be’ [1965] has in part inspired a number of
structuralist positions, whereas his ‘Mathematical truth’ [1973], arguing that a causal
theory of knowledge is incompatible with platonism, is either used to argue for anti-realism
or to offer the challenge for realist positions to account for our knowledge of mathematical
statements.
8
There are in fact a number of different versions of the indispensability argument,
but all are based on the assumption that mathematics is indispensable to the theories of
natural science. See [Panza and Sereni, 2013, Chs 6, 7].
 Philosophy of Mathematical Practice • 5

work is informed by studies of the history and practice of set theory. Maddy
considers, for example, set theorists’ discussions about which criteria should be
used in order to adopt set-theoretical principles such as the continuum hypoth-
esis (CH), Gödel’s constructibility axiom, V = L, and the question concerning

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

which definable sets of reals are Lebesgue-measurable. These principles all have
their origin in analysis. One of the problems that led Cantor to his transfinite set
theory (and so the formulation of CH) concerned representations of functions
by series of trigonometric functions. Lebesgue measure was introduced in order
to extend the type of functions that can be integrated. Maddy’s ‘Indispens-
ability and practice’ [1992] compares (among other things) discussions on the
foundations of set theory with components of the indispensability argument.
She first argues that a naturalist philosopher of mathematics should not only
take the practice of science, but also the practice of mathematics, into account
when assessing claims about mathematics. In this spirit she finds that the math-
ematicians’ discussion of the principles of set theory contradict, for example,
confirmational holism. According to Maddy, the adoption of principles does
not depend on physical facts, for example, whether a particular version of the
continuum has applications in some scientific theory. Instead set theorists refer
to arguments that are based on internal mathematical considerations, such as
how restrictive a particular principle is.

2.1. Reconciliations — Towards PMP

We saw in the previous section that criticism of the foundational perspectives
led to an interest in the development of mathematical ideas, that is, the heuris-
tics of mathematics. In addition a number of scholars have demanded that the
philosophy of mathematics should pay more attention to the human dimen-
sion of mathematics. See for example [Kitcher, 1984] and [Tymoczko, 1998].
The outcomes are positions placing an emphasis on the human agents doing
mathematics and dealing with social aspects of mathematics.9 In particular the
mathematician Reuben Hersh [1979] has been a strong advocate for a shift of
focus to pay attention to social and cultural aspects of mathematical practice.
Today scholars within PMP wish to advance a more balanced position. One
does not have to adhere either to a position that mathematics is at core human
or that it is certain. As we shall see there are scholars in contemporary PMP
taking the assumption that mathematical knowledge is human knowledge while
still trying to explain its special certainty. In addition some of the previously

9
A few scholars have even taken the failure of the foundational schools to provide
mathematics with a secure foundation to support a claim that mathematical knowledge
is fallible. See, for example, [Ernest, 1998; Oliveri, 2010; Hersh, 2013]. Mancosu [2008]
comments on the lack of success of studies turning their back on foundations and reduc-
ing mathematics to a cultural product, that they are unable to account for distinguishing
features of mathematics such as its special certainty. He writes that ‘Logically trained
philosophers of mathematics and traditional epistemologists and ontologists of mathemat-
ics felt that the “mavericks” were throwing away the baby with the bathwater’ [2008,
pp. 5–6]. See also [Leng, 2002] for a criticism of fallibilist theories of mathematics.
6 • Carter

mentioned reactions indicate that studies of foundations are incompatible with

quasi-empiricism (as it was coined by Lakatos). Scholars have since found that
both empiricism and quasi-empiricism may be combined with positions on the
foundations of mathematics.10 In support of this view see [Schlimm, 2010] for

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

an account of how Pasch combined his foundations with empiricism and Fer-
reirós’ exposition [2006] of nineteenth-century mathematicians’, e.g., Riemann’s
use of the word ‘foundation’.
Inspired by Lakatos’s Proofs and Refutations, scholars have started to use
(historical) case studies as a means to investigate philosophical questions. This
has created a number of interrelations between history and philosophy of math-
ematics as will be explained in the next section. Case studies on mathematical
practice are not limited to historical cases and the study of written texts. As
illustrated by the work of Maddy, philosophers also take into account the con-
temporary practice of mathematicians, for example, what they say about their
motivation for adopting various principles. It is therefore clear that ‘practice’
could mean a number of different things depending on what type of question
one wishes to investigate and one’s (philosophical) assumptions. I propose that
it is possible to ask questions at different levels — and as we shall see under
different interpretations of the notion of ‘practice’. At the top level could be
questions about the true nature of mathematical objects and our knowledge of
these. At a lower level are the questions that arise in the practice of engaging
with mathematics. Since there are different ways of engaging with mathematics
(as developed in more detail in [Giaquinto, 2005a], there are different types of
questions to be asked. There is the activity of mathematicians producing new
material (heuristics), the practices of learning and teaching mathematics, and
the practices of using mathematics in science or society.

3. CURRENT SCENE: WHAT IS PMP?

In order to get a handle on PMP, and investigate whether it is possible to
characterise ‘practice’, we start by painting a picture of the contemporary scene.
I shall present examples from PMP and, in order to give the presentation some
structure, sort them into different categories. I find it is possible to identify three
different, sometimes overlapping, strands:11 an agent-based, a historical, and
an epistemological strand. These differ in various respects, e.g., in what aspects
of ‘practice’ they consider and so in which assumptions are built in, what the
aims are for PMP, and which methods are brought in to study the questions
posed. What is offered is a rough classification. It is not always possible to place
philosophers or a single work in any one of these categories as many have broad

10
See Patrick Peccatte’s FOM posting of September 1998
(https://cs.nyu.edu/pipermail/fom/1998-September/002097.html) for an elaboration of
this point.
11
See also [Kjeldsen and Carter, 2015].
 Philosophy of Mathematical Practice • 7

interests.12 It should also be noted that other categorisations could be made,

see, e.g., [Van Bendegem, 2014].

3.1. Agent-Based PMP

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Central to the agent-based strand is the belief that a philosophy of mathematics
is impossible without taking into account the human beings doing mathematics.
The focus here is thus on the practitioners of mathematics: their activities in
developing, using, or learning mathematics. There are two major developments
within this strand. One has strong interconnections with sociology and a built-
in assumption that mathematics is a social activity. The other is a pragmatic
orientation following philosophers like Peirce, Dewey, and Putnam.
Depending on the chosen perspective for such a study, different method-
ologies can be brought in. Sociology is used to study the interactions between
human beings, whereas results from cognitive science are used in order to study
the cognitive functioning of the individual’s mind conveying the ability to form
mathematical concepts. Furthermore there are interrelations with other fields,
such as anthropology and mathematics education.

3.1.1. The Social Strand of Agent-Based PMP

The sociological strand has grown out of works such as those of the sociolo-
gists Bloor [1976], Restivo (e.g., [1993]), and Heintz [2000]. David Bloor and
Sal Restivo (in their different ways) have developed pictures of mathematics
based on the idea that mathematics is a social activity. Bettina Heintz bases
her conclusions about mathematics on a sociological study of mathematicians.
Because of Hersh’s insistence that a philosophy of mathematics should account
for the fact that human beings do mathematics and since he claims mathemati-
cal objects are ‘social-cultural-historical’, I also place Hersh (e.g. [1979]) in this
category of PMP. In his latest book he lists the following five points:

1. Mathematics is human.
2. Mathematical knowledge is fallible.
3. There are different versions of proof or rigour.
4. Empirical evidence, numerical experimentation and probabilistic
proof all can help us decide what to believe in mathematics.
5. Mathematical objects are a special variety of a social-cultural-
historical object [2013, p. 169].

Agent-based, and in particular social PMP, naturally has ties to mathematics

education. Mathematics educators have long been frustrated with the ‘dehu-
manised’ traditional philosophy of mathematics, which some, e.g. Paul Ernest,
claim only offer ‘absolutist’ (referring to platonism and formalism) views of

12
Scholars contributing to at least two of these strands include Andy Arana, José
Ferreirós, Valeria Giardino, Colin McLarty, Marco Panza, Dirk Schlimm, and others
mentioned in the text.
8 • Carter

mathematics. They find such views are unfortunate when teaching mathemat-
ics.13 The exception is Lakatos, who is mentioned as the lone hero.14 Ernest, an
influential figure in the philosophy of mathematics education, has contributed
with an alternative, a social constructivist picture of mathematics (e.g., [1998]).

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Work combining, for example, sociology, philosophy, and education might
not be characterised as truly philosophical, but rather dealing with (philosoph-
ical) questions within the fields of sociology and mathematics education. A
number of studies, though, have tried to bridge the gaps among, e.g., the soci-
ology, history, and philosophy of mathematics and a network has been formed
with this as one of its goals, namely ‘Philosophy of Mathematics: Sociological
Aspects and Mathematical Practice’.15 In addition, several conferences have
been organised (mainly in Brussels) in order to advance such studies (see [Van
Kerkhove and Van Bendegem, 2007; Van Kerkhove, 2009; Van Kerkhove et al.,
2010]). One reason given for rejecting work combining philosophy with other
fields is that they depend on or use empirical facts, e.g., findings from socio-
logical or anthropological studies, when formulating positions on mathematics.
A recent program labelled ‘Empirical Philosophy of Mathematics’ has tried to
develop a method to rationally integrate empirical statements into the philoso-
phy of mathematics. In this program Löwe et al. [2010] first note that philosophy
needs some input.16 Input may consist of philosophers’ (often contradictory)
intuitions, but sometimes, they claim, philosophers also refer to empirical state-
ments. As an example of such a statement they mention ‘most mathematicians
are platonists’.
In the framework of ‘Empirical Philosophy of Mathematics’ Eva Wilhelmus
has performed a study [Wilhelmus, 2007] of mathematicians’ knowledge ascrip-
tions.17 Interestingly, her findings show that even though mathematicians claim
that mathematical knowledge is objective, they seem to hold that the truth
value of a theorem may change depending on available proofs or counterex-
amples. Her study further indicates that the acceptance of the validity of a
theorem depends on, e.g., the importance of the theorem or the reputation of
the mathematician proving it. The study consists of a questionnaire sent to a

