20.09.03 · philosophy / phil-of-math

Mathematical ontology: Platonism, nominalism, structuralism, fictionalism

stub3 tiersLean: nonepending prereqs

Anchor (Master): Benacerraf, P. — What Numbers Could Not Be (1965)

Intuition Beginner

Do numbers exist? Most mathematicians and philosophers are mathematical Platonists. They believe numbers, sets, and functions really exist, but not in the physical world or in our minds. They exist in an abstract realm, outside space and time. The number seven existed before any human thought of it, before the universe began. Plato held the same view about all his Forms: perfect triangles, justice, and beauty all dwell in a transcendent realm of abstract entities.

Paul Benacerraf raised a sharp objection. If numbers are abstract objects outside space and time, how do we know about them? We learn about physical things through causal contact. We see tables, hear music, and touch stone. But we have no causal contact with an abstract number. Mathematical knowledge looks nothing like perception. So how does the Platonist explain that we know arithmetic at all? This is the problem of access.

Nominalists deny that abstract mathematical objects exist. Perhaps mathematics is only about symbols, language, or patterns we find in concrete things. Fictionalists go further. Hartry Field compares mathematics to a novel. Sherlock Holmes is useful for telling detective stories, even though no such person exists. Likewise, numbers are useful for doing physics, even though no such entities exist. Mathematics is a convenient fiction, not a body of truths.

Structuralists, led by Stewart Shapiro, take a different line. Mathematics is not about objects at all. It is about structures: the natural-number structure, the real-number structure, the group structure. The number seven is not a thing but a position inside the natural-number pattern. What matters are the relations between positions, not what the positions are made of. Structuralism sidesteps the question of what a number is by changing the subject.

Visual Beginner

Imagine the four positions as four rooms in one building, each answering the same question differently: what is a number?

Position	Do numbers exist?	In what sense?	Key advocate
Platonism	Yes	Abstract, mind-independent objects	Gödel
Nominalism	No	Symbols, language, or concrete patterns	Goodman, Quine
Structuralism	Structures yes; number-objects no	Positions in an abstract pattern	Shapiro
Fictionalism	No	Useful fictions, like characters in a novel	Field

Platonism and nominalism fight over whether abstract objects exist. Structuralism refuses the question and talks about patterns instead. Fictionalism agrees with nominalism that numbers do not exist, then asks why we keep using them so successfully.

Worked example Beginner

Suppose numbers really are sets, as some logicians propose. Then the number 3 must be one particular set. But which one? Two definitions both work, and they expose the trouble.

John von Neumann built each number from the previous ones. The number 0 starts empty. The number 1 contains only 0. The number 2 contains 0 and 1. The number 3 contains 0, 1, and 2. Ernst Zermelo nested them differently: each number is the single set wrapping the previous number, like boxes inside boxes.

Both definitions satisfy every law of arithmetic. Both let you add, multiply, and prove theorems about 3. Yet they disagree about membership. Is the number 2 inside the number 3? For von Neumann, yes. For Zermelo, no. Arithmetic is silent on the question.

No fact decides which set the number 3 "really" is. This is Benacerraf's identification problem. It pushes toward structuralism: the number 3 is not a set but a place in the counting pattern, and any system filling those places equally well counts as the natural numbers. The takeaway is that numbers have no identity beyond their structural role.

Check your understanding Beginner

Formal definition Intermediate+

The four positions are defined by their answers to two questions: do mathematical objects exist, and what are the truth conditions for mathematical statements?

Mathematical Platonism (Gödel, Hardy). Mathematical objects — numbers, sets, functions, spaces — exist as abstract entities: non-spatiotemporal, acausal, and mind-independent. A mathematical statement is true or false in virtue of the properties of these objects. Gödel held that we possess a faculty of mathematical intuition, analogous to sense perception, that gives access to the abstract realm. Platonism secures the objectivity and necessity of mathematics but inherits the epistemological burden of explaining how a mind makes contact with causally inert objects.

Nominalism (Goodman and Quine, "Steps Toward a Constructive Nominalism," 1947). No abstract mathematical objects exist; only concrete individuals do. Apparent reference to numbers must be paraphrased away, or ordinary mathematical discourse must be accepted as strictly false. The programme faces the obstacle that large parts of mathematics — measure theory, classical analysis, set theory — appear to require irreducible quantification over abstract totalities. Modern variants include Chihara's constructibility theory, which replaces existence with constructibility, and Bigelow's defence of numbers as universals.

Structuralism (Shapiro, Resnik). Mathematics is the science of structures, not objects. A number is a position in a structure, identified by its relations to other positions, not by any intrinsic nature. Three varieties differ on the status of structures themselves. Ante rem structuralism (Shapiro, Philosophy of Mathematics: Structure and Ontology, 1997; Resnik, Mathematics as a Science of Patterns, 1997) treats structures as freestanding abstract entities that exist even with no instance. In re structuralism grants existence only to structures instantiated in concrete systems. Modal structuralism (Hellman, Mathematics Without Numbers, 1989) replaces "the natural-number structure exists" with "necessarily, any system satisfying the Peano axioms would have these properties," trading abstract objects for modal primitives.

