Axioms and Proof: The Rules of Every Game We Play
Introduction
An axiom is a statement accepted without proof. Everything else in a formal system is derived from axioms through logical rules. This much is uncontroversial. The interesting question is what an axiom is.
Is an axiom a self-evident truth, something so obviously correct that denying it would be absurd? Is it a discovered feature of reality, a law of the universe that humans have identified? Or is it simply a rule of a game, a starting condition chosen by humans because it produces interesting results?
This article takes the question seriously. It catalogs the major axiom systems of mathematics with their complete axioms, works through several proofs to show how theorems emerge from axioms, and then confronts the philosophical question directly: what are we doing when we axiomatize?
Part One: The Major Axiom Systems
1. The Peano Axioms (Arithmetic)
Formulated by Giuseppe Peano in 1889, these axioms define the natural numbers. The system uses two primitive notions: a constant 0 and a function S (successor).
0 is a natural number.
For every natural number n, S(n) is a natural number.
There is no natural number n such that S(n) = 0.
For all natural numbers m and n, if S(m) = S(n), then m = n.
If a property P holds for 0, and if for every natural number n, P(n) implies P(S(n)), then P holds for all natural numbers.
Addition and multiplication are not axioms. They are defined recursively from successor:
$a + 0 = a$
$a + S(b) = S(a + b)$
$a \times 0 = 0$
$a \times S(b) = a + (a \times b)$
Define: $1 = S(0)$, $2 = S(1) = S(S(0))$, $3 = S(2)$, $4 = S(3)$.
$2 + 2$
$= 2 + S(1)$ [since $2 = S(1)$]
$= S(2 + 1)$ [by $a + S(b) = S(a + b)$]
$= S(2 + S(0))$ [since $1 = S(0)$]
$= S(S(2 + 0))$ [by $a + S(b) = S(a + b)$]
$= S(S(2))$ [by $a + 0 = a$]
$= S(3)$ [since $3 = S(2)$]
$= 4$ [since $4 = S(3)$] ■
By induction on $b$.
Base case ($b = 0$): We need $a + 0 = 0 + a$.
$a + 0 = a$ by definition. We must show $0 + a = a$.
This requires its own induction on $a$:
Base: $0 + 0 = 0$. ✓
Step: Assume $0 + n = n$. Then $0 + S(n) = S(0 + n) = S(n)$. ✓
So $0 + a = a$ for all $a$. Therefore $a + 0 = 0 + a$. ✓
Inductive step: Assume $a + b = b + a$ for all $a$. Show $a + S(b) = S(b) + a$.
$a + S(b) = S(a + b)$ by definition.
$S(a + b) = S(b + a)$ by the inductive hypothesis.
We need $S(b + a) = S(b) + a$. This requires a lemma: $S(n) + a = S(n + a)$.
(The lemma itself is proved by induction on $a$.)
With the lemma: $S(b) + a = S(b + a) = S(a + b) = a + S(b)$. ✓
Commutativity holds for all natural numbers. ■
2. Euclid's Postulates (Geometry)
Written around 300 BCE. Euclid also listed five "common notions" (axioms of equality) alongside his five geometric postulates.
Common Notions:
Things equal to the same thing are equal to each other.
If equals are added to equals, the results are equal.
If equals are subtracted from equals, the remainders are equal.
Things that coincide with one another are equal to one another.
The whole is greater than the part.
Postulates:
A straight line can be drawn from any point to any other point.
A finite straight line can be extended continuously in a straight line.
A circle can be described with any center and any radius.
All right angles are equal to one another.
If a straight line falling on two straight lines makes the interior angles on the same side less than two right angles, the two straight lines, if produced indefinitely, meet on that side on which the angles are less than two right angles.
Given: A line segment AB.
Construction:
1. Draw circle C1 with center A and radius AB. [Postulate 3]
2. Draw circle C2 with center B and radius BA. [Postulate 3]
3. Let C be a point where C1 and C2 intersect.
4. Draw line segments AC and BC. [Postulate 1]
Proof that triangle ABC is equilateral:
AC = AB (both are radii of C1). [Definition of circle]
BC = BA (both are radii of C2). [Definition of circle]
AB = BA. [CN4, a line segment coincides with itself]
Therefore AC = AB = BC. [CN1, things equal to the same thing are equal]
Triangle ABC is equilateral. ■
3. Hilbert's Axioms (Modern Geometry)
In 1899, David Hilbert published Grundlagen der Geometrie, replacing Euclid's system with a rigorous axiomatization that closed all the implicit gaps. Hilbert's system uses three undefined primitives (points, lines, planes) and 20 axioms organized into five groups.
