CRITICAL STUDIES/BOOK REVIEWS 95

importance of Netz’s analysis but rather calls for more research. When
stylistic features of other periods of mathematical creativity, such as the
Renaissance or the seventeenth century, are better understood, it will be
possible to answer the question: to what degree have the aspects identified
by Netz to do with style and to what degree are they more general features
of mathematical creativity?
A further problem is Netz’s terminology. He likes rich, expressive

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

labels. It seems that a neutral technical terminology would be more helpful.
The question put forward in his book, the question about style in mathe-
matics, is important, and to try to develop a kind of poetics of mathematics
is a fascinating idea. But for this we would need a neutral terminology,
which would not distort the observation of facts, rather than a picturesque
one that catches attention and leads our imagination.
The book as a whole is a stimulating and rich text. In my Patterns of
Change, Linguistic Innovations in the Development of Classical Mathe-
matics (Birkhäuser Verlag, 2008), I analyzed linguistic innovations in the
development of mathematics, and the consideration of style gives a totally
new dimension to these analyses. Comparison of mathematics with poetry
is new territory, which can bring new insights and enrich the more stan-
dard areas of comparisons of mathematics with the arts. The book contains
many totally new and original ideas and has a great potential to stimu-
late further research on both sides of the divide separating science and
humanities.

doi: 10.1093/phimat/nkq025

XAVIER SABATIER. Les formes du réalisme mathématique. Paris: Vrin,

2009. ISBN 978-2-7116-2193-4. Pp. 304.
Reviewed by ANDRÉ LEBEL∗

I
The book under review discusses the most standard forms of mathemati-
cal realism, a.k.a. ‘Platonism’, elaborated in the mainstream of twentieth-
century philosophy of mathematics. Its purpose is therefore to introduce
franco phone readers to an important set of ideas that have been, for the
most part, cultivated by analytic philosophers. It originates from a doc-
toral dissertation at Université Paris 1. It comprises 287 pages of text

∗ Department of Mathematics, Cegep Champlain - St. Lawrence, Québec, Québec G1V

4K2, Canada. [email protected]
96 PHILOSOPHIA MATHEMATICA

divided into four main parts, completed by an eight-page bibliography.

Beginning with Frege, Russell, and Gödel, Xavier Sabatier discusses the
work of several contemporary Platonists, such as Quine, Resnik, Shapiro,
and Maddy. Since the last no longer considers herself a Platonist in math-
ematics (p. 219), the author restricts his commentary to her Realism in
Mathematics (published in 1990).
The introductory first part ends with a discussion of the standard objec-

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

tions to realism. The fourth part of the book, which is also its conclusion,
revisits those objections and shows how, in the light of what has been said,
they can be overcome to some extent from within a naturalist stance. Fol-
lowing the first part, one finds two long sections entitled ‘A priori realism’
and ‘Naturalized realism’. The first focuses on the Frege-Russell brand of
logical Platonism, prefigured in the work of Bolzano, and on Gödel’s ver-
sion of realism. The second examines the renewal of realism emanating
from Quine’s project of a naturalized epistemology. It is in this section
that Sabatier considers some contemporary authors like Penelope Maddy,
Michael Resnik, and Stewart Shapiro. Some might find it questionable to
associate the structuralism of Resnik and Shapiro with Platonism, given
the naturalistic stance of those philosophers (at least in Resnik’s case) but
it is nevertheless an excellent idea to discuss their work in such a context.
Indeed, one can give interpretations of Plato’s Sophist and of other late dia-
logues that are somewhat akin to mathematical structuralism.1 The author,
though, has chosen not to discuss historical Platonism or even varieties
of mathematical Platonism that do not fit well into the analytic tradition,
such as one can find in the work of the philosopher Albert Lautman, or
in the realist reconstruction of Husserl’s philosophy of mathematics given
by G. Rosado Haddock. Moreover, he does not discuss in any detail the
writings of Platonist mathematicians, such as René Thom, Roger Penrose,
or Alain Connes. Therefore the book remains well within the confines of
academic philosophy.
Sabatier opens the first part of his essay with a definition of realism
or Platonism that involves two components. The first one, the ‘semantic’
component, entails the truth of mathematical propositions and their non-
conventional nature. Typical of this thesis is the unrestricted application of
tertium non datur. The second, ‘ontological realism’, states that the truth of
mathematical propositions stems from the fact that they correctly describe
properties of objects or state of affairs that are independent in one way
or another from human subjectivity. A crucial example is given by the
axiom of infinity. The usefulness of this definition of Platonism is then put
to the test and confirmed by a quick review of the relevant philosophical