Fictionalism (Field, Science Without Numbers, 1980). Mathematical objects do not exist, and statements quantifying over them are, strictly, false — as statements about Sherlock Holmes are false because there is no Sherlock Holmes. The fictionalist explains the usefulness of mathematics through conservativeness: a mathematical theory $M$ is conservative over a nominalistic physical theory $N$ when every nominalistic consequence of $N \land M$ is already a consequence of $N$ alone. If mathematics is conservative, using it adds no new physical conclusions; it merely streamlines derivation. Field attempted to demonstrate conservativeness by reconstructing Newtonian gravitation using only qualitative relations among space-time points.

Position	Existence claim	Truth conditions	Central advocate
Platonism	Abstract objects, mind-independent	Correspondence to abstract reality	Gödel
Nominalism	No abstract objects	Paraphrase or rejection	Goodman, Quine
Structuralism	Structures (ante rem); or only instances (in re); or modal	Structural relations	Shapiro, Resnik, Hellman
Fictionalism	No objects; useful fictions	Strictly false; conservative over nominalist science	Field

Key argument — Benacerraf's access problem, the identification problem, and the turn to structuralism Intermediate+

Paul Benacerraf's two papers — "What Numbers Could Not Be" (1965) and "Mathematical Truth" (1973) — are the pivot around which contemporary mathematical ontology turns. The first attacks the objecthood of numbers; the second attacks the knowability of any abstract mathematical object. Together they make structuralism the most attractive compromise.

The access problem (1973). Benacerraf frames a dilemma with two constraints. The semantic constraint demands a uniform account of truth: mathematical statements should be true or false in the same general way as other statements, by virtue of how things are with the objects they concern. The epistemological constraint demands an account of how we come to know mathematical truths, given that knowledge ordinarily requires some access — causal or otherwise — to the objects known.

Platonism satisfies the semantic constraint cleanly: "there are infinitely many primes" is true because the primes, an abstract totality, are thus arranged. But Platonism strains the epistemological constraint. Abstract objects are acausal and non-spatiotemporal; the causal theory of knowledge, the default in post-Gettier epistemology 20.01.02 pending, delivers no route to them. Gödel's proposed faculty of mathematical intuition is the canonical response, but it is widely judged under-described: it offers no account of its reliability, no error theory, and no criterion for distinguishing genuine intuition from mere conviction.

Constructivism and formalism satisfy the epistemological constraint — mathematical objects are mental constructions or formal marks we produce — but struggle with the semantic constraint, because the resulting truth conditions look nothing like ordinary referential truth. The dilemma is that no single position satisfies both constraints at once.

The identification problem (1965). Suppose, with the Platonist, that numbers are objects, and suppose, with the logicist, that numbers are sets. Then the number 3 is some one set. Benacerraf exhibits two reductions that both satisfy every arithmetical truth. The von Neumann reduction identifies $3$ with ${0, 1, 2}$ , where $0 = \emptyset$ ; the Zermelo reduction identifies $3$ with ${{{\emptyset}}}$ . They agree on every law of arithmetic and disagree on every set-theoretic property: for von Neumann, $2 \in 3$ ; for Zermelo, $2 \in / 3$ .

If numbers were objects, there would be a fact of the matter whether $2 \in 3$ . Arithmetic is silent, and the two reductions are equally good at underwriting arithmetic. The conclusion is that numbers are not sets, and more generally that numbers are not objects with intrinsic identity. A number has only the structural role it plays — successor, addition, ordering — and nothing more. This is a direct argument for structuralism: what exists is the natural-number structure, of which each number is a position.

The turn to structuralism. Structuralism inherits the strengths of Platonism (objective, mind-independent truth) while sidestepping the identification problem. The number 7 is not a set but the seventh position in any omega-sequence satisfying the Peano axioms. The access problem remains, but is softened: the structuralist need not explain contact with a unique abstract object 7, only grasp of a pattern. Ante rem structuralism keeps the pattern as an abstract entity; modal structuralism (Hellman) keeps only its possibility; in re structuralism keeps only its instances.

Counterexamples to common slips

"Structuralism is just a weakened Platonism." Ante rem structuralism is a genuine ontological commitment to structures as abstract entities, comparable in strength to Platonism. Modal and in re structuralism are weaker, but they are not "Platonism minus something" — they replace existence claims with modal or instantiation claims that change the logic of mathematical discourse.
"Fictionalism says mathematics is meaningless." Fictionalism says mathematical statements are false (because their quantifiers range over nothing), not meaningless. The fictionalist treats mathematical theories as internally coherent and useful, exactly as a literary critic treats a novel as coherent and illuminating without believing its characters exist.
"Benacerraf proves numbers do not exist." The 1965 argument proves that numbers are not sets with determinate membership structure, and motivates the stronger claim that numbers are not objects at all. It is silent on whether structures exist. Platonism about structures survives even when Platonism about individual numbers does not.
"The access problem refutes Platonism." It poses a demand — supply a non-causal epistemology — that Platonists have not met to general satisfaction. It does not deductively refute the position. The indispensability argument 20.08.02 pending is one route by which Platonists try to meet the demand, and it remains live.