13
Ernest [1996] characterises ‘absolutist views’ as being ‘concerned with the epis-
temological project of providing rigorous systems to warrant mathematical knowledge
absolutely.’
14
It has been argued that philosophy plays a key role when teaching and learning
mathematics. For one thing, the philosophical beliefs of both teacher and student influence
teaching/learning processes (see [Lerman, 1990; Prediger, 2007]).
15
See http://www.lib.uni-bonn.de/PhiMSAMP/.
16
Even when it is accepted that input could come from mathematics there is disagree-
ment concerning which areas of mathematics are relevant to study in order to say what
mathematics is. It could be debated whether the study of arithmetic suffices or whether
one should also include other areas of mathematics. For discussions concerning this issue,
see [Corfield, 2003, Introduction] and [Tappenden, 2008].
17
[Inglis and Aberdein, 2015] is another exemplary study employing empirical data
in a philosophical context. Inglis and Aberdein have systematically studied how math-
ematicians employ different terms such as ‘beauty’ and ‘simplicity’ to characterise
proofs.
 Philosophy of Mathematical Practice • 9

number of practising mathematicians. The first part asks about the mathemati-
cians’ conceptions of proofs. The majority of the mathematicians responding
to the study characterise a proof as something close to a formal proof, i.e., that
it consists of a list of statements, some of which are axioms, and the rest are

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

logical consequences of these. The second section of the questionnaire presents
a variety of scenarios. After each part of the story, the respondents are asked
to comment on the state of knowledge of the main character of the story. One
such story is about a PhD student, John, who finds a proof of his supervisor’s
conjecture, the ‘Jones conjecture’. After the proof of the conjecture has been
accepted for publication in a major journal, most participants of the study
answer that John knows that the theorem is true. The next part of the scenario
explains that John a few years after attends a seminar where he realises that
he can construct a counterexample to the theorem. As one would expect, a
majority of the participants (61%) then claims that John now knows that the
theorem is false. What surprises Wilhelmus, however, is their answer to ‘Did
John know that the Jones conjecture was true on the morning before the talk’
(p. 14): 71% answer positively. Wilhelmus concludes that mathematicians in
general do not use ‘formalizability of a proof’ as a criterion to assess that a
theorem is true.

3.1.2. The Pragmatic Strand of Agent-Based PMP

The pragmatic line of PMP sees it as central for philosophy to characterise
mathematical knowledge — with the qualification that it should be charac-
terised as knowledge possessed by human beings. This can in part be seen as a
rejection of the idea of having an ‘epistemology without subjects’ and opening
up considering knowledge as a product of human activities. Even though knowl-
edge is thus agent-based, proponents of a pragmatist version of knowledge hold
that mathematics cannot simply be reduced to human conventions. Further-
more they reject that mathematical knowledge is something that can only be
studied by sociological methods. They argue that at least parts of mathematics
are non-arbitrary and that, although mathematics is sometimes based on con-
ventions or hypotheses, mathematical knowledge is still objective. One source of
these ideas is C.S. Peirce, who famously characterised mathematics as the nec-
essary reasoning concerning hypothetical states of affairs, i.e., that it is possible
to draw necessary conclusions from formed hypotheses (see [Carter, 2014]). A
number of scholars have used Peirce’s ideas to develop pragmatist pictures of
mathematics. Among them are Michael Otte (also in relation to mathematics
education), Gerhard Heinzmann and Fernando Zalamea. A recent attempt at
formulating an agent-based theory of knowledge can be found in José Ferreirós’s
book Mathematical Knowledge and the Interplay of Practices [2016]. Ferreirós
explains how basic arithmetical truths (based on counting practices) are both
objective and certain. Other, more advanced, areas of mathematics are based
on hypotheses — but such hypotheses are often integrated with other practices
of mathematics, and so cannot be entirely arbitrary. One much discussed exam-
ple is the axiom of choice (AC). Formally it is independent of the axioms of ZF;
10 • Carter

so in principle one could adopt AC or choose not to. One reason it is widely
accepted among contemporary mathematicians is because it has many conse-
quences that they are not willing to give up, consequences that are integral to
the practice of, e.g., analysis. A simple example is that AC implies that two

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

different (standard) ways of defining continuity of functions are equivalent.18
Furthermore, Ferreirós [2011; 2016] has argued that there are very intimate
links between AC and the classical conception of the number continuum. Thus
some mathematical concepts may be rooted in our practical dealings with the
physical world (and others are based on hypotheses), but this does not entail
that we can define concepts at random. When trying to capture such ideas
mathematically, i.e., when defining new mathematical concepts or introduc-
ing new principles, there will often be certain constraints. A view like this is
expressed by David Hilbert in Natur und mathematisches Erkennen, where he
describes how one is led by experience and thought to the concept of continuum,
writing,

furthermore it is seen that the formation of concepts in mathematics is

constantly led by thought and experience, so that mathematics as a whole
is a non-arbitrary closed structure [Hilbert, 1992, p. 5].19

Both agent-based perspectives may find it relevant to study how human beings
form mathematical concepts, theories, and knowledge thereof and so draw on
results from cognitive science, see, for example, [Giaquinto, 2005b; 2007]).20
Marcus Giaquinto [2007] draws on cognitive theories in order to explain
how visualisation may contribute to the formation of beliefs in mathematics.
Although cognitive science could be an important contributor to PMP, scholars
have made a case that cognitive scientists have not yet been able to account
for how (complex) mathematical concepts are formed, see [Doridot and Panza,
2004] and [Ferreirós, 2016, e.g., pp. 65–67].
In addition anthropological studies are drawn upon in combination with
cognitive science as seen, for example, in the work of Helen De Cruz. De Cruz
[2009] investigates the claim made by a number of philosophers that geometric
knowledge is innate or, as proposed by Kant, that geometric knowledge is based
on our intuitions of space. She splits this claim into two parts. The first part,

18
Note that adopting AC (or any other axiom for that matter) because it has con-
sequences one is not willing to give up is only acceptable if it is consistent with, and
independent of, one’s underlying foundation.
19
The author’s translation of the sentence: Vielmehr zeigt sich, dass die Begriffsbil-
dungen in der Mathematik beständig durch Anschauung und Erfahrung geleitet werden,
so dass in großen und ganzen die Mathematik ein willkürfreies Gebilde darstellt.
20
Conversely there are also philosophically oriented scholars in cognitive science,
employing philosophical positions when developing theories of how human beings form
mathematical concepts. An example is Susan Carey (e.g. [Carey, 2004]) who, inspired by
Quine’s position on the formation of conceptual schemes, formulates her bootstrapping
account on how natural numbers are acquired. Others refer to Peano’s characterisation of
the natural numbers.
 Philosophy of Mathematical Practice • 11

‘do we have innate intuitions of space?’, is compared to findings from cognitive

science, neuroscience ,and anthropology. Based on these studies she finds that it
is fair to claim that human beings have intuitions of space. Second she considers
parts of the history of non-Euclidean geometry in order to argue for her second

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

claim, i.e., that until the nineteenth century it seems reasonable to say that
geometry was based on our intuitions of space. In this story she starts with the
fifth postulate of Euclid’s geometry and presents aspects of the long history
that led to the formulation and acceptance of the non-Euclidean geometries
by Lobachevsky, Bolyai, and Gauss in the nineteenth century. De Cruz notes
that, in the many attempts at proving that the Euclidean geometry was the
only possible, mathematicians often unconsciously based their arguments on
principles that are equivalent to the fifth postulate. Another component of her
argument consists of the strong resistance to accepting alternative versions of
geometry.
As illustrated by the above examples, ‘practice’ in agent-based PMP refers
mainly to the fact that mathematics is (also) a human activity focussing on
the agents, real or idealised, doing mathematics. Furthermore this perspective
draws on a number of different methodologies and fields such as empirical stud-
ies, cognitive science, and anthropology (as in the studies by Wilhelmus [2007],
De Cruz [2009], and Heintz [2000]).

3.2. Historical PMP

According to the historical PMP it is a major concern that mathematics has a
history and this gives rise to a number of interesting questions that belong to
philosophy. Whereas the agent-based strand stresses the fact that human beings
do mathematics, one could claim that historical PMP focuses on the outcome of
these activities and, most importantly, questions related to how this outcome,
that is, mathematics, has been shaped across time. An underlying assumption
of this perspective is thus that mathematics is the product of certain activities
and not a static theory. Answers to historical questions could involve adopting
a number of different perspectives. According to recent understandings of the
history of mathematics, cultural and social factors play a more central role; see
[Aspray and Kitcher, 1988, pp. 20–31].
Themes within this strand include how history and philosophy influence
each other, i.e., that philosophical views may shape mathematics, and that the
development of mathematics at any given time may raise philosophical ques-
tions. Indeed, given the special status that mathematical knowledge has had
throughout time, the development of mathematics has inspired much philosoph-
ical thought. The history of mathematics may also be seen as a natural provider
of cases for philosophy as was advised by Lakatos. Concerning the nature of
the relation between history and philosophy, however, there is disagreement.
Opinions on this range from the view that history is inherently philosophically
laden [Oliveri, 2010] to a position that history — or studies of practice — are
independent of philosophical concerns [Maddy, 1997, pp. 200–205]. Conversely,
Moritz Epple — as will be shown below — claims that history makes good
12 • Carter

use of philosophical categories. Finally Mancosu [2008, pp. 17–18] writes that
it is still too early to say anything conclusive concerning this ‘metaphysical
relation’.
In the following I present a few illustrative examples of the above