1 See the remarkable book by Verity Harte, Plato on Parts and Wholes: The Metaphysics
of Structure. Oxford: Clarendon Press, 2002.
 CRITICAL STUDIES/BOOK REVIEWS 97

literature and by a glance at mathematical practice. For example, stan-

dard mathematical methods such as proofs by contradiction, set-theoretic
extensional definitions of functions with infinite domains, and applications
of Zorn’s lemma, all seem to require some sort of realist assumption or
intuition. The positions of the various schools of mathematical realism are
understood here as specific blends of semantic and ontological theses.
Toward the end of the first part, the author introduces two basic ideas

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

that play a very important role in the rest of the book. The first, ‘Logi-
cal universalism’ (universalisme logique), is the doctrine that there is a
single logic, a single set of logical principles that applies to the whole of
knowledge, both mathematical and empirical. It has been extracted from
the works of Frege and Russell by authors like van Heijenoort, de Rouil-
han, and Rivenc. It is clearly consonant with the contemporary concern
with homogeneity of semantics. Adopted in a way by Quine (p. 211), it is
rejected by contemporary mathematical structuralism (p. 265). The second
fundamental idea finds its roots in the acknowledgement of the difficulty
of distinguishing in a physical theory what is purely empirical from what
belongs properly to mathematics. Both ideas are made more precise when
the author discusses the now classical Benacerraf’s problems and Quine’s
indispensability argument. For Sabatier, Platonism seems to entail a defi-
nite homogeneity and unity throughout the various branches of scientific
knowledge, and it is those very aspects that render problematic the inclu-
sion of structuralism into Platonism.
The author concludes his book by looking at the standard objections to
Platonism in the light of the ideas considered in the two central sections. It
is, indeed, by studying the solutions given to those objections by various
schools of realism that one can most conveniently work out in what essen-
tial way they differ. Let me thus briefly recall, as this constitutes the bulk
of this review, the nature of those objections and give a sketch as to how
they can be obviated.

II
Sabatier’s book is best understood as being structured by a set of basic
distinctions. In ontological matters, things revolve about the opposition
between apriorism and naturalism, and also around the acceptance, or the
rejection, of logical universalism, the latter being closely connected with
what the author calls ‘l’univocité de la notion d’existence’. The epistemo-
logical question seems exhausted by the triad intuition, grasp, and pos-
tulation, while the indispensability/dispensability alternative apparently
dominates the applicability problem.
(a) Ontology (la profusion des objets mathématiques)
Our sense of the word ‘object’ is acquired through our daily dealings with
various types of spatio-temporal objects. According to Ockham’s razor,
98 PHILOSOPHIA MATHEMATICA

one needs very good reasons to extend our basic ontology to include
‘objects’ that apparently do not belong to space-time. But common sense
may not be our best guide in scientific matters. According to the author,
Quine has convincingly shown that the gap between our beliefs in the exis-
tence of spatio-temporal items and our beliefs in the existence of some
mathematical items can actually be bridged, and that those beliefs are in
fact situated on a continuum, a hierarchy which also contains beliefs in the

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

existence of objects with rather ambiguous status, like elementary parti-
cles. So it does not seem unscientific to commit ourselves ontologically to
mathematical objects. To do away with them, i.e., to adopt nominalism in a
Quinean context, will necessitate showing that they are not actually indis-
pensable in the conceptual scheme (p. 225). Indispensability arguments
are also given by the structuralists, even if some of them (Shapiro, p. 241)
dissociate their views from the naturalistic framework while trying not to
enter into open contradiction with it.
Maddy’s ideas on this matter are different from the structuralists. For
her, some simple mathematical objects are actually part and parcel of the
spatio-temporal realm and are therefore ‘concrete’. Still, she has to rely
on the indispensability argument to justify the existence of more complex
objects (pp. 225, 233) and, according to Sabatier, this constitutes the main
weakness of her approach (p. 278). However, her views seem particularly
relevant for the so-called ‘access problem’.
An important consequence of those broadly Quinean views of Sabatier
is that, as he himself points out, logic and mathematics can no longer play
a foundational role for science as they certainly did for Frege and Russell.
(b) Access (l’accès aux objets mathématiques)
Here one faces Benaceraff’s epistemic argument. Given the independent
existence of mathematical objects, explaining our knowledge of them
requires showing how we can get in touch with those objects. To accom-
plish this, the argument goes, causal interaction is necessary. However, for
Platonism, mathematical objects are outside the causal network, outside
space and time. They are therefore causally inert, and cannot enter into
any of our epistemic plays. How then is it possible to explain the knowl-
edge of mathematical objects that is so obviously ours? While pointing out
certain difficulties of the ‘causal theory of knowledge’, the author, never-
theless, takes this ‘lack of access’ argument seriously. One of the possible
‘solutions’ would be to make things worse and generalize the predicament
to include empirical objects. Now, it seems that Quine and Gödel could
agree on the fact that even our most elementary beliefs about the world
involve some interactions with non-sensuous entities, unifying or organiz-
ing concepts that can hardly be accounted for by perception. Since empiri-
cal objects are the very paradigm of what truly exists, they are not affected
by the ‘access problem’ and therefore are not seriously threatened by any
 CRITICAL STUDIES/BOOK REVIEWS 99