Exercises Intermediate+

Exercise 2 (easy, multiple choice).

The core of Benacerraf's 1965 identification problem is that:

A. Numbers must be physical, but no physical object is a number.

B. Two equally adequate set-theoretic reductions of the numbers disagree about membership facts, so no set is the number.

C. The indispensability argument fails because mathematics is not tested by experiment.

D. Constructivists cannot recover classical analysis.

Hint

The argument compares von Neumann and Zermelo reductions, which agree on arithmetic and disagree on which sets belong to which.

Answer

Option B. Two reductions both satisfy the Peano axioms yet disagree about whether, for instance, $2 \in 3$ . Since arithmetic fixes no such fact, no particular set is identical with a number. This motivates structuralism. A restates a nominalist slogan; C concerns indispensability; D concerns constructivism.

Exercise 3 (medium, argument reconstruction).

Reconstruct Benacerraf's 1973 dilemma in premise-conclusion form. State the two constraints, then explain which constraint each of Platonism and constructivism satisfies and which it fails.

Hint

One constraint is semantic (uniform truth conditions), the other epistemological (an account of access to the objects known).

Answer

P1 (semantic constraint). A satisfactory account of mathematical truth must assign truth conditions uniform with truth conditions in general: mathematical statements are made true by how things are with the objects they concern.

P2 (epistemological constraint). A satisfactory account of mathematical knowledge must explain how we know mathematical truths, given that knowledge ordinarily requires access — causal or otherwise — to the objects known.

Platonism satisfies P1: abstract objects make mathematical statements true or false. It strains P2: abstract objects are acausal and non-spatiotemporal, so the causal theory of knowledge gives no access, and Gödelian intuition is not a satisfactory substitute.

Constructivism satisfies P2: mathematical objects are mental constructions we produce and inspect. It strains P1: truth becomes tied to construction rather than to an independent reality, so the semantics is non-uniform.

Conclusion. No extant position satisfies both constraints simultaneously. The dilemma drives the search for positions — structuralism, fictionalism, indispensability-based Platonism — that aim to meet both at once.

Exercise 4 (medium, short answer).

Explain Field's conservativeness claim and how it supports fictionalism. Then state Malament's central objection.

Hint

Conservativeness says that adding mathematics to a nominalist physical theory yields no new nominalist conclusions. Malament's objection concerns whether the nominalisation extends beyond Newtonian gravitation.

Answer

Conservativeness. A mathematical theory $M$ is conservative over a nominalistic theory $N$ when every nominalistic consequence of $N \land M$ is already a consequence of $N$ alone. If mathematics is conservative, then using it is harmless: it adds no new physical conclusions, only shortens derivations. Field reconstructs Newtonian gravitation in Science Without Numbers using only qualitative relations among space-time points (betweenness, congruence) and proves a representation theorem matching the standard real-valued formulation. This shows, for one theory, that mathematics is dispensable in principle, so its usefulness does not commit us to its objects.

Malament's objection. The nominalisation requires substantival space-time points and assumes structural properties (continuity, infinite divisibility) that are themselves mathematically loaded. More seriously, the programme does not extend to theories that are essentially mathematical: general relativity's metric tensor, quantum mechanics' Hilbert spaces, and quantum field theory have no known nominalistic reconstruction. Field's programme demonstrates dispensability for a restricted fragment, not for the whole of physics.

Exercise 5 (medium, short answer).

Distinguish ante rem, in re, and modal structuralism. For each, state what exists and what the number 7 is.

Hint

The three differ on whether the natural-number structure exists independently, only in its instances, or only possibly.

Answer

Ante rem (Shapiro, Resnik): the natural-number structure exists as a freestanding abstract entity, even with no instance. The number 7 is the seventh place in this abstract pattern, with identity fixed by its relations to the other places.

In re: structures exist only insofar as they are instantiated in concrete systems. There is no abstract pattern without an instance. The number 7 is a role played by the seventh element of any concrete omega-sequence; without such a sequence there is no 7.

Modal (Hellman): the structure need not actually exist; it need only be possible. The number 7 is the seventh place in any possible system satisfying the Peano axioms. The modal operator replaces the existential commitment, trading abstract objects for the primitives of modal logic.

Exercise 6 (hard, argument reconstruction).

Reconstruct the Quine-Putnam indispensability argument in premise-conclusion form. Then state Field's response and Baker's strengthening.

Hint

The argument runs from the indispensability of mathematics to our best science, via Quine's criterion of ontological commitment, to the existence of mathematical objects. Baker's cicada example concerns explanatory indispensability, not merely deductive.

Answer

P1. We ought to be ontologically committed to all and only the entities indispensable to our best-confirmed scientific theories (Quine's criterion: "to be is to be the value of a bound variable").