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

themes.21

3.2.1. Historical Circumstances for the Development of Philosophical Views

It is generally acknowledged that mathematics underwent profound changes
during the nineteenth century. The three foundational schools tried each in
their own way to deal with these changes. Whereas the mathematical back-
ground and historical motivation of Brouwer and Hilbert are often referred to,
one is rarely told much about Frege’s. Exceptions to this are Marco Panza
[2015] and Jamie Tappenden [2006]. According to Tappenden, the stories that
are told about Frege’s motivation are not correct. Tappenden calls them myths
and even goes so far as to write that they are ‘wildly wrong’. Stories give a one-
sided and narrow account about mathematical developments in the nineteenth
century. They focus on the so-called ‘arithmetisation of analysis’ as done by
Weierstrass, that is the project of freeing analysis from its geometrical foun-
dation by introducing the real numbers and basing reasoning on epsilon-delta
calculations. Since it is possible to base the real numbers on the natural num-
bers, the mathematical task becomes to found them. The later counter myth
states that Frege’s concerns were philosophical rather than mathematical, ques-
tioning his motivations for asking for a higher degree of rigour in mathematics.
This myth claims that (via the work of Weierstrass) the problems of analysis
had already been solved when Frege began his work. Tappenden points out that
many problems had been solved in real analysis. Frege’s mathematical work was
within the areas of geometry and complex analysis. When Frege writes ‘analy-
sis’, he refers to complex analysis. Tappenden shows that in the last part of the
nineteenth century this field was in wild disarray, with diverging definitions of
fundamental concepts, in particular that of a function, giving plenty of reasons
why one would want to engage in philosophical queries. Tappenden places Frege
as a follower of the Riemann tradition (and a critic of Weierstrass).22 With the
work of Riemann, geometrical intuition was reintroduced into analysis and it
was by no means clear at the time how to define analysis, geometry, and their
interrelation. Given the state of complex analysis one understands the urge to
supply a foundation. What is particularly worth emphasizing is Tappenden’s

21
For other themes as well as ways that history and philosophy interact, see the volume
edited by Breger and Grosholz [2000].
22
Many facts support this. To mention a few, Frege did his doctorate at Göttingen,
where Riemann worked. Many of Frege’s friends and teachers, e.g., Abbe and Schering,
were followers of the Riemannian tradition. It is also possible to document that Frege regu-
larly taught courses on Riemannian topics and in several places criticised both Weierstrass
and his followers Biermann and Kossak.
 Philosophy of Mathematical Practice • 13

agenda of making readers aware of the many other themes under debate, for
example:

the issue of the fruitfulness of concepts was of paramount importance,

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

this was bound up in intricate and sometimes surprising ways with
the development of geometry and geometric interpretations of analysis.
[Tappenden, 2006, p. 99]

Like Tappenden, Panza takes up issues beyond the traditional accounts of

Frege’s logicism. The motive is to explain the role functions play in Frege’s
foundational program. In disagreement with Tappenden, Panza rejects the Rie-
mannian influence when it comes to Frege’s notion of a function. One aim of
Panza’s paper is to point out crucial differences from the current set-theoretic
foundations. A careful reading of Frege’s work as well as a historical study of
his mathematical predecessors reveal that Frege’s characterisation of the notion
of a function, and the role functions played, are closer to Lagrange’s concep-
tion. To both Lagrange and Frege a function is a primitive in their respective
foundations. According to Frege the notion of a function is a prerequisite for his
system of logic as presented in his Grundgesetze der Arithmetik. For Lagrange
functions serve to give analysis an algebraic foundation. In addition both regard
a function as somehow associated with its expression, but whereas Lagrange
identified a function with its expression, to Frege a function is not an object
and so cannot be an expression — although ‘the idea of a function detached
from any appropriate expression is merely inconceivable’ [Panza, 2015, p. 92].
The use of the expression is to present the law specifying from which primitives
the function is composed. As a consequence the notion of an arbitrary function
is not available to Frege.
Historical studies may also be based on questions that have philosoph-
ical leanings, such as attempts to understand how ‘simplicity’, ‘generality’
[Chemla et al., 2016], or ‘abstractness’ [Marquis, 2015] can be conceived of.23
Other examples include the studies of ‘structure’ in Noether’s, Mac Lane’s, or
Grothendieck’s mathematics [McLarty, 2006; 2007; 2008].24 A question that
arises in this context is whether to place such studies under the heading of His-
tory or Philosophy. The answer depends on the focus of the study. In general
historical studies are concerned with the understanding of a given notion, such
as ‘generality’, in a particular setting. That is, historians are after the particu-
lar, whereas philosophers are more concerned with whatever general lesson can
be drawn from particular examples — and if possible providing general features

23
Some of the concepts related to epistemology mentioned in the next section under
Epistemological PMP can also be taken in this sense. One could also mention the spe-
cial issue of Philosophia Mathematica (3) 23 No. 2, (2015) devoted to the notion of
‘mathematical depth’.
24
A volume edited by Erich Reck and Georg Schiemer [forthcoming] will uncover the
roots of contemporary structuralism. It contains chapters on a number of mathemati-
cians and philosophers from the nineteenth and early twentieth century (e.g., Dedekind,
Grassmann, Husserl, and Peirce) holding pre-structuralist or even structuralist views.
14 • Carter

of the notion in question. A historian could be interested in documenting how

a notion such as ‘simplicity’ or ‘generality’ is understood and perhaps used as
a guideline for research in a particular period, community, or even by a sin-
gle mathematician. The philosopher interested in capturing a particular notion

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

could occupy herself with how it is understood in different contexts, with the
intent of including possible understandings of it. It could also be remarked that
both philosophers and historians talk about ‘mathematical practice’ but that
they may have very different understandings of what ‘practice’ means in view
of the above mentioned differences between aims.

3.2.2. The Influence of Philosophy on Mathematics

Conversely to the above it is possible to consider the philosophical beliefs of
mathematicians themselves and even investigate how or whether these beliefs
influence their mathematics.25 Mathematicians at any given time have (at least
implicit) views on what mathematics is, which methods to use, which are the
interesting problems to solve, and so on. These may be merely views and not
fully fledged philosophical theories. In some cases, though, mathematicians
have developed philosophical positions on mathematics, and sometimes it can
even be argued that these have influenced their mathematics. Many studies
of the (philosophical) views of past mathematicians have been made. Exam-
ples include Bolzano [Rusnock, 2000], Hausdorff [Epple, 2006], Pasch [Schlimm,
2010], and Riemann (e.g., [Scholz, 1982; Bottazzin and Tazziolii, 1995]). It is
also possible to name a few contemporary mathematicians who have expressed
their views in public, such as Reuben Hersh and Alan Connes ([Changeux and
Connes, 1995; Connes et al., 2001]; see also the paper by Kanovei, Katz and
Mormann [Kanovei et al., 2013] for an interesting evaluation of Connes’ views).
As for the question of whether philosophical views influence the mathematics
produced, to my knowledge there have not been any systematic studies. There
are the obvious examples of constructivists (of various sorts) developing math-
ematics along constructivist guidelines, like Brouwer [van Atten, 2004], Weyl
[1918], and Bishop and Bridges [1985].

3.2.3. Philosophical Categories of Historical Studies

Along the lines of Lakatos’s second dogma, ‘the history of mathematics, lacking
the guidance of philosophy, has become blind’, there are also scholars developing
what could be denoted philosophical tools in order to understand the history of
mathematics. One such example is the adaption by Moritz Epple of categories
developed by Rheinberger [1997]. Epple [2004] employs these tools in order to
delve deeper into the history of the introduction of the first invariants of knots,
which were introduced in the 1920s independently by Alexander at Princeton
and Reidemeister in Vienna. A surface, or anachronistic, reading of the first
part of their two papers would make it seem like their work led to the same

25
There is a long list of well-known cases where philosophers are mathematicians (or
vice versa), as for example, Descartes, Leibniz, Newton, Husserl.
 Philosophy of Mathematical Practice • 15

techniques — a belief that has also previously been accepted. Both Alexan-
der and Reidemeister represent knots by a knot diagram.26 In both cases the
presentation of the knot gives rise to a matrix with integer entries and cer-
tain invariants in the form of integers. By instead using the tools of epistemic

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

configurations, techniques, and objects, Epple is able to show that their ideas
were indeed quite different. Epistemic configurations bear some resemblance to
Kuhn’s paradigms and are characterised as:

the entirety of the intellectual resources that are involved in a particular

episode. It comprises the mathematical language, the skills and techniques
at the disposal of the mathematician or the group of mathematicians
engaged in this research, the set of research topics and open problems
under consideration, the horizon of aims and more general heuristic guide-
lines followed by the researchers, etc. . . . An epistemic configuration of
mathematical research, together with the mathematician(s) working in
and with them, thus constitutes a (usually rather small) working unit for
the production of mathematical knowledge. [Epple, 2004, p. 148]

By taking us further into Alexander’s and Reidemeister’s articles Epple demon-

strates that they work within different epistemic configurations. They both use
Poincaré’s topology as a background theory. But whereas Alexander treats the
knots as Riemann spreads or Riemann surfaces, Reidemeister considers open
unbranched coverings of knots or link complements. That is, their objects of
investigation, referred to as the epistemic objects, are different. As they have
different epistemic objects they employ different techniques to arrive at the par-
ticular matrix: Reidemeister uses combinatorial algebra and Alexander linear
systems and homology.27 An epistemic configuration can be seen as a charac-
terisation of ‘a practice’. We shall therefore return to comment on this in the
next section when discussing the possibility of capturing a common notion of
‘practice’.
Summing up, in Historical PMP, ‘practice’ refers in particular to the various
ways that mathematics develops. In this strand we have seen examples of how
philosophy and history support each other. For example, philosophy contributes
with categories to understand history better [Epple, 2004] or philosophical
questions may be answered by referring to historical case studies. Historical

26
Their knot diagrams look similar except for different conventions adopted to show
which strand is on top when lines cross in the diagram.
27
These historiographical tools are also useful in describing how the research focus
may change; so, for example, what was previously considered to be an epistemic technique
becomes the object under study, i.e., the epistemic object. Tinne Kjeldsen [2009] demon-
strates that this was a decisive step in the development of the concept of a convex set by
Minkowski. She documents a shift of focus from the development of a geometric method to
solve certain problems in number theory to the study of the method itself, finally resulting
in the singling out of the concept of a convex body. The shift here is from the designing of
a method — the epistemic context — which later itself becomes the object of study, the
epistemic object.
16 • Carter

cases may, for example, help us understand the mathematical and philosophi-
cal background of the “crisis” that led to the formulations of the foundational
schools as well as the development of structuralist positions in the philosophy
of mathematics.

