Ackermann Set Theory (A)

Origin. Wilhelm Ackermann, "Zur Axiomatik der Mengenlehre" (Mathematische Annalen 131, 1956). Azriel Lévy ("On Ackermann's set theory", 1959; Lévy–Vaught 1961) proved ZF ⊆ A* and showed a strong reflection principle holds in A*, then asked whether the converse holds. William Reinhardt answered it: "Ackermann's set theory equals ZF" (Annals of Mathematical Logic 2, 1970).

Models. Classes and sets in one sort, with membership between classes permitted, and a single schema in place of ZF's set-construction axioms. Ackermann's idea is that a class is a set when it can be defined without reference to the totality of sets — the universe is not a "sharply delimited" object, and anything specified without appeal to it is. This is a motivation from the unknowability of the absolute, not from size, and it is the feature that distinguishes A from NBG and Morse–Kelley, where classes cannot be members.

Formalism.

Language: ∈, and a constant V for the class of all sets. (Ackermann's original used a predicate symbol M(x) for "x is a set"; V is definable from it by comprehension.) "x is a set" abbreviates x ∈ V.

Axioms: A1. Extensionality: ∀x∀y (∀z(z ∈ x ↔ z ∈ y) → x = y) A2. Heredity: x ∈ V ∧ (y ∈ x ∨ y ⊆ x) → y ∈ V (elements and subclasses of sets are sets) A3. Class comprehension: ∃x ∀y (y ∈ x ↔ y ∈ V ∧ φ(y)), for every formula φ with x not free (any collection of sets is a class) A4. Ackermann's schema. For every ∈-formula φ(y, z₁,…,zₙ) in which V does not occur and which contains no free variables besides those shown: z₁,…,zₙ ∈ V ∧ ∀y (φ(y, z⃗) → y ∈ V) → ∃w ∈ V ∀y (y ∈ w ↔ φ(y, z⃗)) (a class defined without mentioning V, from set parameters, and containing only sets, is itself a set) A5. Foundation (optional; A* denotes A with the foundation schema).

What A4 replaces: Pairing, Union, Power set, Separation, Replacement, Infinity. Each is derivable. The condition "V does not occur in φ" is the entire content — it is what stops φ(y) := (y = y) from making V a set.

The equivalence (Lévy 1959, Reinhardt 1970): The theorems of A* about well-founded sets are exactly the theorems of ZF. Consequently: A* is consistent iff ZF is consistent, and A* proves nothing about sets that ZF does not. The proof direction that is Reinhardt's translates A*-proofs into ZFC-proofs via the Reflection Principle — one chooses a limit rank M with as many closure properties as the given proof requires, and interprets V as M. Lévy's direction shows ZF ⊆ A* and that a strong reflection principle is available in A*.

Where A falls short:* the strong replacement axiom of NBG is not provable in A* (Lévy). Reinhardt's A⁺ — A*'s language plus a second constant V′ with V ⊆ V′ — was introduced to repair this and initially looked stronger; it is equiconsistent with ZF (Reinhardt for the upper end via indescribability, then improved).

Relatives: A_∞, Ackermann's own extension with a class V_n of sets of type n for each natural n; the same equivalence holds with modifications.

Symbols.

Metatheory. Reinhardt's theorem is the entry's point and it is a deflation: a theory motivated by a genuinely different picture of what a set is — definable without reference to the absolute, rather than small — turns out to prove exactly ZF's theorems about well-founded sets. The mechanism is reflection: Ackermann's schema is a reflection principle in disguise, since "definable without V" is what reflection exploits, and ZF's reflection theorem is strong enough to simulate it. So A is not an alternative to ZF in strength or content but in presentation, and the interest is that the two motivations converge. Placement relative to Class Theories: NBG and MK make classes objects but forbid class membership and stratify by size; A permits classes to be members and stratifies by definability, so it is the third option in that space and the one that turns out to be ZF again.

Applies to. The reflection principle and its role as a generator of ZF's axioms — A is the cleanest statement that reflection is the content of Replacement and Power set. Structural reflection programmes in large cardinal theory, where Ackermann's schema is the ancestor. Alternative set theories, as the definability-based option alongside NF's stratification and GPK's polarity. The philosophy of the absolute infinite, which is Ackermann's stated motivation and which the schema formalizes.

Limitations. Proves nothing new: Reinhardt's theorem means A* is ZF with different axioms, so there is no mathematical reason to work in it. NBG's strong replacement fails in A*, which was long taken as a defect and required Reinhardt's A⁺ to repair — and A⁺ is equiconsistent with ZF too. The restriction "V does not occur in φ" in A4 is syntactic and brittle: the schema's content depends entirely on it, and small variations produce inconsistency or triviality. Overlaps Class Theories; it is here because its individuation is a schema over ∈ and V, not a two-sorted apparatus.

© 2026 Lingenic LLC