SHAPIRO, THINKING ABOUT MATHEMATICS

Chapter X. Structuralism

Mathematics is defined by the central slogan that it is the science of structure. This
perspective is shared by several prominent philosophers, including Paul Benacerraf,
Geoffrey Hellman, Michael Resnik, and Stewart Shapiro. While these thinkers vary
on the existence of mathematical objects, they generally agree that mathematical
sentences have objective truth-values independent of the language, mind, and social
conventions of the mathematician. The structuralist enterprise provides a unique way
to understand the nature of mathematical objects, such as numbers, sets, and points,
by viewing them as positions within patterns rather than isolated entities.

1. The Underlying Idea

Traditional platonists, or realists in ontology, typically hold that the subject-matter of

a given branch of mathematics is a collection of objects that possess a sort of
ontological independence. For such a philosopher, numbers are on a par with
physical objects like automobiles, suggesting that one could theoretically understand
the essence of the number 2 without invoking its relationship to 6 or 6,000,000. The
structuralist, however, vigorously rejects this notion of independence. They argue that
the essence of a natural number lies entirely in its relations to other natural
numbers. In this view, arithmetic is not the study of individual objects, but the study
of a single abstract structure: the pattern common to any infinite collection of
objects that has a unique initial object and a one-to-one successor relation satisfying
the principle of induction.

To clarify this, Shapiro distinguishes between a system and a structure. A system is a

collection of objects with specific relations among them, such as a corporate
hierarchy, a chess configuration, or a basketball defense. A structure is the abstract
form of such a system, highlighting the interrelationships while ignoring features of
the objects that do not affect those relations. We apprehend these patterns through a
process of abstraction, where we focus on the spatial relations and roles—such as the
point guard in a defense—while ignoring irrelevant features like hair color or field
goal percentage. In mathematics, these structures are studied as such, independent of
any specific physical instances. As Michael Resnik puts it, mathematical objects are
structureless points or positions that have no identity outside of their respective
structures.

This perspective effectively eliminates many traditional philosophical puzzles, such as

the Benacerraf-Kitcher problem. This problem arises from the fact that there are
multiple ways to reduce arithmetic to set theory—such as the Zermelo or von
Neumann reductions—which lead to conflicting answers about whether the number 1
is a member of the number 3. For the structuralist, these questions about the "internal
composition" of a number are misguided because the number 3 is nothing more than a
place in a pattern. Asking if 1 is a member of 3 is as nonsensical as asking if the
point guard position is taller than the power forward position; height applies to the
people playing the roles, not the offices themselves. This leads to a form of
ontological relativity where mathematical objects are strictly tied to the structures
that constitute them.
3. Structuralism Without Structures

While ante rem structuralism treats structures as independent objects that exist prior
to any systems exemplifying them, other philosophers prefer an in re approach. This
more conservative view, known as eliminative structuralism, seeks to avoid
countenancing structures as bona fide objects in their own right. Instead, it reinterprets
mathematical statements as convenient shorthand for generalizations over all
systems of a certain type. For instance, a statement like "3 + 9 = 12" is rephrased to
mean that in any natural number system, the object in the 3-place added to the
object in the 9-place results in the object in the 12-place.

The primary challenge for the eliminative structuralist is the requirement of a robust
background ontology. If the program is to avoid being vacuous, there must actually
be enough objects in the universe to form the systems being discussed. If the
background ontology were finite, then all generalizations over infinite systems—like
those required for arithmetic or real analysis—would be technically true but
mathematically useless, as there would be no systems to satisfy the premises. To
resolve this, some structuralists adopt the ontological option, which postulates the
existence of enough abstract objects (usually via set theory) to exemplify any
structure under study. In this framework, the set-theoretic hierarchy serves as the
"universal" background ontology that fills the places of other structures.

Alternatively, many ontological anti-realists favor the modal option, which replaces
the requirement of actual systems with possible systems. Geoffrey Hellman provides
the most detailed execution of this in his modal eliminative structuralism, famously
articulated in his work Mathematics Without Numbers. Rather than asserting that the
natural numbers exist, Hellman asserts that it is logically possible for such a system
to exist. Mathematical theorems are then understood as claims about what would
necessarily hold in any logically possible system of that type.

To maintain the integrity of this nominalist program, Hellman must ensure that the
logical modalities he uses are not themselves defined in terms of set theory, which
would result in an unacceptable circularity. Consequently, he treats logical possibility
and necessity as primitive notions rather than reducing them to model-theoretic
existence within a set-theoretic hierarchy. This allows the modal structuralist to
account for the objectivity of mathematics without committing to a platonistic
ontology of abstract objects, though it trades that ontology for a complex modal
ideology. Ultimately, both ante rem and eliminative versions of structuralism
highlight that mathematics is less about "things" and more about the logical
dependencies and patterns that define them.