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

3.3. Epistemological PMP
For lack of a better term, I denote the last strand Epistemological PMP. The
qualifier ‘epistemological’ is intended to indicate that this strand is in some
respects close to traditional philosophy. It tries to avoid assumptions, e.g., that
talking about mathematics requires reference to human agents (although this is
not precluded). In addition it still wishes to answer traditional questions within
the philosophy of mathematics, such as what is the nature of mathematical
objects and what constitutes our knowledge of them. It does not, however, entail
a traditional way of viewing epistemology in the sense of “a view from nowhere”
and the possibility of obtaining absolute knowledge. Furthermore complement-
ing traditional philosophy of mathematics, philosophers in this strand ask for
an extension of topics that can be studied under the heading of philosophy of
mathematics (so another name could be ‘Extension-of-topics-studied’ PMP).
It is noted (e.g., by David Corfield [2003]) that when considering the actual
practice of doing mathematics there is a range of other questions that can be
asked and are being asked by contemporary mathematicians. In contrast to
the traditional methodology there is a stress on ‘mathematical practice’, and
case studies are involved when answering questions. ‘Practice’ in this strand is
probably best understood in a more abstract way than in the other two, since
in principle it could refer to anything that has to do with (real) mathematics.
What is important is the aim when taking this perspective, implying a shift
of ‘focus as to what is the subject under study, “mathematics” as it mani-
fests itself, not some idealised version or something based on our prejudices’
[Panza, 2003]. In epistemological PMP it is thus recognised that the aim of
investigation is to obtain a better understanding of the mathematics we have
access to. Characterising the phenomenological approach to the philosophy of
mathematics, Mary Leng writes:

The motivation for the so-called ‘phenomenological’ approach to the phi-

losophy of mathematics is thus the idea that philosophers should be
seriously concerned with ‘real’ mathematics and not some idealized notion
of the discipline. [Leng, 2002, p. 3]

— and so yet other headings capturing some contributions in this strand could
be ‘phenomenological PMP’ or ‘philosophy of real mathematics’.28
Taking this perspective does not from the outset exclude the possibility of
obtaining answers concerning the “true nature of mathematics”, although it

28
The philosophy of real mathematics is also the title of [Corfield, 2003].
 Philosophy of Mathematical Practice • 17

can be — and indeed has been — debated whether this is possible. Leng finds
that according to the phenomenological approach

if you want to understand what mathematics is you must first find out

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

what it is that mathematicians do. Conclusions may later be drawn about
ontology and epistemology, but that is a further step. [Leng, 2002, p. 5]

Based on a case study on the classification of C ∗ -algebras she does draw

conclusions about the epistemology and ontology of mathematics. To classify
C ∗ -algebras means to determine up to isomorphism the different types of alge-
bras that are possible to define. This is done by defining invariants, usually
“simpler” objects so that any two algebras are isomorphic if and only if their
corresponding invariants are isomorphic. The mathematician Georg Elliott has
initiated a program where the hope is that K-theory will contribute to obtaining
such invariants. As part of her case study Leng followed two research seminars
conducted by Elliott. One purpose of her study was to test Lakatos’s claim that
the formulation of a proposition is developed as a response to found counterex-
amples. The process observed in the seminars did not support this picture. In
these Elliott started by formulating and proving a weak version of the theorem
in question, gradually strengthening the formulation whenever it was possible to
find a proof of it. That is, whereas it still makes sense to say that the theorem is
proof generated, it did not develop by formulating a series of counterexamples.
Towards the end of her paper Leng argues that an anti-realist picture emerges
from the case study. She finds that the mathematicians do not care whether
the objects referred to exist or not when accepting a statement as true. What
matters is that it is possible to find a convincing proof. In contrast Brendan
Larvor [2001] claims that from the perspective of a ‘dialectical philosopher of
mathematics’ it is not possible to reach a conclusion on the ontology of math-
ematics. The ‘dialectical philosopher of mathematics’ is inspired by Lakatos’s
writings. Larvor comments that Lakatos’s most important contribution was to
turn our attention to the inner life of mathematics and in this context writes
that:

Whether we adopt fictionalism; or embrace a kind of emergentism in which

the mathematics produces itself out of the activities of the mathemati-
cians; or whether we think of progress as ever-closer approximation to a
pre-existing Platonic reality, makes no difference to our study of the inner
logic of mathematical development. The dialectical stories turn out the
same regardless of any ontological commitment. [Larvor, 2001, p. 218]

Others, like Oliveri [2010], find that the history of mathematics should be
brought in in order to say something about the philosophy of mathematics,
but stress that this is only justified if it contributes to the solution of philo-
sophical problems that remain unsolvable by previous methods. Finally there
are philosophers who find that the only way to do philosophy of mathemat-
ics is to start from mathematics itself, taking a bottom-up approach. See, for
18 • Carter

example, [Cellucci, 2012] for a critical attitude towards ‘top-down approaches’.

Given this lack of consensus, and especially the lack of convincing arguments for
either side, the question concerning the range of conclusions possible from, for
example, the use of case studies, is something that should be carefully assessed

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

from case to case.
In what follows a few more exemplary cases will be presented illustrating
the range of practices dealt with in this strand. For clarity they are divided
into epistemology, ontology and foundations.

3.3.1. Epistemology in the Epistemological PMP

The prime example of work asking questions related to the epistemology of
mathematics is [Mancosu, 2008]. In his introduction, Mancosu notes that a
number of different epistemological issues can be addressed besides the tradi-
tional question of how knowledge of acausal entities is possible (referring to
[Benacerraf, 1973]). Topics he proposes to study, orthogonal to those of tra-
ditional philosophy, include ‘fruitfulness, evidence, visualization, diagrammatic
reasoning, understanding, explanation’ [2008, p. 1]. Among these, the topics of
diagrammatic reasoning and the role of visualisation have received much atten-
tion. They serve as a good example for illustrating how a topic has been treated
from various perspectives. At first philosophers were mainly concerned with
questions related to whether reasoning using diagrams is rigorous. Referring to
Pasch and Hilbert, scholars claimed that figures should play no justificatory
role in proofs. More recently a number of scholars have defended and explained
their use, for example, Ken Manders [2008] and John Mumma [2010] in the
context of Euclidean geometry. Jon Barwise and John Etchemendy [1996] have
shown how a rigorous axiomatic system including diagrams can be set up. More
radically, in a recent article Jody Azzouni writes: ‘Talk of pictorial proofs lack-
ing “rigor” is false’ [2013, p. 325]. Azzouni bases his discussion on a pictorial
proof of the identity 12 + 14 + 18 + . . . + 21n + . . . = 1 and the (history of the)
Intermediate Value Theorem in analysis.
Besides a focus on proofs and rigour, attention has been drawn to the
many other roles visualisation plays in mathematics. Studies have been made
on past mathematicians’ views on the use of diagrams. For example, Ivahn
Smadja [2012] notes that Hilbert found that geometric intuition is impor-
tant in mathematics and that he praised Minkowski’s work in Geometry of
Numbers where connections between geometry and number theory are fruit-
fully exploited to develop both fields. Jeremy Heis tries to locate local reasons
for why diagrams were given up, not only in analysis, but even in geometry
in the nineteenth century. He finds that in projective geometry mathemati-
cians successfully employed the practice of exploiting ‘special features of the
drawn figure’ [Heis, 2012, Section III, p. 38], something that is often used as an
argument against diagram use. Furthermore he explains how the practice of rea-
soning using diagrams breaks down once one considers curves of degree three or
higher (the duality principle cannot be upheld), a reason that could be given for
the rejection of diagrams in further development of the field. Recent studies by
 Philosophy of Mathematical Practice • 19

Silvia De Toffoli and Valeria Giardino demonstrate how diagrammatic reason-

ing is a rigorous and integrated practice in contemporary mathematics, e.g.,
in low-dimensional topology [De Toffoli and Giardino, 2015]. They illustrate
how formal expressions in this area make little sense without accompanying

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

pictures or mental images. Some of their examples illustrate the construction
of 3-dimensional manifolds such as the torus. These diagrams show which sides
are supposed to be joined, or glued, in order to obtain the desired manifold.
Formally gluings can be expressed by certain equivalence relations. But the
equivalence relations do not contribute to the topological intuition needed to
work in the field. Their role is to provide a formal justification for the manip-
ulations done on diagrams. Further examples include a pictorial proof showing
certain constructions to be performed on different presentations on 3-manifolds.
Whereas the studies of De Toffoli and Giardino illustrate that diagrams play
an epistemic role, other scholars have pointed to the many other roles visu-
alisation plays in mathematics, for example, for understanding and heuristics
[Giaquinto, 2005b; Carter, 2010; Starikova, 2010].29 For up-to-date treatments
of this topic refer to [Giaquinto, 2015] and [Giardino, 2017].

