A Choice That Seems Obvious
Suppose you have a collection of boxes, each containing at least one ball. Can you choose one ball from each box?
If there are finitely many boxes: obviously yes. Open each one, pick a ball, done.
If there are infinitely many boxes — say, one box for each natural number — can you still choose one ball from each? Intuitively, yes. Just keep choosing.
But “just keep choosing” is not a mathematical proof. A proof requires an explicit rule, or a guarantee that such a rule exists. And it turns out that for some infinite collections, no explicit rule can be given — yet mathematicians want to talk about choosing anyway.
The axiom of choice provides that guarantee. It states: for any collection of non-empty sets, there exists a choice function — a function that picks exactly one element from each set. No rule required. The function simply exists.
This sounds harmless. It is, in some sense, the most powerful and most troublesome assumption in modern mathematics.
Why We Need It: The Sock Problem
Bertrand Russell illustrated the need for the axiom with a vivid example. Suppose you have infinitely many pairs of shoes. Choosing one shoe from each pair is easy: always pick the left shoe. There is an explicit rule.
Now suppose you have infinitely many pairs of socks — identical, indistinguishable within each pair. Now there is no rule. Left and right are meaningless. To choose one sock from each pair, you need the axiom of choice: the guarantee that a selection exists, even without a recipe for making it.
This is the heart of the matter. The axiom of choice is needed precisely when no explicit selection rule exists but you want to assert that a selection is possible anyway.
The Axioms Behind Modern Mathematics
Modern mathematics rests on a foundation called ZFC — the Zermelo-Fraenkel axioms together with the Axiom of Choice. The nine Zermelo-Fraenkel axioms (ZF) specify the basic properties of sets: what counts as a set, when two sets are equal, how sets can be formed from other sets.
The ZF axioms alone are sufficient for most of everyday mathematics. But certain important theorems — some central, some surprising — require the additional power of the axiom of choice.
Theorems that require the axiom of choice (or cannot be proved without it):
- Every vector space has a basis
- The product of any collection of non-empty compact topological spaces is compact (Tychonoff’s theorem)
- Every surjective function has a right inverse
- Every infinite set contains a countably infinite subset
- The real numbers can be well-ordered
- There exist non-measurable subsets of the real line
That last item is the door to the Banach-Tarski paradox.
Non-Measurable Sets: Where Intuition Breaks
A measurable set is one to which we can assign a well-defined “size” (measure) consistent with our intuitions. The interval [0, 1] has measure 1. The set of rational numbers has measure 0. Nice sets have measures; pathological sets may not.
In 1905, Giuseppe Vitali used the axiom of choice to construct the first known non-measurable set — a subset of the real line to which no consistent notion of length can be assigned.
The construction: partition the interval [0, 1] into equivalence classes, where two numbers are equivalent if their difference is rational. Each class is a translation of the rationals — a countably infinite, dense set. Now use the axiom of choice to pick exactly one representative from each class. The resulting set V cannot be assigned a measure. If it had one, contradictions follow immediately from basic properties of measure.
Vitali’s set exists only because the axiom of choice guarantees we can pick one representative from each of uncountably many equivalence classes — with no rule for which to pick.
The Banach-Tarski Paradox
In 1924, Stefan Banach and Alfred Tarski proved one of the most startling theorems in mathematics:
A solid ball in ℝ³ can be decomposed into a finite number of disjoint pieces that can be reassembled into two solid balls, each the same size as the original.
Five pieces suffice. The reassembly uses only rigid motions — rotations and translations, no stretching.
This is not a physics claim. The “pieces” are non-measurable sets — geometrically intricate, infinitely complicated objects that cannot be physically realized. They have no well-defined volume. The paradox does not violate conservation of mass because these pieces have no mass in any coherent sense.
The proof uses the axiom of choice in a fundamental way: it selects representatives from infinitely many orbits of a group acting on the sphere, producing the non-measurable pieces. Without the axiom of choice, the Banach-Tarski decomposition cannot be performed.
The paradox is a theorem, not an error. Its lesson: the axiom of choice allows mathematical objects to exist that have no physical interpretation and behave in deeply counterintuitive ways.
Equivalent Formulations: A Zoo of Strange Theorems
The axiom of choice is equivalent — within ZF — to a remarkable number of seemingly different statements. Proving any one of them is equivalent to proving the axiom of choice itself.
Well-ordering theorem (Zermelo, 1904): Every set can be well-ordered — arranged in a sequence where every non-empty subset has a least element. For the natural numbers this is obvious (their natural order works). For the real numbers it is deeply non-constructive: no one can write down an explicit well-ordering of ℝ, but the axiom of choice guarantees one exists.
Zorn’s lemma: If every chain (totally ordered subset) in a partially ordered set has an upper bound, then the set has at least one maximal element. This formulation is the one most used in practice — it appears in proofs across algebra, topology, and analysis. Every vector space has a basis (proved using Zorn’s lemma). Every ring has a maximal ideal. Every field has an algebraic closure.
Tychonoff’s theorem: The product of any collection of compact topological spaces is compact. For finite products this follows from ZF alone. For infinite products — even uncountably infinite — the axiom of choice is necessary.
These equivalences were not obvious. Proving that the well-ordering theorem implies the axiom of choice, or that Zorn’s lemma implies the well-ordering theorem, requires real mathematical work.