P2. Mathematical entities are indispensable to our best-confirmed scientific theories: those theories quantify over real numbers, functions, Hilbert spaces, and the like, and cannot be formulated without them.

C. We ought to be ontologically committed to mathematical entities.

Field's response. P2 is false in principle. Mathematics is dispensable because it is conservative over nominalistic science; a nominalistic reconstruction of the physics removes the quantification over abstract objects while preserving the physical consequences. The argument therefore fails to force commitment.

Baker's strengthening. Even if mathematics is deductively dispensable, it is explanatorily indispensable. Baker's cicada example: certain cicadas have prime-numbered life cycles (13 and 17 years), and the best explanation of this fact appeals to the number-theoretic property of primality. If mathematics is doing genuine explanatory work, then removing it removes the explanation, which is a cost an indispensability theorist can levy against the nominalist. Saatsi replies that explanatory value does not entail ontological commitment.

Indispensability, explanation, and mathematical practice Master

The Quine-Putnam indispensability argument

The strongest argument for mathematical Platonism is not Gödel's intuition but the indispensability argument, developed by Quine in "On What There Is" (1948) and "Two Dogmas of Empiricism" (1951), and sharpened by Putnam in "What is Mathematical Truth?" (1975) and Philosophy of Logic (1971). The argument applies Quine's naturalised epistemology — on which we believe in electrons because our best physics cannot be stated without them — to mathematics. If our best science quantifies over real numbers, functions, tensors, and Hilbert spaces, and if we accept that science as true, then by Quine's own criterion of ontological commitment we are committed to those mathematical entities. To accept the physics while denying the mathematics is, on the Quinean picture, an unstable dualism: the mathematics and the physics come as a package in a single confirmed theory. The argument turns anti-Platonism into a cost to be paid, not a default to be enjoyed.

Field's Science Without Numbers (1980) is the most sustained attempt to pay that cost. The strategy is to show that mathematics is dispensable in principle by producing a nominalistic physics — a formulation of the science that quantifies only over concrete space-time regions and their qualitative relations. Where the standard theory says "the gravitational potential is a real-valued function," the nominalist theory says "these regions stand in such-and-such betweenness and congruence relations," and a representation theorem guarantees that the two agree on every physical prediction. If the programme succeeded across all of physics, indispensability would collapse and the argument would lose its second premise.

It has not succeeded. David Malament's 1982 review argued that even the Newtonian case requires substantival space-time points and structural assumptions (continuity, infinite divisibility) that are themselves mathematical, and that the extension to general relativity — whose metric tensor is not paraphrasable away — and to quantum mechanics is not in prospect. Mark Colyvan's The Indispensability of Mathematics (2001) surveys the damage and concludes that the argument survives, strengthened by the observation that mathematical entities play an explanatory, not merely deductive, role in science 20.08.02 pending, 20.08.03 pending.

Alan Baker's "Mathematical Indispensability and Scientific Practice" (2005; 2009) sharpens the point with the cicada example. Periodical cicadas of the genus Magicicada have life-cycle lengths of 13 and 17 years — both prime. The best explanation is evolutionary: a prime cycle minimises the frequency with which the cicadas encounter predators or competing broods whose own cycles are smaller integers, because a prime number shares no factors with those integers. The explanation is irreducibly number-theoretic; removing the appeal to primality removes the explanation itself. Baker argues that explanatory indispensability is a stronger commitment than deductive indispensability: a nominalist can tolerate mathematics as a deductive shortcut, but tolerating it as the only available explanation is harder. Juha Saatsi responds that explanatory power does not entail ontological commitment — a model can explain without its ontology being read literally — and the exchange remains live.

Fictionalism and its discontents

Field's fictionalism is the most thoroughgoing nominalism on offer, and its fate is bound to the conservativeness programme. If mathematics is conservative over nominalistic science, then the fictionalist can use mathematics freely, treating it as a calculating device that adds no new physical content. The programme demands a conservativeness proof for each mathematical theory over each nominalistic reconstruction, and these proofs are hard: Field proved one for a fragment of Newtonian gravitation, and no proof exists for the theories that dominate contemporary physics. Mary Leng's Mathematics and Reality (2010) defends a fictionalism that concedes indispensability for the formulation of science while denying it for the content: the physical content of general relativity can in principle be stated without tensors, even if tensors are the natural vehicle. Mark Balaguer's Platonism and Anti-Platonism in Mathematics (1998) argues that full-blooded Platonism and fictionalism are both defensible and that no consideration decides between them — a pluralism that leaves the ontological question genuinely open. Gideon Rosen's modal fictionalism treats mathematical discourse as governed by a prefix "according to the mathematical fiction," preserving the surface logic of mathematics without its existential commitments.