3.3.2. Ontology in the Epistemological PMP

As regards questions related to the ontology of mathematics, the focus on math-
ematical practice dictates that answers to the traditional questions be based on
studies of actual mathematics rather than on philosophical arguments based on
assumptions. There have been some attempts at this, e.g. the above-mentioned
study on the development of C ∗ -algebra by Leng [2002] and myself (e.g. [Carter,
2004; 2013]). Here I will mention the recent study by Jeremy Avigad and
Rebecca Morris [2013; 2016] concerning the introduction of the concept of a
character in the proof of Dirichlet’s theorem on the number of prime numbers
in an arithmetical progression. More precisely the theorem states that if m and
k are relatively prime (i.e., have no common factors), then the arithmetical
progression m, m + k, m + 2k, . . . contains infinitely many primes. In addition
to directing attention to this nice and thorough work, the intention is to high-
light how concrete studies, such as this, of the actual practice of mathematics
may lead to new questions and concerns. Avigad and Morris find that although
characters play an important role in later proofs, there is no mention of them
in Dirichlet’s original proof from 1837. A character — in a modern treatment
of the theorem — is a function, or more precisely a group homomorphism, that
maps elements of an abelian group to the non-zero complex numbers. Avigad
and Morris investigate the development of the proofs leading to the modern
formulation and the gradual development of characters. In addition they place
the development of characters in a wider context as they regard this study

29
The concerns mentioned all belong to epistemology. Scholars have also focused on
ontological issues related to diagrams, for example, on the status of diagrams in Euclid
[Panza, 2012]. Another topic concerns the cognitive origin of Euclidean geometry; see
[Hamami and Mumma, 2013].
20 • Carter

as part of a larger quest to understand the development leading to the mod-

ern concept of a function. What makes me characterise this work as belonging
also to philosophy, is what I conceive of as their motivation for engaging in this
study, namely to understand why the concept of a character — or more broadly

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

a function — was introduced into mathematics. Furthermore the detailed case
study enables reflection on what it means for something to be counted as an
object in mathematics and the benefits achieved by adopting something as an
object rather than treating it via other types of representations.30 Avigad and
Morris formulate a list of operations and ways of treating them made possible
when considering characters as objects in themselves, such as formulating prop-
erties of characters, forming sets of characters and being able to sum over them
[2013, p. 21]. As for the benefits gained by considering characters as objects,
Avigad and Morris point to an increased conceptual understanding:

These innovations play an equally important role in fostering a better

understanding of Dirichlet’s proof itself, by highlighting key features of
the concepts and objects in question, motivating the steps of the proof,
and reducing cognitive burden on the reader by minimizing the amount
of information that needs to be kept in mind at each step along the way.
[ibid., p. 54]

3.3.3. Logic and Foundations in the Epistemological PMP

It has been compellingly argued31 that the motivation of many of the
researchers in the early foundational schools came from developments of actual
mathematics. Similarly, because of developments within mathematics, other
types of foundations have been proposed, e.g., within category theory (followed
by a fierce debate between set theorists and category theorists). Recently a new
program has been formulated, called ‘Univalent foundations’, which is based on
surprising links that have been found between Martin Löf type theory and the
mathematical discipline homotopy theory. Furthermore, links found between
type theory and computer proof assistants are exploited. The new combination
is called ‘Homotopy Type Theory’ (HoTT). In brief, Homotopy Type Theory
is an interpretation of type theory in homotopy theory, where types are inter-
preted as spaces and identity of types is expressible via the notion of homotopies
between paths. This bridge gives rise to the univalence axiom. One consequence
of this axiom is that it allows the identification of isomorphic structures or
objects, something that is often done by mathematicians, but does not follow
from the Zermelo-Fraenkel axioms for set theory (ZFC). For further details, see
the introduction of the HoTTbook [Univalent Foundations Program, 2013] or
the articles by James Ladyman and Stuart Presnell [2015; 2016].

30
For example, as expressions.
31
See, e.g., [Gray, 2008] for a historical account of the period from the nineteenth
century and into the early twentieth when mathematics developed into what is referred to
as ‘modern’ mathematics.
 Philosophy of Mathematical Practice • 21

As a second example of enquiries within logic and PMP one could mention
the work of Colin McLarty [2010] in determining the strength of the system
needed in order to prove Fermat’s last theorem (FLT). Whereas the aforemen-
tioned example provides a foundation in its traditional sense, McLarty’s results

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

have practical bearings for mathematicians, in the sense that they show that it
might be possible to find proofs of FLT within arithmetic. It is well known that
Andrew Wiles’s 1995 proof of FLT is very long and uses techniques that go
well beyond what one normally counts as belonging to arithmetic. McLarty has
pointed out that Wiles’s proof uses tools that go beyond even ZFC. One assump-
tion is the so-called axiom of Universes (U), formulated by Grothendieck. This
axiom is needed in order to provide a foundation for the operations performed,
and structures considered, in category theory.32 To found these operations a
universe is defined in ZFC33 and a category may then be regarded as an element
of some universe. Universes go beyond ZFC: Grothendieck has shown that U is
the set Vα for some strongly inaccessible ordinal α and since ZFC + U models
ZFC, the existence of U cannot be proved from ZFC. According to McLarty,
Grothendieck used universes in cohomology theory to show certain results in
number theory needed in order to prove the Weil conjectures. It has been proved
that universes are not needed for these proofs, but they make things “simpler”,
i.e., provide what McLarty calls ‘conceptual order’ [ibid, p. 361]. In the proof of
FLT Wiles refers to results that are obtained by using this machinery. McLarty
has shown that it should be possible to rework the proof into one that is only
based on finite-order arithmetic. There are even hopes that it might be proved
that first-order PA suffices, or even only fragments of PA. If such translations of
the proof were actually to be written down, they would be longer than the orig-
inal. The purpose of using the axiom of universes is to organise; so in addition,
such a proof would lack this structure.
One implication of these results for the practice of doing mathematics is
that one can tell that it is in principle possible to find a proof of FLT using
only arithmetic. It is not possible to tell, however, that only long proofs are
possible from PA. It has not been excluded that someone might find a short
arithmetical proof of FLT.
A final interesting study, related to the question of how strong a system is
needed in order to prove certain theorems, is the programme of reverse mathe-
matics designed by Stephen Simpson and Harvey Friedman (see e.g. [Simpson,

32
In category theory one considers categories, such as the category of Abelian groups
with group homomorphisms, a totality that is too big to be regarded as a set. In addition
constructions are performed on these and maps, the so-called functors, are defined between
categories, and functor categories are considered.
33
McLarty describes the universe as follows:

an uncountable transitive set U such that U, ∈ satisfies the ZFC axioms in the
nicest way: it contains the powerset of each of its elements, and for any function
from an element of U to U the range is also an element of U . [2010, p. 359]
22 • Carter

2009]). John Stillwell [2018] gives an accessible introduction to this programme.

Placing it in a broader context, Stillwell traces its roots to the discussions about
the status of the parallel postulate in Euclidean plane geometry. The main ques-
tion of reverse mathematics is: ‘Which set existence axioms are needed to prove

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

the theorems of ordinary, non-set-theoretic mathematics?’ [Simpson, 2009, p. 2].
The programme investigates which subsystems of second-order arithmetic are
sufficient to prove a number of specific theorems of mathematics. The results
considered are taken from many different branches of mathematics, i.e., logic,
analysis, and algebra. Subsystems are constructed by restricting the compre-
hension axiom schemes, that is, the axioms stating which sets exist in the
system. Simpson and Friedman establish that a given subsystem is the weakest
possible by reversing the normal proof procedure: in addition to proving that a
specific theorem follows from the axioms of a subsystem, they also prove that
the theorem implies the axioms of the subsystem (modulo the system in which
reasoning takes place). As an example, one could mention that it is possible
to prove the existence of an algebraic closure of any countable field from the
base system, named ‘recursive comprehension’. But in order to prove that this
closure is unique a stronger system is required. Further results from algebra
that require even stronger systems include structure theorems about abelian
groups such as ‘Any reduced Abelian p-group has an Ulm resolution’ which is
provably weaker than the more general statement ‘Any Abelian p-group has
an Ulm resolution’. This result follows from the theorem that ‘any group is a
direct sum of a divisible group and a reduced group’.34
It might already be clear from the above, that a wide range of practice,
that is cases, is considered under epistemological PMP: they go from consider-
ing examples from contemporary mathematics (C ∗ -algebra, proof of Fermat’s
theorem) to exploiting and even developing contemporary mathematics as is
done in homotopy type theory. Note also the different perspectives chosen in
the cases mentioned; whereas McLarty studies the presentation of the proof
of FLT, Leng is interested in how proofs are found in analysis. Furthermore
historical studies are done for various purposes, e.g., the studies mentioned on
Hilbert, projective geometry, and the development of characters. Finally there
are also studies of (historical) texts independent of historical concerns as for
example, the studies mentioned of Euclid’s Elements.

34
The system that is equivalent to the theorem ‘Any countable field has a unique
algebraic closure is ‘Weak König’s lemma’ (WKL0 ), which states ‘Any infinite subtree
of 2<N has an infinite path’ [Simpson, 2009, p. 36]. ‘Any reduced Abelian p-group has an
Ulm resolution’ is equivalent to ‘Arithmetical transfinite recursion’ whereas ‘Any group is a
direct sum of a divisible group and a reduced group’ requires the strongest existence axiom,
Π11 CA0 . The systems (in order of strength) are the ‘recursive comprehension axiom’, RCA0 ,
‘weak König’s lemma’, WKL0 , ‘Arithmetical comprehension’ ACA0 , ‘Transfinite Recur-
sion’ ATR0 , and finally Π11 CA0 denoting ‘Π11 -comprehension’. In analysis it is, for example,
possible to prove the intermediate value theorem in RCA0 [Simpson, 2009, pp. 77–91] and
the Bolzano-Weierstrass theorem is equivalent to ACA0 [ibid., p. 107].
 Philosophy of Mathematical Practice • 23

4. PRACTICE
We return to the question of whether it is possible to describe ‘PMP’ through
a characterisation of ‘practice’. In the light of the different interpretations of
‘practice’ one is tempted to say that the answer is no. When referring to practice