I.1. Two distinct points determine a unique line.
I.2. Any two distinct points on a line determine that line.
I.3. Every line contains at least two points. There exist at least three points not on the same line.
I.4. Three non-collinear points determine a unique plane.
I.5. Any three non-collinear points of a plane determine that plane.
I.6. If two points of a line lie in a plane, then every point of that line lies in the plane.
I.7. If two planes have a point in common, they have at least one more point in common.
I.8. There exist at least four points not all lying in the same plane.
II.1. If point B lies between points A and C, then A, B, C are distinct points on the same line, and B also lies between C and A.
II.2. For any two distinct points A and C, there exists a point B on line AC such that C lies between A and B.
II.3. Of any three points on a line, exactly one lies between the other two.
II.4. (Pasch's Axiom) If a line enters a triangle through one side, it must exit through one of the other two sides.
III.1. A line segment can be laid off from any point on any ray, and the result is unique.
III.2. Segment congruence is transitive.
III.3. Segment addition: if $AB \cong A'B'$ and $BC \cong B'C'$ and B is between A and C and B' is between A' and C', then $AC \cong A'C'$.
III.4. An angle can be laid off from any ray on either side, and the result is unique.
III.5. (SAS) If two sides and the included angle of one triangle are congruent to two sides and the included angle of another, the triangles are congruent.
IV.1. Through a point not on a given line, at most one line can be drawn parallel to the given line.
V.1. (Axiom of Archimedes) Given segments AB and CD, there exists a natural number n such that laying off CD on the ray from A, n times, passes beyond B.
V.2. (Axiom of Completeness) The points on a line cannot be extended to a larger system while preserving all the above axioms.
When Hilbert was asked to clarify what "points," "lines," and "planes" meant, he reportedly said: "One must be able to say at all times, instead of points, straight lines, and planes, tables, chairs, and beer mugs." The axioms define the relationships between the primitives. The primitives themselves are empty labels.
4. Zermelo-Fraenkel Set Theory with Choice (ZFC)
The standard foundation of modern mathematics. All mathematical objects (numbers, functions, spaces, structures) can be constructed as sets within ZFC. The system uses one primitive relation: $\in$ (membership).
$\forall A\, \forall B\, (\forall x\, (x \in A \leftrightarrow x \in B) \to A = B)$
Two sets are equal if and only if they have exactly the same elements.
$\exists A\, \forall x\, (x \notin A)$
There exists a set with no elements. We call it $\emptyset$.
$\forall a\, \forall b\, \exists C\, \forall x\, (x \in C \leftrightarrow x = a \lor x = b)$
For any two sets a and b, the set {a, b} exists.
$\forall A\, \exists B\, \forall x\, (x \in B \leftrightarrow \exists C\, (C \in A \land x \in C))$
For any set of sets $A$, the union $\bigcup A$ exists.
$\forall A\, \exists B\, \forall x\, (x \in B \leftrightarrow x \subseteq A)$
For any set A, the set of all subsets of A exists.
$\exists A\, (\emptyset \in A \land \forall x\, (x \in A \to x \cup \{x\} \in A))$
There exists an infinite set. Specifically, a set containing $\emptyset$, $\{\emptyset\}$, $\{\emptyset, \{\emptyset\}\}$, $\{\emptyset, \{\emptyset\}, \{\emptyset, \{\emptyset\}\}\}$, ... This is von Neumann's construction of the natural numbers from the empty set.
$\forall A\, \exists B\, \forall x\, (x \in B \leftrightarrow x \in A \land \phi(x))$
For any set $A$ and any property $\phi$, the set of elements of $A$ satisfying $\phi$ exists.
If $F$ is a function-class and $A$ is a set, then $F[A] = \{F(x) : x \in A\}$ is a set.
Applying a definable function to every element of a set produces a set.
$\forall A\, (A \neq \emptyset \to \exists x\, (x \in A \land x \cap A = \emptyset))$
Every non-empty set contains an element disjoint from itself.
For any collection of non-empty sets, there exists a function that selects exactly one element from each set.
Define: $0 = \emptyset$
$1 = \{\emptyset\} = \{0\}$
$2 = \{\emptyset, \{\emptyset\}\} = \{0, 1\}$
$3 = \{\emptyset, \{\emptyset\}, \{\emptyset, \{\emptyset\}\}\} = \{0, 1, 2\}$
In general: $n + 1 = n \cup \{n\}$
$\emptyset$ exists by ZF2. $\{\emptyset\}$ exists by ZF3 (pair $\emptyset$ with itself). $\{\emptyset, \{\emptyset\}\}$ exists by ZF3. The infinite set of all of them exists by ZF6.
