Abstract Objects & Platonism

What an abstract object is

The standard diagnostic comes in three negations. An abstract object is non-spatial (not located anywhere), non-temporal (not in time, has no age), and causally inert (does not push or pull anything). The number 7 satisfies all three. Your dog satisfies none. Universals like redness, sets, propositions, types of words, fictional characters, possible worlds, and pure geometric shapes are usually counted in.

The point of the diagnostic is to mark off a kind of entity whose existence cannot be settled by perception. You can find your dog by looking; you cannot find the number 7 by looking. So if abstract objects exist, the way we know about them must be very different from the way we know about ordinary things. That epistemological puzzle is what makes platonism philosophically interesting and, to many, suspicious.

From Plato to Frege

Plato launched the tradition with the theory of Forms: each kind of thing in the sensible world participates in an eternal, perfect, non-spatial Form. The triangles you draw approximate the Form of Triangle; the just acts in our city instantiate the Form of Justice. Plato's argument was partly mathematical (the geometers' theorems are about something more exact than any drawn figure) and partly ethical (moral knowledge requires invariant standards).

The modern revival is mostly Frege. The Foundations of Arithmetic (1884) argued that numbers cannot be physical marks, mental images, or properties of heaps — they must be objective entities of a third realm, knowable through reason. Frege's logicism aimed to derive arithmetic from pure logic plus the existence of certain abstract objects (extensions of concepts). Russell's paradox cracked Frege's specific proposal in 1902, but the platonist intuition survived. Gödel made it explicit in his 1947 essay on Cantor's continuum hypothesis: mathematical objects are "perceived" by a faculty of intuition analogous to sense perception.

The indispensability argument

The strongest contemporary case for platonism does not rest on intuition. Quine, refined by Putnam, argued as follows.

1. Ontological commitment. A theory is committed to the entities its quantifiers range over. To accept "there are prime numbers greater than ten" is to accept primes.
2. Naturalism. Our best science is the standard for what to believe exists. There is no first-philosophy court of appeal beyond science.
3. Indispensability. Our best scientific theories — physics, biology, economics — quantify over numbers, functions, and other mathematical entities, and we cannot reformulate them without doing so.
4. Conclusion. By naturalism plus ontological commitment, we ought to believe that mathematical entities exist.

The argument flips the burden of proof. The platonist used to be on the back foot, asked to justify a strange ontology. After Quine, the nominalist is on the back foot, asked to show that science can be done without the supposedly strange ontology. That is exactly what Hartry Field tried, and the response defines a generation of philosophy of mathematics.

Platonism vs its rivals

	Platonism	Nominalism (Field)	Fictionalism	Structuralism	Aristotelian realism
Numbers exist?	Yes, abstract	No	No (useful fiction)	Yes, as positions	Yes, in instances only
Math sentences true?	Literally	Vacuously / false	True-in-the-fiction	About patterns	Literally, when instantiated
Causally inert objects?	Required	Forbidden	None to be inert	Required	Avoided
Handles Benacerraf access?	With effort	Trivially	Trivially	Better than classical platonism	Yes — sense perception suffices
Indispensability rebuttal?	Embraces it	Rewrite physics	Pretence is enough	Patterns suffice	Instances suffice
Famous proponent	Frege, Gödel, Quine	Field, Hellman	Yablo, Balaguer	Resnik, Shapiro	Aristotle, Armstrong
Cost	Heavy ontology	Heavy revision of science	Heavy revision of semantics	Strong modal commitments	Limits to applied math

No row is dominant; each view trades a different problem for a different one. The dialectic since the 1980s is largely about which trade-off is least bad.

Worked example: do prime numbers exist?

Suppose we ask "are there infinitely many prime numbers?" The mathematician answers yes — Euclid proved it around 300 BCE. The philosophical question is what we are doing when we say so.

Platonist reading. The sentence is literally true. There is a structure called the natural numbers, every natural number is either prime or composite, and the primes form an infinite subset. Euclid's proof is a discovery about a mind-independent realm. The infinitude of primes was a fact long before Euclid existed.