BENACERRAF, WHAT NUMBERS COULD NOT BE

The central objective of Paul Benacerraf’s seminal work is to challenge the traditional
realist assumption that numbers are specific objects, particularly the claim that they
can be identified with sets,. Benacerraf begins by examining the logic of
reductionism, noting that while it is possible to provide a rigorous foundation for
arithmetic within set theory, this very success creates a profound philosophical
dilemma,. If numbers were truly objects, there would have to be a unique, determinate
set that corresponds to each number; however, the availability of multiple, equally
adequate but mutually exclusive reductions suggests that numbers lack the "internal
composition" required of bona fide objects,.

The Tale of Two Reductions

To illustrate this, Benacerraf introduces two competing set-theoretic accounts of the

natural numbers: those of Ernst Zermelo and John von Neumann. In Zermelo's
system, the number 0 is identified with the empty set ($\emptyset$), the number 1
with the singleton of the empty set (${\emptyset}$), and each subsequent number
$n$ is defined as the singleton of its predecessor. Thus, for Zermelo, the number 2 is
${{\emptyset}}$ and 3 is ${{{\emptyset}}}$. Conversely, von Neumann defines each
number as the set of all its predecessors. In this account, 0 is $\emptyset$, 1 is
${\emptyset}$, 2 is ${0, 1}$ or ${\emptyset, {\emptyset}}$, and 3 is ${0, 1, 2}$ or
${\emptyset, {\emptyset}, {\emptyset, {\emptyset}}}$.

Both systems are mathematically impeccable; they both satisfy the Peano axioms
and allow for the derivation of all arithmetical truths,. However, they disagree on non-
arithmetical, set-theoretic properties. For example, in von Neumann’s account, the
number 1 is a member of the number 3, whereas in Zermelo’s account, it is not,.
Benacerraf argues that if numbers were indeed sets, there would have to be a "right"
answer to the question of whether $1 \in 3$, yet arithmetic itself remains indifferent
to such a choice,.

The Argument Against Numbers as Objects

Because there is no principled reason to prefer Zermelo’s sets over von Neumann’s,
Benacerraf concludes that the search for the "real" set-theoretic identity of numbers is
a misguided enterprise,. The fact that we can choose any system that forms a
progression (an infinite sequence with a starting point and a successor relation)
proves that the specific nature of the elements in the sequence is irrelevant to their
function as numbers,. If $n$ were a particular object, it would have to be that object
regardless of the system it belongs to; but since "2" can be represented by any object
playing the role of the third element in a progression, it follows that "2" is not an
object at all, but rather a position in a structure,.

This leads Benacerraf to a form of ontological relativity. He asserts that identity

statements, such as "the number 2 is ${{\emptyset}}$," lack a determinate truth-value
because they attempt to equate entities from different categories without a shared
framework for individuation,. Just as it makes no sense to ask if the "point guard"
position on a basketball team is taller than the "power forward" position—since height
is a property of the office-holders and not the offices themselves—it is nonsensical to
ask about the set-theoretic properties of numbers,. Numbers, according to Benacerraf,
are structureless points.

Numbers as Positions in a Structure

The positive thesis emerging from this critique is a foundational version of

structuralism. Benacerraf suggests that what mathematicians are truly studying are
not objects but relations. Arithmetic is the "elaboration of the properties of all
systems of the order type of the numbers". This perspective shifts the focus from
metaphysics to structural form; the "essence" of a natural number lies entirely in its
relationship to other numbers within the pattern.

This eliminative or in re structuralism implies that mathematical statements are

essentially generalizations over all possible systems that exemplify a particular
structure,. When a mathematician asserts that $2 + 2 = 4$, they are not making a
claim about a specific entity "2," but rather stating that in any system that satisfies the
natural number structure, the object in the second place added to itself will result in
the object in the fourth place,. Consequently, the ontological commitment of
mathematics is not to a realm of ghostly, abstract individuals, but to the logical
dependencies that define the system as a whole,.

Conclusion: The Rejection of Platonism

Ultimately, Benacerraf's argument serves as a powerful critique of Platonism (or

realism in ontology),. By showing that any attempt to "pin down" numbers as sets
results in arbitrary choices, he demonstrates that numbers do not have the kind of
independent existence usually attributed to physical objects,. He identifies a
significant dilemma for the realist: if we insist on a uniform semantics where "2"
refers to an object just as "Julius Caesar" does, we are left with the epistemic mystery
of how we could ever know which specific set "2" is,.

By concluding that numbers are not objects, Benacerraf provides an alternative that
preserves the objectivity of mathematical truth while avoiding the pitfalls of a bloated
ontology,. Mathematics is the science of structure, and its "objects" are merely
positions or "offices" within the patterns that describe the world,. This work remains
a cornerstone of contemporary philosophy, forcing a reevaluation of what it means for
a mathematical entity to exist,.