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

in social and pragmatic PMP, I stressed the practitioners and their activities. In
historical PMP, it was noted that the starting point is the product of the activi-
ties of mathematicians, but that the emphasis is on the relation between various
factors and activities and how they contribute to this product. In addition in
historical PMP the temporality of mathematics acts as a central underlying
assumption. Finally for epistemological PMP practice is understood in a more
abstract way as referring to a methodology in general that takes seriously ‘math-
ematical practice’ in a rather broad sense. In support of this observation, Van
Bendegem [2014] identifies eight different perspectives within the practice turn
of mathematics (as held by the mavericks, naturalists, sociologists, cognitive
scientists, in education, etc.) that he finds are in opposition to each other. In
this section, however, I shall suggest some unifying characterisations.
There are a few attempts in the literature to characterise ‘a practice’. Philip
Kitcher [1984] and later José Ferreirós [2016] have both described ‘practice’ by
drawing on the disciplinary matrices of Thomas Kuhn.35 As noted the char-
acterisation of an epistemic configuration by Moritz Epple could also be taken
as a candidate — although his motive for introducing it is different. Kitcher
in The Nature of Mathematical Knowledge [1984] breaks with what he refers
to as a traditional account of knowledge in a number of ways. First he rejects
an ‘apriorist’ account of mathematical knowledge. Second he wishes to take
into account that mathematical knowledge depends on a community of know-
ers and third, he bases his account on examples drawn from the history of
mathematics. Kitcher defends the position that to ‘understand the epistemo-
logical order of mathematics one must understand the historical order’ [ibid.,
p. 5]. The main case study that informs the description of knowledge concerns
the development of differential and integral calculus from around 1600 until its
rigorous treatment by, e.g., Cauchy and Weierstrass during the nineteenth cen-
tury. One of Kitcher’s points in this story is that the need for rigourisation of
mathematics came from mathematics itself and not from ‘apriorist epistemolog-
ical ideas’ [ibid., p. 246]. One major problem was that most theories depended
on an inconsistent handling of the infinitesimals. This problem was solved by
the introduction of limits by Cauchy and the real numbers by, for example,
Weierstrass during the nineteenth century.
Based on these concerns Kitcher defines a practice in terms of a quintuple,
L, M, S, Q, R. In the tuple L stands for language, M for meta-mathematical
views, S for the set of accepted statements, Q for important questions pursued
by the ‘practice’, and R for accepted reasoning. One aim of his project is to

35
See also [Van Bendegem and Van Kerkhove, 2004] for an extension of Kitcher’s pro-
posal. They introduce the ‘MathPract-structure’ consisting of a number of components
in order to characterise the practice of proving theorems.
24 • Carter

argue that changes in mathematics can be explained as transitions from one

practice to another by alterations in one or more of these components. Further-
more he argues that such transitions are ‘rational’ and as such contribute to
knowledge. It should be noted that his characterisation of a practice holds a

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

number of assumptions that renders it inappropriate for a general definition.36
The first assumption is that it is a practice without practitioners. Kitcher refers
to ‘practices of idealised agents’ and so (real) agents are left out. This in part
opposes the view of agent-based PMP. In addition Kitcher wishes to argue for
an anti-realist view of mathematics and so he excludes mathematical objects
from the tuple. Overall the inclusion of all five components renders the charac-
terisation too fine grained to pass as a general characterisation of ‘a practice’.
In addition, such level of detail would make one wonder why some categories
are included whereas others, for example ‘objects’, are left out.
Ferreirós [2016], discussing various views on ‘practice’, stresses, in opposition
to Kitcher, that practice should be taken in the sense of ‘the activities of human
agents’ and that knowledge does not make sense without the knowers, i.e., the
human agents (p. 21). In order to analyse knowledge he thus considers the
pair, framework-agent, where a framework is taken to include something like
Kitcher’s four-tuple, L, S, Q, R. ‘M’, meta-mathematical views, are excluded
from this list. Ferreirós instead places them with the agent. A further important
qualification is that Ferreirós holds that it does not make sense to claim that
practices are all-encompassing and unique at any given time. He bases his
account on the fact that ‘several different levels of practices and knowledge are
coexistent’ (p. 25) and that their links and interplay are crucial to mathematical
knowledge. In support of these claims Ferreirós draws on a number of historical
case studies from the theories of numbers, in particular the real numbers, set
theory, and geometry. One major development considered is the transformation
of geometry in the hands of Euclid to the theory of the real numbers and the
further development leading to contemporary set theory.
As in Ferreirós’s framework-agent pair, an epistemic configuration as devel-
oped by Epple consists in part of the human agents, in this case the mathe-
maticians developing the results under investigation. In addition it includes the
mathematical objects, the epistemic objects, which are not explicitly included
in the other two practices.
Variations of such characterisations are evidently of value when investigat-
ing issues related to the activities of doing mathematics as is manifested by
the works of Kitcher, Epple, and Ferreirós. All three include certain assump-
tions not held by all the strands; so as they stand they are insufficient as a
general description of ‘a practice’. But if one makes a generalisation of these, a
rough characterisation of ‘a practice’ emerges (one quite similar to Ferreirós’s
framework-agent): In general a practice is captured by a tuple consisting of
‘agents’ and ‘mathematics’, which can be written as A, M . The component
‘agents’ consists of agents, real or idealised, and their activities and beliefs.

36
See also the discussion of Kitcher’s contribution in [Ferreirós, 2016, Chs 2, 3].
 Philosophy of Mathematical Practice • 25

‘Mathematics’ refers to the content of mathematics — theories, theorems,

proofs, mathematical models, etc. In general ‘a practice’ consists of relations
between agents and what constitutes mathematics. This gives a simple picture
of the field characterising some overall structures. In order to capture the rich-

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

ness of each strand this picture should be supplemented with the descriptions
given in the previous sections. Agent-based PMP obviously has as a starting
point the agent(s) and regards mathematics only in relation to those agents.
Historical PMP starts at the level of mathematics and considers the develop-
ment of a particular part of mathematics through an interrelation between the
agents and mathematics. Finally I understand epistemological PMP as start-
ing somewhere in the space set up between the two, with most emphasis on
the mathematical case studies (as is seen from the examples given), but with
the intent of transcending the mathematical examples and arriving at gener-
alisations, or perhaps even idealisations, based on the considerations of actual
practice. In this way philosophical categories, concepts, or even a position may
be formed, so that one obtains a relation between philosophy and the practice
tuple in the sense that the philosophical considerations are based on mathemat-
ical practice.37 I propose also to understand agent-based PMP in the sense that
the intended outcome of the investigation is a philosophical position based on
the assumption of agents doing mathematics. Similarly historical PMP, as we
have seen, formulates philosophical positions held by mathematicians, creates
philosophical categories to understand historical cases, tries to understand why
certain philosophical questions were asked, and investigates how to understand
certain concepts related to mathematics. I therefore propose that PMP intends
a relation between philosophy and practice tuple consisting of mathematics and
agents.
A complement to the abstract characterisation of practice can be found in
the preface to a collection of articles assembled under the title From practice
to results in logic and mathematics [Giardino et al., 2012]. Léna Soler presents
different characterisations of ‘practice’ as follows:

(a) Practices as processes, dynamical actions and procedures, contrasted

with the propositional products of these practices (theories, experimental
facts, results of a mathematical theorem, etc.); (b) Ongoing day-to-day,
real time actual science (science as it is really practiced) contrasted with
science as it is a posteriori reconstructed by practitioners, notably in
their publications; (c) Real science opposed to idealized accounts provided
by philosophers — most of the time in a derogatory sense of the term
‘idealized’, that is in the sense implying the allegation to have produced

37
This could be written as P : A, M . One of the reviewers proposed an extension
of this notation: one could write AM , M  stressing that the focus is on mathematical
agents doing mathematics; or AP , M  for philosophical agents studying mathematics.
Philosophers of PMP can then be described as a nested couple: AP , AM , M . In this
way metamathematics could be a special case, namely where instead of a philosophical
agent, a mathematical agent is studying mathematical practices, i.e., AM , AM , M .
26 • Carter

a one-sided and truncated, if not a completely deceitful account of what

practitioners indeed do and what the science under scrutiny really is . . . .
[Giardino et al., 2012, Preface, p. 2, emphasis original].

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Elaborating on (a) one could note that the development of mathematics can
be examined at three levels: macro, meso, and micro [Giardino et al., 2012,
Introduction, pp. 7–8]. According to my understanding of these levels, the
macro level asks questions regarding what can be said about the development
of mathematics generally. At the meso level questions regarding research styles,
schools, and in general questions relating particular groups can be asked. Finally
at the micro level, one may investigate the development of particular con-
cepts, such as the notion of proof at particular times, validity of diagrammatic
reasoning, and such like.
Items (a) and (b), focussing on processes or day-to-day science in contrast
to the products of these processes, fit well with many of the examples pre-
sented in the previous section. For example, Epple’s [2004] study went beyond
the presented results of Reidemeister and Alexander, looking instead at their
epistemic objects and methods. Leng [2002] participated in research seminars
in order to see how theorems are developed. Maddy considered the arguments
of set theorists about which principles of set theory to adopt. Kitcher’s [1984]
and Ferreirós’s [2016] emphasis on the practices of agents for the development
of mathematical knowledge is yet another example. In order to accommodate
the mentioning of the human element in the last case one could propose a mod-
ification of (a) stressing that the focus is not only on processes but on human
beings:

(d) Focus on the human agents doing mathematics and the implications
that this has for philosophy in opposition to regarding mathematics as
independent of human activities and mathematical knowledge without
knowers.