Fictionalist reading. The sentence is true within the standard arithmetical fiction, just as "Sherlock Holmes lived at 221B Baker Street" is true within the Conan Doyle fiction. We can use the fiction to make true predictions about physical systems without believing it literally describes them. Mary Leng and Stephen Yablo have developed this in detail.

Nominalist reading. Strictly speaking the sentence is false (or vacuous), since there are no numbers to be prime. But there is a nominalistically acceptable reformulation in terms of, say, abstractable equivalence classes of physical configurations or modal claims about what counting procedures could yield.

Structuralist reading. The sentence is true and is about a pattern — the natural-number structure — instantiated in any sufficiently rich progression of objects. The primes are positions in that pattern, not particular entities.

Notice that all four parties agree on every observable consequence. The disagreement is metaphysical. That is what makes the question hard and what gives Carnap-style sceptics their opening.

Major objections to platonism

• Benacerraf's access problem (1973). If abstract objects are causally inert, no causal chain runs from them to our brains. How can we have justified beliefs about them? Reliable belief requires some explanation of why our beliefs co-vary with the truths, and platonism seems to owe one.
• Benacerraf's identification problem (1965). Is the number 2 the set {{∅}} (Zermelo) or {∅, {∅}} (von Neumann)? Both reductions work. If neither is privileged, perhaps numbers are not objects at all but positions in a structure.
• Carnap's anti-metaphysics. External questions about whether numbers "really" exist are ill-formed; only internal questions within a mathematical framework are meaningful. Quine rejected this distinction; neo-Carnapians revive it.
• Ockham's razor. A non-physical realm is a heavy ontological commitment. If a nominalist or fictionalist account does the same explanatory work with fewer kinds of entity, parsimony favours it.
• Field's reformulation challenge. If physics can be done in nominalist terms (Field 1980 attempts Newtonian gravitation), the indispensability premise fails and the chief argument for platonism collapses.
• Easy ontology pushback. Amie Thomasson and others argue that "there is a prime number" follows trivially from "there is a number divisible only by one and itself." If platonism is this easy, it cannot be the substantive metaphysical thesis it is sold as.

Variants of platonism

• Plenitudinous platonism. Mark Balaguer: every consistent mathematical structure exists. Solves the access problem because, given so many structures, our beliefs cannot help but match some of them.
• Structuralism (ante rem). Stewart Shapiro: mathematical objects are positions in abstract structures. Mitigates Benacerraf's identification problem.
• Neo-Fregean / neo-logicism. Crispin Wright and Bob Hale: arithmetic can be derived from second-order logic plus Hume's Principle, securing the existence of numbers without Frege's paradoxical comprehension.
• Modal structuralism. Geoffrey Hellman: mathematical claims are translatable into modal claims about possible structures, removing the need for actually existing abstracta.
• Universals platonism (beyond math). Properties and relations exist outside their instances. The redness of this apple is a particular instance of the universal Redness, which exists whether or not any apple does.

Common confusions

• Platonism does not require Plato's Forms. Modern platonism is closer to Frege's "third realm" than to Republic-style ethics-and-politics metaphysics.
• "Mathematical realism" is broader than platonism. Aristotelians and structuralists can be realists without being platonists.
• Abstract does not mean "vague" or "general." The number 7 is precise and particular; "abstract" is a technical term for non-spatial, non-temporal, causally inert.
• Indispensability is empirical, not a priori. If physics could be reformulated, the argument loses force; this is why Field's project matters even if you find it unconvincing.
• Fictionalism is not anti-realism about science. A fictionalist about math can be a full realist about electrons; the fiction is the mathematical apparatus, not the empirical predictions it lets you make.
• Platonism does not entail any particular set theory. Whether ZFC, ZF + large cardinals, or something else describes "the universe of sets" is internal to mathematics; platonism just says there is a fact of the matter.