Every natural number is literally the set of all smaller natural numbers. $3 = \{0, 1, 2\}$. This means $2 \in 3$ (2 is an element of 3) and $2 \subset 3$ (2 is a subset of 3). Membership and ordering coincide. The less-than relation is just the membership relation. ■
5. The Axioms of Probability (Kolmogorov, 1933)
Andrey Kolmogorov axiomatized probability theory, placing it on the same rigorous footing as the rest of mathematics. The system uses three primitives: a sample space $\Omega$ (the set of all possible outcomes), a sigma-algebra $\mathcal{F}$ (the collection of events we can assign probabilities to), and a probability function $P$.
For every event $A$ in $\mathcal{F}$, $P(A) \geq 0$.
$P(\Omega) = 1$.
For any countable sequence of mutually exclusive events $A_1, A_2, A_3, \ldots$
$P(A_1 \cup A_2 \cup A_3 \cup \ldots) = P(A_1) + P(A_2) + P(A_3) + \ldots$
From these three axioms, every result in probability theory follows: Bayes' theorem, the law of large numbers, the central limit theorem, conditional probability, random variables, distributions, stochastic processes. Three axioms generate an entire branch of mathematics.
$\Omega = \Omega \cup \emptyset \cup \emptyset \cup \ldots$ [$\Omega$ union with infinitely many empty sets]
These are mutually exclusive (the empty set is disjoint from everything).
$P(\Omega) = P(\Omega) + P(\emptyset) + P(\emptyset) + \ldots$ [by K3]
$1 = 1 + P(\emptyset) + P(\emptyset) + \ldots$ [by K2]
$0 = P(\emptyset) + P(\emptyset) + \ldots$
Since $P(\emptyset) \geq 0$ by K1, and an infinite sum of non-negative terms equals 0,
each term must be 0. Therefore $P(\emptyset) = 0$. ■
6. The Field Axioms (Algebra)
A field is the algebraic structure that captures the arithmetic of the rational numbers, real numbers, and complex numbers. It has two operations (addition and multiplication) satisfying these axioms. Let $F$ be a set with elements $a, b, c$ and operations $+$ and $\times$.
F1. Closure: $a + b \in F$
F2. Associativity: $(a + b) + c = a + (b + c)$
F3. Identity: There exists $0 \in F$ such that $a + 0 = a$
F4. Inverses: For every $a$, there exists $-a$ such that $a + (-a) = 0$
F5. Commutativity: $a + b = b + a$
F6. Closure: $a \times b \in F$
F7. Associativity: $(a \times b) \times c = a \times (b \times c)$
F8. Identity: There exists $1 \in F$ (with $1 \neq 0$) such that $a \times 1 = a$
F9. Inverses: For every $a \neq 0$, there exists $a^{-1}$ such that $a \times a^{-1} = 1$
F10. Commutativity: $a \times b = b \times a$
F11. Distributivity: $a \times (b + c) = (a \times b) + (a \times c)$
$a \times 0 = a \times (0 + 0)$ [F3: $0 + 0 = 0$]
$= (a \times 0) + (a \times 0)$ [F11: distributivity]
Now add $-(a \times 0)$ to both sides:
$a \times 0 + (-(a \times 0)) = (a \times 0) + (a \times 0) + (-(a \times 0))$
$0 = a \times 0 + 0$ [F4: additive inverse]
$0 = a \times 0$ [F3: additive identity] ■
7. The Axioms of Group Theory
A group is the simplest algebraic structure: a set $G$ with a single operation $\cdot$ satisfying four axioms.
G1. Closure: For all $a, b \in G$, $a \cdot b \in G$.
G2. Associativity: $(a \cdot b) \cdot c = a \cdot (b \cdot c)$.
G3. Identity: There exists $e \in G$ such that $e \cdot a = a \cdot e = a$ for all $a$.
G4. Inverses: For every $a \in G$, there exists $a^{-1} \in G$ such that $a \cdot a^{-1} = a^{-1} \cdot a = e$.
Groups are everywhere. The integers under addition form a group (identity: $0$, inverse of $n$: $-n$). The nonzero rationals under multiplication form a group (identity: $1$, inverse of $a$: $1/a$). Symmetries of a geometric figure form a group. Permutations of a set form a group. The concept is so general that it captures the abstract structure of symmetry itself.
Suppose $e$ and $e'$ are both identity elements of $G$.