of the anti-platonist arguments. Hence, our whole conceptual scheme jus-

tifies our beliefs in mathematical objects.
The author rightly notes (p. 77) that the access problem is understood
differently in logical Platonism than in mathematical Platonism. The dif-
ference is rooted in the absolutely central distinction between ‘grasp’ of
propositions and ‘intuition’ of numbers or mathematical items, both pro-
cesses being deemed ‘mysterious’. In a way, the former appears to contem-

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

porary analytical philosophy to be more general and more acceptable to
the sober philosophical mind than the latter. Actually, it seems absolutely
indispensable: to explain away any grasp of propositions is the plight of the
empiricist, and such an elimination is bound to have serious consequences
for the very possibility of objective science. Sabatier is here touching on an
issue that has played a tremendously important role in twentieth-century
philosophy: can and should the problem of reference be somehow dissoci-
ated from ‘intuition’? The debate over that has been rather acrimonious, so
much so that a prominent contemporary philosopher has qualified Gödel’s
views on intuition as ‘rank medievalism’. This shows clearly that some-
thing absolutely fundamental is here at play.
The problem of access is tackled differently by Penelope Maddy. She
innovates with respect to standard Quineism by describing some kind of
‘perception’ of (small) sets, going back in a way to some of Gödel’s ideas
on intuition, but this time in a naturalist rather than aprioristic framework.
Some mathematical objects are therefore part and parcel of the spatio-
temporal realm. Maddy’s perception is meant to explain the origins of
the ‘first truths’ of mathematics. An interesting aspect of Maddy’s con-
ceptions is her recognition of some pre-linguistic knowledge, which is in
opposition to Quine’s holistic scheme. This knowledge, which is fallible,
is the basic material used in the axiomatizations of set theory, the latter
being themselves fallible (p. 236). According to Sabatier, despite Maddy’s
differences with Quine, she accepts the basic tenets of his epistemology.
Sabatier thinks of her work as an interesting complement to the influential
views of the Harvard master, particularly with respect to the genesis of
basic knowledge about sets (pp. 223, 235, 237).
Not relying on ‘grasp’ and ‘intuition’, and in the wake of Quine,
the structuralists are rather ‘postulationist’ when it comes to solving the
access problem. For them, the access problem is more formidable since
they accept the traditional idea that mathematical items are not concrete
spatio-temporal objects and are thus causally inert. Like Maddy, they rec-
ognize the importance of explaining the genesis of basic mathematical
beliefs, but their solution in terms of pattern recognition is more linguisti-
cally minded than Maddy’s intuition, as for the structuralists, mathematical
structures and language are closely connected (p. 266). Indeed, they reject
intuition, and to them, with the development of mathematics, it becomes
100 PHILOSOPHIA MATHEMATICA

increasingly clear that the causes of beliefs are replaced by reasons of an

essentially discursive nature (p. 273).