Item (c) resembles the concerns of the Epistemological PMP. All three
strands, however, seem to oppose idealised versions of mathematics as a basis
for the philosophy of mathematics. Besides (c) above, and in the light of the
common motivation of all strands, I propose the following to capture some of
the common threads of PMP:
(i) Practice — topics: A demand for an extension of questions asked in the
philosophy of mathematics; questions ought to be asked from all ranges of
mathematical practice and not simply concern ontology and foundations, and
(ii) Practice — methods: The methodologies brought in to study and answer
these questions should likewise be extended, as long as they rationally con-
tribute to the questions asked. In addition, when it makes sense, results from
other sciences, such as history, cognitive science, and sociology, may be used.
The proposal is for philosophy of mathematics to mean philosophy of math-
ematics, where mathematics is taken in a broad sense. This approach opens up
a number of different perspectives as we have seen (dealing with the product,
 Philosophy of Mathematical Practice • 27

that is, the theories presented in any preferred way, the activities of producing,
applying, or learning the product as well as the role of the practitioners in all
this). Underlying this proposal is the role of philosophy as our way of asking
the most fundamental questions and allowing for these questions to concern all

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

aspects of mathematics.
From this perspective the philosophy of mathematical practice could be
regarded as a proper subset of the philosophy of mathematics stressing
‘practice’.38 This would mean two things:

(1) ‘Mathematics’ in the philosophy of mathematics is to be understood not

only in an ideal sense. It should also be possible to deal with questions
related to mathematics at different levels and different types of activities.
(2) Extension of methodologies brought in to deal with these questions. It
is here the real challenge lies, namely to uphold philosophical standards
(of reasoning). A careful scrutiny of which conclusions can be made from,
e.g., empirical studies and to which “level” of investigation they apply is
required.

5. CONCLUSION
In conclusion it has been proposed that the PMP starts when philosophical
investigations are based on studies or considerations of mathematical practice.
Taking a ‘bottom-up’ approach, starting with the many contributions within
this field, three different strands have been characterised, the agent-based, the
historical, and the epistemological PMP. They differ in various respects, in
particular on how ‘practice’ is understood. It has been noted that practice
can be characterised in terms of (i) focussing on the practitioners, the human
beings engaging in mathematics, (ii) as the mathematics itself conceived of as
the product of certain events that can be understood historically, or finally (iii)
as simply any relevant aspect of mathematics brought in to investigate whatever
question has been proposed to study. In order to state in general what PMP is, a
general framework is proposed, that is a relation that holds between philosophy
and a general practice consisting of agents and mathematical content. The idea
is to capture what is taken to count as ‘practice’ in each strand by varying
the stress on the components of this tuple. Some versions of PMP emphasise
agents whereas versions of the epistemological PMP focus primarily on the
mathematics. Furthermore when referring to the philosophy of mathematical
practice, it is proposed that the output of studies of such practices belongs to
philosophy.

38
I assume that whether one accepts that PMP is a subset of the philosophy of math-
ematics depends on how inclusive (or narrow) one takes the philosophy of mathematics in
particular and philosophy in general to be. As I hope to have conveyed here, PMP wishes
to be inclusive and open-minded and so by making the above statement I transfer this view
to philosophy in general. If this is unacceptable one may regard PMP as a complement to
the traditional philosophy (of mathematics).
28 • Carter

In general it is found that PMP raises a number of new questions that can
(fruitfully) be treated within philosophy of mathematics. In the text a number
of such questions and studies of these have been presented. It should be noted
that PMP means a shift in focus as to what is the subject under study, not

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

idealisations of mathematics, but mathematics in the multiple ways it presents
itself to us in our various engagements with it. Furthermore acknowledging that
the aim of investigation is to obtain a better understanding of mathematics that
we practice — and that there is not one true picture of mathematics.

REFERENCES
Antonelli, A. [2001]: ‘Introduction’ [to issue ‘Mathematical philosophy’], Topoi 20,
1–3.
Aspray, W., and P. Kitcher, eds [1988]: History and Philosophy of Modern Mathemat-
ics. Minnesota Studies in the Philosophy of Science; IX. Minneapolis: University
of Minnesota Press.
Avigad, J., and R. Morris [2015]: ‘The concept of “character” in Dirichlet’s
theorem on primes in an arithmetic progression’. Available at MathArXiv,
http://arxiv.org/pdf/1209.3657v3.pdf. Accessed December 2018.
[2016]: ‘Character and object’, Review of Symbolic Logic 9, 480–510.
Azzouni, J. [2013]: ‘That we see that some diagrams are perfectly rigorous’,
Philosophia Mathematica (3) 21, 323–338.
Barwise, J., and J. Etchemendy [1996]: ‘Visual information and valid reasoning’, in
G. Allwein and J. Barwise, eds, Logical Reasoning with Diagrams: Studies in Logic
and Computation, pp. 3–25. Oxford University Press.
Benacerraf, P. [1965]: ‘What numbers could not be’, Philosophical Review 74, 47–73.
[1973]: ‘Mathematical truth’, Journal of Philosophy 70, 661–679.
Benis Sinaceur, H. [2006]: ‘From Kant to Hilbert: French philosophy of concepts in
the beginning of the twentieth century’, in [Ferreirós and Gray, 2006], pp. 311–337.
Oxford University Press.
Bishop, E., and D. Bridges [1985]: Constructive Analysis. Springer.
Bloor, D. [1976]: Knowledge and Social Imagery. London: Routledge & Kegan Paul.
Bottazzini, U., and R. Tazzioli [1995]: ‘Naturphilosophie and its role in Riemann’s
mathematics’, Revue d’histoire des mathématiques 1, 3–38.
Breger, H., and E. Grosholz, eds [2000]: The Growth of Mathematical Knowledge.
Dordrecht: Kluwer.
Carey, S. [2004]: ‘Bootstrapping & the origin of concepts’, Daedalus 133, 59–68.
Carter J. [2004]: ‘Ontology and mathematical practice’, Philosophia Mathematica (3)
12, 244–267.
[2010]: ‘Diagrams in analysis’, International Studies in the Philosophy of
Science 24, 1–14.
[2013]: ‘Handling mathematical objects: Representations and context’, Syn-
these 190, 3983–3999.
[2014]: ‘Mathematics dealing with “Hypothetical states of things” ’,
Philosophia Mathematica (3) 22, 209–230.
Cellucci, C. [2012]: ‘Top-down and bottom-up philosophy of mathematics’, Found.
Sci. 18, 93–106.
Changeux, J., and A. Connes [1995]: Conversations on Mind, Matter, and Mathe-
matics. Princeton University Press.
 Philosophy of Mathematical Practice • 29

Chemla, K., R. Chorley, and D. Rabouin, eds [2016]: The Oxford Handbook on
Generality in Mathematics and the Sciences. Oxford University Press.
Connes, A., A. Lichnerowicz, and M.P. Schutzenberger [2001]: Triangle of Thoughts.
Jennifer Gage, trans. Providence, R.I.: American Mathematical Society.

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Corfield, D. [2003]: Towards a Philosophy of Real Mathematics. Cambridge University
Press.
De Cruz, H. [2009]: ‘An enhanced argument for innate elementary geometric knowl-
edge and its philosophical implications’, in [Van Kerkhove, 2009], pp. 185–206.
De Toffoli, S., and V. Giardino [2015]: ‘An inquiry into the practice of proving in
low-dimensional topology’, in G. Lolli, M. Panza, and G. Venturi, eds, From Logic
to Practice, pp. 315–336. Boston Studies in the Philosophy and History of Science;
308. Springer.
Doridot, F., and M. Panza [2004]: ‘À propos des apports des sciences cognitives à la
philosophie des mathématiques’, Intellectia 39, 263–287.
Epple, M. [2004]: ‘Knot invariants in Vienna and Princeton during the 1920’ies.
Epistemic configurations of mathematical research’, Science in Context 17,
131–164.
[2006]: ‘Felix Hausdorff’s considered empiricism’, in [Ferreirós and Gray, 2006],
pp. 263–289.
Ernest, P. [1996]: ‘The nature of mathematics and teaching’, Philosophy of Mathemat-
ics Education Newsletter 9. Available at http://people.exeter.ac.uk/PErnest/pome/
pompart7. Accessed December 2018.
[1998]: Social Constructivism as a Philosophy of Mathematics. Albany, New
York: State University of New York Press.
Ferreirós, J. [2006]: ‘Riemann’s Habilitationsvortrag at the crossroads of mathematics,
physics and philosophy’, in [Ferreirós and Gray, 2006], pp. 73–105.
[2010]: ‘In or about mathematics: Concerning some versions of mathematical
naturalism’, in [Van Kerkhove et al., 2010], pp. 117–154.
[2011]: ‘On arbitrary sets and ZFC’. The Bulletin of Symbolic Logic 17, 361–
393.
[2016]: Mathematical Knowledge and the Interplay of Practices. Princeton
University Press.
Ferreirós, J., and J. Gray, eds [2006]: Architecture of Modern Mathematics: Essays in
History and Philosophy. Oxford University Press.
François, K., and J.P. Van Bendegem, eds [2007]: Philosophical Dimensions in
Mathematics Education. Springer Science and Business Media.
Giaquinto, M. [2005a]: ‘Mathematical activity’, in [Mancosu et al., 2005], pp. 75–87.
[2005b]: ‘From symmetry perception to basic geometry’, in [Mancosu et al.,
2005], pp. 31–55.
[2007]: Visual Thinking in Mathematics. Oxford University Press.
[2015]: ‘The epistemology of visual thinking in mathematics’, The Stanford
Encyclopedia of Philosophy (Winter 2015 Edition), Edward N. Zalta, ed. Avail-
able at http://plato.stanford.edu/archives/win2015/entries/epistemology-visual-
thinking/. Accessed Dec. 2018.
Giardino, V. [2017]: ‘Diagrammatic reasoning in mathematics’, in L. Magnani and
T. Bertolotti, eds, Springer Handbook on Model-Based Science, pp. 499–522.
Springer.
Giardino, V., A. Moktefi, S. Mols, and J.P. Van Bendegem [2012]: ‘From practice to
results in logic and mathematics’, issue of Philosophia Scientiae 16, 1–210.
Gray, J. [2008]: Plato’s Ghost. Princeton University Press.
30 • Carter

Hamami, Y., and J. Mumma [2013]: ‘Prolegomena for a cognitive investigation of

Euclidean diagrammatic reasoning’. Journal of Logic, Language and Information
22, 421-448.
Heintz, B. [2000]: Die Innenwelt der Mathematik. Zur Kultur und Praxis einer