Structuralism and the problem of structure identity

Structuralism answers the identification problem cleanly — numbers are positions, not objects — but inherits a deeper question: when are two structures the same? The natural answer is up to isomorphism, but isomorphism is itself a structural notion, defined in terms of maps that preserve structure. If structures are defined by isomorphisms and isomorphisms by structures, the account threatens circularity. Shapiro's ante rem structuralism bites the bullet: structures are coherent abstract patterns, and coherence is the structuralist analogue of existence. The cost is that "coherence" must be made precise without presupposing the set-theoretic machinery it is meant to ground. Hellman's modal structuralism avoids abstract structures but owes an account of the modal primitives it substitutes; Charles Parsons (Mathematical Thought and Its Objects, 2008) presses the worry that modal operators are no less obscure than the abstract objects they replace, and distinguishes structures grounded in intuitive abstraction (the natural numbers, reachable by counting) from those requiring stronger commitment (arbitrary ZFC sets). Category theory offers a third option 20.09.02 pending: Steve Awodey and others treat structural identity as primitive, made precise by the univalence axiom in homotopy type theory, on which identity of types is equivalent to equivalence of types. Whether this dissolves or relocates the circularity is itself contested.

Mathematical explanation and non-causal accounts

The cicada case is one instance of a broader category: mathematical explanations of physical phenomena. Honeycomb hexagons are explained by the isoperimetric theorem (hexagons minimise perimeter for a tessellating cell). Bridge stability, the stability of the solar system, and the spacing of atomic energy levels all receive explanations in which the mathematics is doing work that no causal story replaces. Marc Lange's Because Without Cause (2017) develops a theory of non-causal explanation in which mathematical explanations form a distinctive class: they explain by showing that the phenomenon could not have been otherwise, given the relevant mathematical constraints, rather than by citing causal mechanisms 20.08.03 pending. This bears directly on indispensability: if mathematics is sometimes the only available explanation, the nominalist must either supply a non-mathematical replacement or accept that some physical facts have no explanation, which is a high cost. The explanatory turn has moved the indispensability debate from "is mathematics deductively needed?" to "is mathematics explanatorily needed?", and the latter is harder for the nominalist to answer.

Mathematical practice and naturalism

Penelope Maddy's Naturalism in Mathematics (1997) and Second Philosophy (2007) argue that the ontological question should be settled by attending to mathematical practice itself, not by importing external philosophical constraints. On Maddy's naturalism, the mathematician's own standards — proof, rigour, fruitfulness — are the relevant arbiters, and the philosopher has no standing to overrule them on metaphysical grounds. Set theorists who adopt large-cardinal axioms do so because those axioms yield a richer, more unified theory, and that internal justification suffices; the question of whether large cardinals "really exist" is, for the naturalist, a question internal to the practice. This dissolves the access problem by rejecting its premise: there is no extra-mathematical epistemological constraint to satisfy. William Thurston's "On Proof and Progress in Mathematics" (1994) complements the position empirically, documenting how working mathematicians actually think — in geometric images, in analogies, in computations — rather than in the formal proofs that philosophy of mathematics has traditionally taken as the whole of the activity. David Corfield's Towards a Philosophy of Real Mathematics (2003) and Reuben Hersh's What Is Mathematics, Really? (1997) press the same point: a philosophy that addresses only formal proofs and set-theoretic foundations ignores most of what mathematicians do.

Diagrams, visualisation, and thought experiments

The naturalist attention to practice has drawn the epistemic role of diagrams into focus. Marcus Giaquinto's Visual Thinking in Mathematics (2007) argues that diagrams are not mere aids to intuition but genuine sources of mathematical evidence, on which proofs sometimes depend. Kenneth Manders's study of the Euclidean diagram shows that diagrammatic reasoning in Euclid is rigorous when read in its own terms: the diagram licenses exactly those inferences that survive the practice's controlled reading, and the modern demand that all rigour be symbolic is a local prejudice, not a universal norm. James Robert Brown's Philosophy of Mathematics (1999) goes further, defending mathematical thought experiments — imagined constructions that establish results without literal proof — as a legitimate and irreducible mode of mathematical knowledge. These positions connect mathematical ontology to the epistemology of perception and imagery 29.03.02 pending: if diagrams yield knowledge, then mathematical knowledge has a visual component that pure Platonism and pure formalism struggle to accommodate.

The applicability of mathematics

Eugene Wigner's 1960 essay "The Unreasonable Effectiveness of Mathematics in the Natural Sciences" names a puzzle that every ontological position must address. Mathematical concepts developed for internal reasons — complex numbers for algebra, group theory for the solvability of equations, Riemannian geometry for its own sake — turn out, decades or centuries later, to describe the physical world with a precision no one mandated. Mark Steiner's The Applicability of Mathematics as a Philosophical Problem (1998) sharpens the puzzle: the physicist's strategy of choosing formalisms for their beauty, analogy, and formal elegance, then reading off physical predictions, succeeds far more often than it should if mathematics and the world had no special relation. Steiner calls the success "anthropocentric" and offers it as evidence against any purely naturalistic account.