Then $e \cdot e' = e'$ [because $e$ is an identity: $e \cdot x = x$]
And $e \cdot e' = e$ [because $e'$ is an identity: $x \cdot e' = x$]
Therefore $e = e \cdot e' = e'$. ■
8. Propositional Logic
Logic itself rests on axioms. One standard axiomatization of propositional logic (the Lukasiewicz system) uses three axiom schemas and one rule of inference.
L1. $\phi \to (\psi \to \phi)$
L2. $(\phi \to (\psi \to \chi)) \to ((\phi \to \psi) \to (\phi \to \chi))$
L3. $(\neg\phi \to \neg\psi) \to (\psi \to \phi)$
From $\phi$ and $\phi \to \psi$, conclude $\psi$.
From these three schemas and one rule, every tautology of propositional logic can be derived. The entire apparatus of logical reasoning reduces to three patterns and one operation.
9. Non-Euclidean Geometries
Replace Euclid's fifth postulate and everything changes.
Replace Postulate 5 with: Through a point not on a given line, there exist at least two lines parallel to the given line.
Replace Postulate 5 with: Through a point not on a given line, there exist no lines parallel to the given line.
All three geometries are internally consistent. None is more "true" than the others in a mathematical sense. The physical universe, as it turns out, uses all three: flat (Euclidean) spacetime far from massive objects, positively curved spacetime near them (general relativity), and the geometry of the universe as a whole remains an open empirical question (current measurements suggest it is very close to flat).
Part Two: What Is an Axiom?
The Self-Evidence View
"We hold these truths to be self-evident."
The oldest view of axioms, going back to Aristotle, is that they are self-evident truths. An axiom is a statement so obviously correct that no rational person could deny it. Euclid's first four postulates have this character. Of course a line can be drawn between two points. Of course a circle can be drawn with any center and radius. These feel less like assumptions and more like descriptions of how space necessarily works.
The self-evidence view was dominant for over two thousand years. It aligned with the Platonic idea that mathematical objects exist independently of human minds, in an abstract realm of perfect forms. Axioms, in this view, are not chosen. They are perceived. The mathematician does not invent axioms any more than the astronomer invents stars. The axioms are there, and the mathematician's job is to see them clearly.
This view has a problem. Euclid's fifth postulate does not feel self-evident. It is complex, verbose, and its truth is not immediately obvious. For two thousand years, mathematicians tried to derive it from the other four, believing it could not be a genuine axiom because it lacked the self-evidence that axioms were supposed to have. When Lobachevsky and Bolyai showed that denying the fifth postulate produced a consistent alternative geometry, the self-evidence view suffered a serious blow. If two contradictory statements (the fifth postulate and its negation) both produce consistent mathematics, then at most one can be a truth about reality, and neither can be self-evident in the way Aristotle meant.
A deeper problem: the Axiom of Choice. It is independent of the other ZFC axioms. You can accept it or reject it, and both choices give consistent mathematics. Accepting it gives you the Banach-Tarski paradox (a solid ball can be decomposed into finitely many pieces and reassembled into two balls of the same size). Rejecting it means that not every vector space has a basis. Neither option is "self-evident." The choice is pragmatic, not perceptual.
The Formalist View
David Hilbert proposed a radically different view: axioms are rules of a game. They have no inherent meaning. They are not "true" or "false" in any metaphysical sense. They are starting positions from which we derive consequences using agreed-upon rules. Mathematics is the game of deriving theorems from axioms. It is a formal activity, like chess.
Hilbert's point about "tables, chairs, and beer mugs" is the formalist manifesto. The axioms of geometry do not describe space. They describe abstract relationships between undefined terms. If the axioms are satisfied by points and lines, fine. If they are satisfied by tables and chairs, equally fine. The mathematics is the same because the mathematics is in the structure, not in the objects.
Formalism explains how contradictory axiom systems can coexist. Euclidean and hyperbolic geometry are not competing claims about reality. They are different games with different rules. Asking which is "true" is like asking whether chess or checkers is "true." The question does not apply. Both are consistent. Both are interesting. Both have applications.
Formalism has its own problems. If mathematics is just a game, why does it apply to the physical world with such uncanny precision? Eugene Wigner called this "the unreasonable effectiveness of mathematics in the natural sciences." Maxwell's equations, formulated as pure mathematics, predicted electromagnetic waves before anyone detected them. General relativity, a geometric theory, predicted gravitational lensing, time dilation, and black holes decades before they were observed. If mathematics is merely a game, it is a game that the universe happens to play by. This requires explanation, and formalism does not provide one.