The Independence Result: Gödel and Cohen
For decades after Zermelo formulated the axiom of choice in 1904, mathematicians asked: is it provable from the other axioms? Necessary? Optional?
Kurt Gödel, 1938: Constructed the “constructible universe” L — a model of set theory in which every set is built explicitly from simpler sets in a definite transfinite process. In L, the axiom of choice holds. This showed that the axiom of choice is consistent with ZF: you can add it without introducing contradictions (assuming ZF itself is consistent).
Paul Cohen, 1963: Invented the technique of forcing — a method for constructing new models of set theory by adding “generic” elements. Using forcing, Cohen built a model of ZF in which the axiom of choice fails. This showed the axiom of choice is independent of ZF: it cannot be proved from the ZF axioms.
Together, Gödel and Cohen established that the axiom of choice is genuinely optional. ZF with choice (ZFC) and ZF without choice (ZF¬C) are both consistent, and they describe genuinely different mathematical universes.
Cohen’s forcing technique, developed to resolve this question, became one of the most important tools in modern set theory and won him the Fields Medal in 1966 — the only Fields Medal ever awarded for a result in mathematical logic.
Life Without the Axiom of Choice
What happens if we reject the axiom of choice? Mathematics becomes more complicated — and in some ways more honest.
Without choice:
- There may exist vector spaces with no basis
- The real numbers may not be well-orderable
- The countable union of countable sets may not be countable
- Tychonoff’s theorem fails for infinite products
- Non-measurable sets may not exist (the Banach-Tarski paradox disappears)
Some mathematicians — constructivists and intuitionists — embrace this. For them, asserting the existence of an object without constructing it is not mathematics. The axiom of choice is the paradigm of non-constructive reasoning: it guarantees a choice function exists without providing any recipe for it.
The axiom of determinacy (AD) is an alternative to the axiom of choice that some set theorists study. It states that every infinite two-player game of perfect information is determined — one player or the other has a winning strategy. AD is inconsistent with full choice but consistent with a weakened form. Under AD, all subsets of the real line are measurable — the Banach-Tarski paradox is impossible. The mathematical landscape is different, arguably cleaner, but also more constrained.
Why Most Mathematicians Accept It
Despite the controversy, the vast majority of working mathematicians use the axiom of choice freely. The reasons are pragmatic and aesthetic:
It simplifies proofs. Enormous swaths of algebra, topology, and analysis rest on Zorn’s lemma or equivalent tools. Replacing these with constructive arguments would require rebuilding large parts of mathematics from scratch, with vastly more complexity.
Its consequences are mostly benign. The Banach-Tarski paradox is strange, but it involves non-measurable sets that play no role in physics or most of mathematics. The useful consequences — every vector space has a basis, products of compact spaces are compact — are essential.
The alternative is worse. Without choice, mathematics loses elegance and simplicity. Many natural statements become false or unprovable. The cure seems worse than the disease.
The position of most mathematicians is essentially pragmatic: the axiom of choice is a useful tool whose paradoxical consequences are confined to a mathematical corner most of us never visit.
Conclusion: An Assumption Worth Knowing
The axiom of choice is present in almost every corner of modern mathematics, usually invisibly. When a textbook says “let B be a basis for this vector space,” it is invoking the axiom of choice. When a topology proof uses Tychonoff’s theorem, it is invoking the axiom of choice. When an analyst constructs a non-measurable set, same story.
Understanding it matters not because you will encounter it in practice — you may not — but because it reveals something deep about the nature of mathematical reasoning. Mathematics does not simply describe a fixed, given reality. It is built on axioms — assumptions we choose to make. The axiom of choice is the most consequential choice among those choices.
And unlike the party problem or the Pythagorean theorem, it is a choice no proof can force you to make.
Related topics: Set theory, Gödel’s incompleteness theorems, Banach-Tarski paradox, measure theory, foundations of mathematics
Frequently asked
What does the axiom of choice actually say?
Given any collection of non-empty sets, it is possible to choose exactly one element from each set simultaneously — even if there are infinitely many sets and no rule specifying which element to pick. For finite collections this is trivially true. For infinite collections it is an additional assumption that cannot be proved from the other axioms of set theory.
Why is the axiom of choice controversial?
It asserts the existence of objects that may not be constructible — you cannot always give an explicit rule for which element to choose, yet the axiom guarantees a choice exists. Some mathematicians find non-constructive existence proofs philosophically unsatisfying. The axiom also implies results that seem paradoxical, like the Banach-Tarski paradox. Constructivists and intuitionists reject it entirely.
What is the Banach-Tarski paradox?
The Banach-Tarski paradox states that a solid ball in three-dimensional space can be decomposed into a finite number of pieces (five suffice) and reassembled — using only rotations and translations — into two balls of the same size as the original. This is not physically possible: the 'pieces' are non-measurable sets with no well-defined volume, and the decomposition requires the axiom of choice. It is a theorem, not an error, but it reveals how strange non-measurable sets can be.
Is the axiom of choice true?
This question has no objective answer in standard mathematics. Kurt Gödel showed in 1938 that the axiom of choice is consistent with the other axioms of set theory (ZF): assuming it introduces no contradiction. Paul Cohen showed in 1963 that its negation is also consistent. The axiom is independent of ZF — neither provable nor disprovable. Most working mathematicians accept it because it simplifies proofs and its denial leads to a more complicated, less elegant theory.