A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
The axiom of extensionality states that two sets are equal if and only if they have exactly the same members: ∀A ∀B (∀x (x ∈ A ↔ x ∈ B) → A = B). This axiom eliminates any notion of internal structure, ordering, or multiplicity — a set is determined entirely by its elements, nothing more. It is the identity criterion for sets and is presupposed by virtually every proof that two sets are equal: show mutual containment. Without extensionality, {1, 2} and {2, 1} could in principle be different objects; the axiom guarantees they are one and the same set.
Practice proving set equalities by double inclusion (A ⊆ B and B ⊆ A). Then consider what mathematics would look like without extensionality — multisets and sequences are structures that intentionally violate it, which clarifies what the axiom rules out. Compare sets with other collection-like objects (bags, lists, types) to see that extensionality is a genuine choice, not a tautology.
From your overview of ZFC, you know that set theory is built on a single primitive relation: membership (∈). Everything in ZFC — numbers, functions, ordered pairs, sequences — is ultimately defined in terms of which things belong to which sets. The Axiom of Extensionality is the rule that says what it means for two sets to be *the same set*: ∀A ∀B (∀x (x ∈ A ↔ x ∈ B) → A = B). Two sets are equal if and only if they have exactly the same members. Equivalently, a set is determined entirely by its members — nothing else about it matters.
This is a genuine substantive claim, not a definition or tautology. Consider the following objects in everyday mathematics: the set {1, 2, 3}, the sequence (1, 2, 3), the multiset ⟨1, 1, 2⟩, and the tuple (1, 2, 3). All of them "contain" the numbers 1, 2, 3 in some sense, but they are different kinds of objects because they encode additional structure — order and multiplicity. The axiom of extensionality asserts that *sets* carry none of this additional structure: {1, 2, 3} and {3, 1, 2} and {1, 2, 3, 2} (if we could write that) are all the same set, because they have exactly the same members. The axiom rules out any notion of "internal arrangement" or "how many times something appears" — membership is binary, and that is all.
The practical consequence is the double-inclusion proof technique. To prove A = B, show that every element of A is an element of B (A ⊆ B) and every element of B is an element of A (B ⊆ A). Extensionality guarantees that this is sufficient: if the membership conditions agree for all x, then the sets are equal. This pattern appears constantly in mathematical proofs — whenever you need to prove two sets are equal, you unfold their definitions and verify the same elements qualify on both sides. Without extensionality, this argument wouldn't work: you could have two distinct sets with identical members, like two different empty sets.
The axiom also resolves the question of the empty set: there is exactly one set with no members. If ∅₁ and ∅₂ were two empty sets, then vacuously every member of ∅₁ is a member of ∅₂ and vice versa (there are none to check), so extensionality gives ∅₁ = ∅₂. Extensionality thus guarantees that the empty set is unique — you don't need a separate axiom asserting uniqueness. This is characteristic of how extensionality works throughout ZFC: it collapses all set-theoretic constructions to a canonical form determined solely by membership, which is what lets ZFC serve as a foundation for mathematics where "same mathematical object" has an unambiguous meaning.