Several responses are available. The Fregean response holds that applicability follows from generality: mathematics describes any possible structure, so any physical structure will instantiate some mathematical description. The selection response holds that the apparent miracle is survivorship bias — of the vast landscape of mathematical theories, a few fit the world by chance, and those are the ones we remember. Christopher Pincock's Mathematics and Scientific Representation (2012) and Jody Azzouni's Tracking Reason (2008) treat applied mathematics as idealisation: the mathematics fits because the physical situation has been abstracted and idealised until a tractable structure emerges, so the fit is constructed, not discovered. None of these responses is universally accepted; the puzzle of applicability remains a productive interface between the philosophy of mathematics and the philosophy of science, and it constrains what any satisfactory ontology must explain.

Category theory, aesthetics, and pluralism

Category theory and homotopy type theory raise an ontological question distinct from the Platonism-nominalism axis 20.09.02 pending. Jean-Pierre Marquis's From a Geometrical Point of View (2009) and Colin McLarty's work on Grothendieck and the logic of number theory argue that category-theoretic foundations take structural identity as primitive: objects are determined only up to isomorphism, and the univalence axiom ( $A ≃ B \to A = B$ ) makes the structuralist slogan a theorem of the foundation. Elaine Landry extends the point to structural realism in physics 20.08.02 pending, [20.03.*]. Whether this constitutes a fourth ontological option or a refinement of structuralism is open.

Mathematical aesthetics intersects ontology through the role of beauty in mathematical discovery. G. H. Hardy's A Mathematician's Apology (1940) treats beauty as a mark of mathematical significance, and Gian-Carlo Rota's "The Phenomenology of Mathematical Beauty" (1997) analyses the concept without reducing it to utility [20.04.]. If aesthetic judgement guides the selection of axioms and conjectures, then the ontology of mathematics is shaped in part by an aesthetic faculty whose epistemology is itself unaccounted for. George Lakoff and Rafael Núñez's Where Mathematics Comes From (2000) locate mathematical cognition in embodied conceptual metaphor [29.05.], [20.06.*], offering a naturalistic account of how the mind generates mathematical structures, at the cost of explaining why those structures apply so successfully to a world the mind did not design.

Mathematical pluralism, defended by Colyvan in "The Ontological Commitments of Inconsistent Theories" and developed by Chris Mortensen's Inconsistent Mathematics (1995), treats inconsistent but useful mathematical theories (naive set theory, infinitesimal calculus before rigorisation) as ontologically on a par with consistent ones, formalised within paraconsistent logic [42.*]. The pluralist holds that there is no single privileged mathematical ontology; different theories carry different commitments, and the philosopher's task is to map the commitments rather than adjudicate a single winner. This connects to the neologicism of Crispin Wright and Bob Hale, who revive Frege's definition of number via abstraction principles (Hume's Principle) and treat numbers as abstract objects introduced by definition rather than discovered — a position that sits between Platonism and structuralism and that bears on the wider question of whether numbers are objects or concepts 20.09.02 pending.

Connections Master

Philosophy of mathematics: Platonism and constructivism 20.09.01. The direct prerequisite. 20.09.01 surveys the major positions at an introductory level; this unit deepens the four ontological positions — Platonism, nominalism, structuralism, fictionalism — and the arguments (Benacerraf, Quine-Putnam) that organise them. Read 20.09.01 for the map; read this unit for the arguments that drive commitment.
Foundations of mathematics 20.09.02 pending. The immediate neighbour. Foundations debates (logicism, formalism, intuitionism) generate the ontological questions this unit takes up: if numbers are not logical objects (logicism's collapse), not mere marks (formalism's loss), and not only mental constructions (intuitionism's restrictiveness), then what are they? Structuralism, fictionalism, and indispensability-based Platonism are the post-foundational answers. Category-theoretic foundations and neologicism appear in both units and are cross-referenced.
Scientific realism and the indispensability argument 20.08.02 pending. The indispensability argument is the principal bridge between philosophy of mathematics and philosophy of science. Scientific realism about electrons and the Quinean criterion of ontological commitment jointly imply realism about mathematical entities, if mathematics is indispensable to science. Structural realism 20.08.02 pending and the debate over idealisation and models 20.08.03 pending bear directly on whether the mathematics in our best theories should be read literally.
Causation and explanation 20.08.03 pending. The cicada argument, mathematical explanation of physical phenomena, and Lange's non-causal explanations all turn on what counts as an explanation in science. The explanatory-indispensability strengthening of the Quine-Putnam argument depends on accounts of scientific explanation developed here.
Space, time, and relativity 20.03.03 pending. Field's nominalisation of Newtonian gravitation, Malament's critique, and the claim that general relativity is essentially mathematical all turn on the substantivalism-relationalism debate about space-time 20.03.03 pending. Whether space-time points are concrete (and so available to the nominalist) or abstract is load-bearing for fictionalism.
Epistemology: knowledge, justification, and truth 20.01.01–20.01.03 pending. Benacerraf's access problem is an application of the causal theory of knowledge and the Gettier problem to mathematics. Responses to the access problem — Gödelian intuition, indispensability, Maddy's naturalism — are epistemological programmes evaluated in the framework of 20.01.02 pending.
Aesthetics [20.04.*]. Mathematical beauty (Hardy, Rota) and its role in guiding conjecture and axiom-selection connect ontology to the philosophy of aesthetics. The applicability puzzle (Wigner, Steiner) turns partly on whether aesthetic judgement tracks a mind-independent mathematical reality.
Cognition and embodied mathematics [29.05.], [29.06.]. Lakoff and Núñez's embodied account, mathematical cognition in children, and the epistemic role of diagrams 29.03.02 pending connect ontology to cognitive science and the psychology of mathematics.
Mathematical models in the sciences. The applicability question extends to specific domains: mathematical models in medicine [35.], climate models in earth science [27.], and the Lambda-CDM model in cosmology [28.04.*]. Each is a case study in whether the mathematics is indispensable to the science and therefore ontologically committing.
Foundations and formal logic [42.*]. Paraconsistent mathematics, category-theoretic foundations, homotopy type theory, and Tarski's semantic conception of truth are the formal substrates of several positions here, especially structuralism and pluralism.