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

beweisenden Disziplin. Springer.
Heis, J. [2012]: ‘Why did geometers stop using diagrams’. Available at
https://vdocuments.mx/why-did-geometers-stop-using-diagrams.html. Accessed
December 2018.
Hersh, R. [1979]: ‘Some proposals for reviving the philosophy of mathematics’,
Advances in Mathematics 31, 31–50.
[2013]: Experiencing Mathematics: What Do We Do, When We Do Mathe-
matics? Providence, R.I.: American Mathematical Soc.
Hilbert, D. [1992]: Natur und mathematisches Erkennen: Vorlesungen gehalten 1919–
1920 in Göttingen. David Rowe, ed. Basel: Birkhäuser Verlag.
Inglis, M., and A. Aberdein [2015]: ‘Beauty is not simplicity: An analysis of
mathematicians’ proof appraisals’. Philosophia Mathematica (3) 23, 87–109.
Kanovei, V., M. Katz, and T. Mormann [2013]: ‘Tools, objects and chimeras: Connes
on the role of hyperreals in mathematics’, Found. Sci. 18, 259–296.
Katz, J. [1998]: Realistic Rationalism. Cambridge, Mass: MIT Press.
Kitcher, P. [1984]: The Growth of Mathematical Knowledge. Oxford University Press.
Kjeldsen, T.H. [2009]: ‘Egg-forms and measure bodies: Different mathematical prac-
tices in the early history of the development of the modern theory of convexity’,
Science in Context 22, 85–113.
Kjeldsen, T.H., and J. Carter [2015]: ‘The role of history and philosophy in university
mathematics education’, in M. Matthews, ed., International Handbook of Research
in History, Philosophy and Science Teaching, Vol. 2, pp. 837–871. Springer.
Ladyman, J., and S. Presnell [2015]: ‘Identity in homotopy type theory, Part I: The
justification of path induction’. Philosophia Mathematica (3) 23, 386-406.
[2016]: ‘Does homotopy type theory provide a foundation for mathemat-
ics?’ The British Journal for the Philosophy of Science. Available at doi:
10.1093/bjps/axw006. Accessed Dec. 2018.
Lakatos, I. [1976a]: Proofs and Refutations: The Logic of Mathematical Discovery.
Cambridge University Press.
[1976b]: ‘A renaissance of empiricism in the recent philosophy of mathematics’,
The British Journal for the Philosophy of Science 27, 201–223.
Larvor, B. [2001]: ‘What is dialectical philosophy of mathematics’, Philosophia
Mathematica (3) 9, 212–229.
Lautman, A. [2011]: Mathematics, Ideas and the Physical Real. Simon B. Duffy, trans.
Bloomsbury Continuum.
Leng, M. [2002]: ‘Phenomenology and mathematical practice’, Philosophia Mathe-
matica (3) 10, 3–14.
Lerman, S. [1990]: ‘Alternative perspectives of the nature of mathematics and their
influence on the teaching of mathematics’, British Educational Research Journal
16, 53–61.
Löwe, B., T. Müller, and E. Müller-Hill [2010]: ‘Mathematical knowledge as a case
study in empirical philosophy of mathematics’, in [Van Kerkhove et al., 2010],
pp. 185–203.
Maddy, P. [1992]: ‘Indispensability and practice’, Journal of Philosophy 89, 275–289.
[1997]: Naturalism in Mathematics. Oxford University Press.
 Philosophy of Mathematical Practice • 31

Mancosu, P., ed [2008]: The Philosophy of Mathematical Practice. Oxford University

Press.
Mancosu, P., K.F. Jørgensen, and S.A. Pedersen, eds [2005]: Visualisation, Explana-
tion and Reasoning Styles in Mathematics. Synthese Library; 327. Springer.

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Manders, K. [2008]: ‘The Euclidean diagram (1995)’, in [Mancosu, 2008], pp. 80–133.
Marquis, J.-P. [1995]: ‘Category theory and foundations of mathematics: Philosophical
excavations’, Synthese 103, 421–447.
[2015]: ‘Mathematical abstraction, conceptual variation and identity’, in
P. Schroeder-Heister, G. Heinzmann, W. Hodges, and P.E. Bour, eds, Logic,
Methodology and Philosophy of Science, pp. 229–321. London: College Publica-
tions.
McLarty, C. [2006]: ‘Emmy Noether’s “set theoretic” topology: From Dedekind to the
rise of functors’, in [Ferreirós and Gray, 2006], pp. 187–208.
[2007]: ‘The last mathematician from Hilbert’s Göttingen: Saunders Mac Lane
as a philosopher of mathematics’, British Journal for the Philosophy of Science
58, 77–112.
[2008]: ‘ “There is no ontology here”: Visual and structural geometry in today’s
arithmetic’, in [Mancosu, 2008], pp. 354–406.
[2010]: ‘What does it take to prove Fermat’s last theorem? Grothendieck and
the logic of number theory’, The Bulletin of Symbolic Logic 16, 359–377.
Mumma, J. [2010]: ‘Proofs, pictures and Euclid’, Synthese 175, 255–287.
Oliveri, G. [2010]: ‘For a philosophy of mathematical practice’, in [Van Kerkhove et
al., 2010], pp. 89–116.
Panza, M. [2003]: ‘Mathematical proofs’, Synthese 134, 119–158.
[2012]: ‘The twofold role of diagrams in Euclid’s plane geometry’, Synthese
186, 55–102.
[2015]: ‘From Lagrange to Frege: Functions and expressions’, in H. Benis-
Sinaceur, M. Panza, and G. Sandu. Functions and Generality of Logic: Reflections
on Dedekind’s and Frege’s Logicisms, pp. 59–95. Springer.
Panza, M., and A. Sereni [2013]: Plato’s Problem: An Introduction to Mathematical
Platonism. Basingstoke: Palgrave Macmillan.
Prediger, S. [2007]: ‘Philosophical reflections in mathematics classrooms’, in [François
and Van Bendegem, 2007], pp. 43–59.
Putnam, H. [1967]: ‘Mathematics without foundations’. The Journal of Philosophy
64, 5–22.
Reck, E., and G. Schiemer, eds. [forthcoming]: The Pre-History of Mathematical
Structuralism. Oxford Uiversity Press.
Restivo, S. [1993]: ‘The social life of mathematics’, in S. Restivo, J.P. Van Bendegem,
and R. Fischer, eds, Math Worlds, pp. 247–278. Albany: State University of New
York Press.
Rheinberger, H.-J. [1997]: Toward a History of Epistemic Things: Synthesizing
Proteins in the Test Tube. Stanford University Press.
Rusnock, P. [2000]: Bolzano’s Philosophy and the Emergence of Modern Mathematics.
Amsterdam: Rodopi.
Schlimm, D. [2010]: ‘Pasch’s philosophy of mathematics’, Review of Symbolic Logic
3, 93–118.
Scholz, E. [1982]: ‘Herbart’s influence on Riemann’, Historia Mathematica 9, 413–440.
Simpson, S. [2009]: Subsystems of Second-Order Arithmetic. Cambridge University
Press.
32 • Carter

Smadja, I. [2012]: ‘Local axioms in disguise: Hilbert on Minkowski diagrams’, Synthese

186, 315–370.
Starikova, I. [2010]: ‘Why do mathematicians need different ways of presenting
mathematical objects? The case of Cayley Graphs’. Topoi 29, 41–51.

Downloaded from https://academic.oup.com/philmat/article-abstract/27/1/1/5316111 by Goldsmiths College user on 16 February 2019

Stillwell, J. [2018]: Reverse Mathematics: Proofs from the Inside Out. Princeton
University Press.
Tappenden, J. [2006]: ‘The Riemannian background to Frege’s philosophy’, in
[Ferreirós and Gray, 2006], pp. 97–132.
[2008]: ‘Mathematical concepts: Fruitfulness and naturalness’, in [Mancosu,
2008], pp. 276–301.
Tymoczko, T., ed. [1998]: New Directions in the Philosophy of Mathematics. Revised
and extended version. Princeton University Press.
The Univalent Foundations Program [2013]: Homotopy type theory: Univa-
lent foundations of mathematics. Princeton: Institute for Advanced Study.
http://homotopytypetheory.org/book. Accessed Dec. 2018.
van Atten, M. [2004]: On Brouwer. Belmont, Calif.: Wadsworth.
Van Bendegem, J.P. [2014]: ‘The impact of the philosophy of mathematical practice
on the philosophy of mathematics’, in L. Soler, S. Zwart, M. Lynch, and V. Israel-
Jost, eds., Science After the Practice Turn in the Philosophy, History, and Social
Studies of Science, pp. 215–226. Routledge Studies in the Philosophy of Science;
14. Routledge.
Van Bendegem, J.P., and B. Van Kerkhove [2004]: ‘The unreasonable richness of
mathematics’, Journal of Cognition and Culture 4, 525–549.
Van Kerkhove, B., ed. [2009]: New Perspectives in Mathematical Practices: Essays in
Philosophy and History of Mathematics. Singapore: World Scientific Publishing.
Van Kerkhove, B., J. De Vuyst, J.P. Van Bendegem, eds [2010]: Philosophical Per-
spectives on Mathematical Practice. Texts in Philosophy; 12. London: College
Publications.
Van Kerkhove, B., and J.P. Van Bendegem, eds. [2007]: Perspectives on Mathe-
matical Practices: Bringing Together Philosophy of Mathematics, Sociology of
Mathematics, and Mathematics Education. Springer.
Weyl, H. [1918]: Das Kontinuum. Leipzig: Veit.
Wilhelmus, E. [2007]: Formalizability and Knowledge Ascriptions in Mathemati-
cal Practice. ILLC research report and technical notes series: Logic, philosophy
and linguistics series. University of Amsterdam Institute for Logic, Language and
Computation.