(c) Incompleteness of Mathematical Objects (indétermination des objets

mathématiques)
This is related to another challenge posed by Benaceraff. The whole prob-
lem this time is based on an apparently natural distinction: on the one

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

hand one has the ‘substance’ of an object, what is responsible for its iden-
tity, what I refer to when I say ‘the table’. On the other hand, we have the
set of scientifically discovered properties of that object, which is subject
to change and reorganization (p. 42). When it comes to mathematics, the
distinction is highly problematic. Indeed, what could well be an ‘intrin-
sic’ property of 2? Is not 2 what it is by virtue of its relations with other
numbers? What else could it be? Perhaps, the best way to obviate the dif-
ficulty is to abandon the idea that numbers are objects. Would it not be
more natural to reduce numbers to sets? Then again, there is a problem:
which sets? There are infinitely many valid set-theoretic constructions of
the natural-number structure, i.e., infinitely many systems forming a recur-
sive progression. Since they have different properties that are, of course,
inconsequential as far as arithmetic is concerned, it seems that 2 must have
contradictory properties! Shall we then abandon the idea that numbers are
sets? The author finds an elegant solution to this conundrum in Maddy’s
idea according to which the various set-theoretic reconstructions of nat-
ural numbers are just as many ‘yardsticks’ that can be used to measure,
in different but equivalent ways, the size of sets. As yardsticks, they are
not objects, at least not in a traditional sense, and can therefore tolerate
some degree of ‘incompleteness’ (p. 239). The standard linguistic usage,
in which number terms are treated as nouns, so central for Frege (p. 84,
244), is therefore perhaps misleading.
This solution is clearly rejected by the structuralists. For them, mathe-
matical objects are ‘places’ or ‘positions’ in ‘structures’, and mathematics
is mainly the study of such structures. The incompleteness of mathemati-
cal objects is therefore of no concern for mathematical structuralism and
is indeed, according to Sabatier, ‘the central principle of this philosophy’
(p. 262). Since that position is at least partly grounded in a rejection of
logical universalism, its conformity to Platonism can be questioned.

(d) Applicability (l’applicabilité des mathématiques)

Mathematics is highly applicable to the empirical world and that, oddly
enough, is an argument for both the realist and the antirealist. The fact
that mathematics applies and seems indispensable to science, together with
the difficulty of isolating the purely mathematical contribution to physical
experience, is consonant with platonism. But, on the other hand, the ideal
character of mathematical objects poses a difficulty similar to the one faced
 CRITICAL STUDIES/BOOK REVIEWS 101

above in the access problem: how can physical objects interact with math-
ematical objects, since the latter are causally inert? The problem seems
especially urgent for logical platonism (p. 12), with its double emphasis
on generality and reference to logical objects: Mathematics applies in prin-
ciple to everything, but its basic items, such as numbers, form a realm of
their own. So, is mathematics really ‘general’ or trans-generic, or is it a sci-
ence in the Aristotelian sense, a discipline concerned with a distinct genus

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

or a separate ontological region (p. 33)?
The author reviews the logicist and the structuralist solutions, but
shows a preference for Maddy’s ‘empirical sets’. In all the solutions ema-
nating from a naturalist framework, though, one always starts with empir-
ical knowledge, the consideration of which shows the essential role of
mathematical statements and mathematical items.

III
This completes my review of some of the themes Sabatier discusses in
his fine book. By way of critique, let me point out that this writer seems
to think (pp. 167, 286, and also 247), along with some of the authors he
discusses, that an intuition of (or reference to) the cumulative hierarchy can
be used to explain how mathematics applies so well to the empirical realm.
I would rather think that this explains too much and, in another sense,
too little. Indeed, on the one hand it is hard to accept that the amazing
efficiency of mathematics in quantum field theory, for instance, can be
attributed to the fact that empirical objects can be included as basic items in
the cumulative hierarchy! It just does not seem appropriate. From another
perspective, this idea explains too much, since it does not account for the
relative inefficiency of mathematics in most of the non-physical sciences.
To quote René Thom:

Here we are at the root of the so-called miracle of physical laws,

which enabled sciences to produce its most powerful procedures for
prediction. To what extent can this miraculous situation be found
in other cases than fundamental physics? We shall see later that, in
all likelihood, the miracle of physical laws is unique and isolated;
hence our capacities for quantitative prediction are severely limited
to domains close to physics and mechanics.2

2 See René Thom, ‘The role of mathematics in present-day science’, in L.J. Cohen,
J. Los, H. Pfeiffer, and K.-P. Podewski, eds, Logic, Methodology and Philosophy of Science
VI, pp. 3–13. Amsterdam and Warsaw: North-Holland and Polish Scientific Publishers,
1982.
102 PHILOSOPHIA MATHEMATICA

Sabatier’s approach does not address the fact that a lot of mathematics
is not, and might never be, applied or even applicable. This is why I regard
the explanations given on this problem as rather weak and much too gen-
eral to be really convincing.3 The really surprising fact that demands an
explanation is not so much that mathematics can ‘hook on to the world’
and describe it — natural languages can do it too, and very well indeed —
but the fact that those mathematical descriptions can be used to make