Historical and philosophical context Master

Plato's theory of Forms, set out in the Republic, Phaedo, and Parmenides, is the origin of mathematical Platonism. The Forms are perfect, eternal, non-physical exemplars — the Form of Justice, the Form of the Good, the Form of Triangle — of which physical instances are imperfect copies. Mathematical objects sit naturally in this ontology: the perfect circle is a Form, and every drawn circle approximates it. Aristotle rejected separate Forms in favour of forms immanent in substances, and the medieval debate between realists and nominalists over universals (Roscelin, Abelard, Ockham) rehearsed the same structure in a theological key: do universals exist independently, or are they only names? The medieval nominalist victory in theology did not settle the mathematical case, because mathematics presented the realist with a stronger argument — the necessity and objectivity of arithmetical truth.

The nineteenth century transformed the question. Gottlob Frege's Die Grundlagen der Arithmetik (1884) analysed cardinal numbers as logical objects — the number of $F$ s is the extension of the concept "equinumerous with $F$ " — and set the programme of logicism that aimed to derive mathematics from logic alone. Russell's paradox (1902) refuted Frege's system, and the foundations crisis that followed (Zermelo's axiomatisation, Hilbert's programme, Brouwer's intuitionism) displaced ontology in favour of epistemology and consistency 20.09.02 pending. By the 1930s the ontological question had retreated behind the foundational one: mathematicians asked whether mathematics was consistent, complete, and reducible to logic, not whether its objects existed. Gödel's incompleteness theorems (1931) closed the foundational heroic phase, and Gödel himself read his results as evidence for Platonism — there are truths no formal system captures, so truth outruns provability and is not constituted by it.

The modern ontological debate begins with two papers. W. V. Quine's "On What There Is" (1948) introduced the criterion of ontological commitment — to be is to be the value of a bound variable — and dissolved the question of what exists into the question of what our best theories quantify over. Nelson Goodman and Quine's "Steps Toward a Constructive Nominalism" (1947) attempted, and confessed failure of, a reconstruction of mathematics without abstract objects. Paul Benacerraf's "What Numbers Could Not Be" (1965) and "Mathematical Truth" (1973) then reframed the debate around two problems — the identification problem and the access problem — that every subsequent position has had to address. Putnam's "What is Mathematical Truth?" (1975) and Philosophy of Logic (1971) joined Quine's criterion to the indispensability of mathematics in science, producing the argument that remains the strongest case for Platonism.

The period from 1980 to 2000 produced the major alternatives. Field's Science Without Numbers (1980) launched mathematical fictionalism and the conservativeness programme. Hellman's Mathematics Without Numbers (1989) developed modal structuralism. Shapiro's Philosophy of Mathematics: Structure and Ontology (1997) and Resnik's Mathematics as a Science of Patterns (1997) systematised ante rem structuralism. Maddy's Naturalism in Mathematics (1997) redirected attention to practice, and Balaguer's Platonism and Anti-Platonism in Mathematics (1998) argued the ontological question was, on present evidence, undecidable. The twenty-first century has turned to mathematical explanation (Baker 2005; Lange 2017), category-theoretic and type-theoretic foundations (Voevodsky, Awodey, McLarty), and the cognitive science of mathematics (Lakoff and Núñez 2000; Dehaene), pressing the ontological question from new directions without resolving the Benacerraf-Quine framework that organises it.

Bibliography Master

Plato — Republic, Books VI–VII (the Form of the Good, the divided line); Phaedo (Forms and recollection); Parmenides (self-predication and the third-man argument).
Frege, G. — Die Grundlagen der Arithmetik (Koebner, Breslau, 1884); English trans. J. L. Austin, The Foundations of Arithmetic (Blackwell, 1950).
Quine, W. V. — "On What There Is," Review of Metaphysics 2, 21–38 (1948).
Goodman, N. & Quine, W. V. — "Steps Toward a Constructive Nominalism," Journal of Symbolic Logic 12, 105–122 (1947).
Gödel, K. — "What is Cantor's Continuum Problem?" American Mathematical Monthly 54, 515–525 (1947); revised 1964 version in Benacerraf & Putnam (eds.), Philosophy of Mathematics, 2nd ed. (1983).
Benacerraf, P. — "What Numbers Could Not Be," Philosophical Review 74, 47–73 (1965).
Benacerraf, P. — "Mathematical Truth," Journal of Philosophy 70, 661–679 (1973).
Putnam, H. — "What is Mathematical Truth?" Historia Mathematica 2, 529–533 (1975); reprinted in Mathematics, Matter and Method (Cambridge, 1979).
Putnam, H. — Philosophy of Logic (Allen & Unwin, 1971).
Field, H. — Science Without Numbers: A Defence of Nominalism (Blackwell, 1980); 2nd ed. with new introduction (Oxford University Press, 2016).
Malament, D. — "Review of Field's Science Without Numbers," Journal of Philosophy 79, 523–534 (1982).
Hellman, G. — Mathematics Without Numbers: Towards a Modal-Structural Interpretation (Oxford University Press, 1989).
Shapiro, S. — Philosophy of Mathematics: Structure and Ontology (Oxford University Press, 1997).
Resnik, M. D. — Mathematics as a Science of Patterns (Oxford University Press, 1997).
Parsons, C. — Mathematical Thought and Its Objects (Cambridge University Press, 2008).
Maddy, P. — Naturalism in Mathematics (Oxford University Press, 1997).
Maddy, P. — Second Philosophy: A Naturalistic Method (Oxford University Press, 2007).
Balaguer, M. — Platonism and Anti-Platonism in Mathematics (Oxford University Press, 1998).
Colyvan, M. — The Indispensability of Mathematics (Oxford University Press, 2001).
Baker, A. — "Are There Genuine Mathematical Explanations of Physical Phenomena?" Mind 114, 223–238 (2005).
Saatsi, J. — "The Mathematics of the Impossible: The Explanation of the Honeycomb Conjecture," Foundations of Science 10, 145–161 (2005).
Wigner, E. — "The Unreasonable Effectiveness of Mathematics in the Natural Sciences," Communications on Pure and Applied Mathematics 13, 1–14 (1960).
Steiner, M. — The Applicability of Mathematics as a Philosophical Problem (Harvard University Press, 1998).
Pincock, C. — Mathematics and Scientific Representation (Oxford University Press, 2012).
Azzouni, J. — Tracking Reason: Proof, Discovery, and the Crisis of Foundational Mathematics (Oxford University Press, 2008).
Lange, M. — Because Without Cause: Non-Causal Explanations in Science and Mathematics (Oxford University Press, 2017).
Thurston, W. P. — "On Proof and Progress in Mathematics," Bulletin of the American Mathematical Society 30, 161–177 (1994).
Hersh, R. — What Is Mathematics, Really? (Oxford University Press, 1997).
Corfield, D. — Towards a Philosophy of Real Mathematics (Cambridge University Press, 2003).
Giaquinto, M. — Visual Thinking in Mathematics: An Epistemological Study (Oxford University Press, 2007).
Brown, J. R. — Philosophy of Mathematics: A Contemporary Introduction to the World of Proofs and Pictures, 2nd ed. (Routledge, 2008).
Wright, C. — Truth and Objectivity (Harvard University Press, 1992).
Halbach, V. — Axiomatic Theories of Truth (Cambridge University Press, revised ed. 2014).
Leng, M. — Mathematics and Reality (Oxford University Press, 2010).
Marquis, J.-P. — From a Geometrical Point of View: A Study of the History and Philosophy of Category Theory (Springer, 2009).
McLarty, C. — "What Does It Take to Prove Fermat's Last Theorem? Grothendieck and the Logic of Number Theory," Bulletin of Symbolic Logic 16, 359–377 (2010).
Landry, E. — "Structural Realism and Category Theory," Synthese 190, 1–18 (2013).
Mortensen, C. — Inconsistent Mathematics (Kluwer, 1995).
Chihara, C. S. — Constructibility and Mathematical Existence (Oxford University Press, 1990).
Bigelow, J. — The Reality of Numbers: A Physicalist's Philosophy of Mathematics (Oxford University Press, 1988).
Hardy, G. H. — A Mathematician's Apology (Cambridge University Press, 1940).
Rota, G.-C. — "The Phenomenology of Mathematical Beauty," in The Mathematical Experience (Birkhäuser, 1997).
Lakoff, G. & Núñez, R. E. — Where Mathematics Comes From: How the Embodied Mind Brings Mathematics into Being (Basic Books, 2000).
Shapiro, S. — Thinking About Mathematics: The Philosophy of Mathematics (Oxford University Press, 2000).
Benacerraf, P. & Putnam, H. (eds.) — Philosophy of Mathematics: Selected Readings, 2nd ed. (Cambridge University Press, 1983).