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

astonishingly accurate predictions. It seems to me that philosophers should
finally consider Thom’s proposal seriously. It could be, at the very least, a
stimulating starting point for future investigations.
Appropriately, the author discusses Gödel’s use of Kant. However, he
should also, in my humble opinion, have said something more substan-
tial about Husserl’s influence on Gödel (after 1959). For instance, the
bipartition alluded to on page 128 seems reminiscent of the fundamental
distinction, which can be found in Formal and Transcendental Logic
(§27), between formal apophantics and formal ontology. The reading
of Husserl’s final Logic could perhaps explain the evolution in Gödel’s
ideas about the relations between logic and mathematics. Considering
the great mathematical logician’s emphasis on intuition, a discussion of
categorial intuition would perhaps also have been in order, the more so
since it is known that Gödel regarded very highly Husserl’s Sixth Logical
Investigation. Works by Tieszen, van Atten, Cassou-Noguès, and others
could have been used profitably here.
As to the clarity, it seems to me that a few schematic representations
would have done marvels to express the similarities and differences, often
subtle, between the semantic systems of Bolzano, Russell, and Frege.
Some basic knowledge of Frege and/or Russell is almost indispensable
to understand fully the second part of an otherwise almost self-contained
book.
However, let me emphasize that these are only minor aspects of a very
well-argued discussion. Overall, the book is well edited, with only a few
misprints, none of which are of any significance. I have not found any fac-
tual errors either. Within the limits it sets for itself, the book under review
is quite successful, both rigorous and well written. Besides, there are not
that many books on the topic (as far as I can tell, it is the only such study
in French). It is therefore a welcome addition to the literature, suitable for
graduate students, and even for seasoned philosophers or mathematicians
looking for a global survey of realism, written from the point of view of
mainstream philosophy of mathematics. Its very appearance in French is
of some importance in itself. No doubt it will become a basic reference
for French readers describing what some sectors of analytic philosophy

3 See the sensible remarks by Patras in the final chapter of Fréderic Patras, La pensée
mathématique contemporaine. Paris: P.U.F., 2001.
 BOOKS OF ESSAYS 103

have been up to. Hopefully, Sabatier will complement his excellent sur-
vey by a volume on mathematical realism devoted to the less orthodox
approaches.

doi: 10.1093/phimat/nkq026

Downloaded from http://philmat.oxfordjournals.org/ at University of Wisconsin-Oshkosh on June 7, 2015

Books of Essays

ROMAN MURAWSKI. Essays in the Philosophy and History of Logic and

Mathematics. Poznań Studies in the Philosophy of the Sciences and the
Humanities; 98. Amsterdam; New York: Rodopi, 2010. ISSN 0303-8157.
ISBN 978-90-420-3090-9 (hbk); 978-90-420-3091-6 (e-book). Pp. 343.

AUTHORS AND TITLES

Cantor’s philosophy of set theory, pp. 15–28.

Leibniz’s and Kant’s philosophical ideas vs. Hilbert’s program, pp. 29–39.
Truth vs. provability: Philosophical and historical remarks, pp. 41–57.
Philosophy of mathematics in the 20th century: Main trends and doctrines,
pp. 59–73.
On new trends in the philosophy of mathematics, pp. 75–84.
Remarks on the structuralistic epistemology of mathematics (with Izabela
Bondecka-Krzykowska), pp. 85–93.
From the history of the concept of number (with Thomas Bedürftig),
pp. 95–122.
Church’s thesis and its epistemological status, pp. 123–134.
Phenomenology and philosophy of mathematics, pp. 135–146.
Hoene-Wroński — genius or madman?, pp. 149–160.
Grassmann’s contribution to mathematics, pp. 161–168.
Giuseppe Peano and symbolic logic, pp. 169–182.
E.L. Post and the development of logic, pp. 183–193.
John von Neumann and Hilbert’s school, pp. 195–209.
Contribution of Polish logicians to decidability theory, pp. 211–231.
Contribution of Polish logicians to predicate calculus, pp. 233–243.
The English algebra of logic in the 19th century, pp. 245–269.
The development of symbolism in logic and its philosophical background
(with Thomas Bedürftig), pp. 271–299.
Consolidated references, pp. 301–334.

doi: 10.1093/phimat/nkq022