The Structuralist View
A middle path. Structuralism holds that mathematics is about abstract structures: patterns of relationship that can be instantiated by many different concrete systems. The natural numbers are not any particular collection of objects. They are the structure described by the Peano axioms: anything that has a starting element, a successor function, and satisfies induction is the natural numbers, regardless of what the elements "really are." The von Neumann construction ($\emptyset, \{\emptyset\}, \{\emptyset, \{\emptyset\}\}, \ldots$) and the Zermelo construction ($\emptyset, \{\emptyset\}, \{\{\emptyset\}\}, \ldots$) both satisfy Peano's axioms. Both are equally valid instantiations. The natural numbers are not either construction. They are the shared structure.
On this view, axioms are definitions of structures. They are not truths about reality (Platonism) and not arbitrary game rules (formalism). They are precise specifications of patterns. The Peano axioms specify the natural-number pattern. The group axioms specify the symmetry pattern. The ZFC axioms specify the hierarchy-of-collections pattern. Whether these patterns exist in some Platonic realm, or only in minds and symbols, is a metaphysical question that structuralism deliberately sidesteps.
Why It Matters
The question "what is an axiom?" is not academic in the pejorative sense. It has consequences.
If axioms are self-evident truths, then mathematics is discovery, and there is one correct mathematics. Alternative axiom systems are curiosities at best, errors at worst. Mathematical knowledge is objective and certain in the strongest possible sense.
If axioms are game rules, then mathematics is invention, and there are infinitely many valid mathematics. The choice of axioms is pragmatic (which game is interesting? which game is useful?) rather than epistemic (which game is true?). Mathematical knowledge is certain within each game but says nothing about reality beyond the game.
If axioms are structural definitions, then mathematics is identification: recognizing abstract patterns that can be instantiated in multiple ways. The "truth" of mathematics lies not in the axioms themselves but in the relationships they describe. Mathematical knowledge is about structure, and the physical world is one (but not the only) instantiation of mathematical structures.
Each view has held dominance at different periods. Platonism dominated from antiquity through the 18th century. Formalism dominated the early 20th century (until Gödel's incompleteness theorems showed that no formal system can capture all of mathematical truth). Structuralism has grown in influence since the mid-20th century. None has conclusively won.
Part Three: Self-Evidence and Its Limits
The Declaration
"We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness."
Thomas Jefferson was making an axiomatic argument. The Declaration of Independence is structured like a mathematical proof: state your axioms (self-evident truths), derive your conclusions (the right to revolution), and present the derivation as logically compelling. The power of the argument rests on the claim that the axioms are self-evident, requiring no further justification.
But are they? "All men are created equal" was written by a man who owned slaves. It was not self-evident to most of human history that men were created equal. Aristotle explicitly argued that some men are natural slaves. The caste systems of India, the feudal hierarchies of medieval Europe, and the racial hierarchies of the colonial era all rested on the opposite axiom: that men are created unequal. Jefferson's "self-evident truth" was, in historical context, a radical and controversial axiom that most of the world's civilizations had explicitly rejected.
This illustrates the problem with self-evidence. What feels self-evident depends on who you are, where you are, and when you are. The parallel postulate felt self-evident to Euclid and false to Lobachevsky. The divine right of kings felt self-evident in 1600 and absurd in 1800. "All men are created equal" felt absurd in 1400 and self-evident (to many) in 1800. Self-evidence is historically contingent, not timeless.
The Lesson
This does not mean axioms are arbitrary. Jefferson's axiom proved more productive than Aristotle's. The political systems built on human equality (constitutional democracies) have, on the whole, produced more prosperity, more freedom, and more human flourishing than the systems built on natural hierarchy (feudalism, caste systems, slave states). The axiom is justified not by self-evidence but by its consequences.
The same is true in mathematics. The Peano axioms are not "self-evidently true." They are productive. They generate all of arithmetic and everything built on arithmetic. ZFC is not self-evidently the correct foundation for mathematics. It is the foundation that has proven most useful, most powerful, and most consistent with mathematical practice over the last century. If a better foundation emerged (and some argue that homotopy type theory or category theory may provide one), mathematicians would adopt it, not because the old axioms became false, but because the new ones proved more productive.
Axioms are judged by their fruits. A self-evident axiom that produces nothing is worthless. A controversial axiom that produces deep, beautiful, and applicable mathematics is invaluable. The Axiom of Choice is the clearest example: it is not self-evident, its consequences are sometimes paradoxical (Banach-Tarski), and it is accepted by almost all working mathematicians because the mathematics it generates is too powerful and too useful to abandon.
Conclusion
An axiom is a point where explanation stops and construction begins. Below the axioms, there is no further justification. Above them, there is an entire universe of consequences. The axioms themselves sit at the boundary between the unjustifiable and the derivable.
The major systems of mathematics (Peano arithmetic, Euclidean and non-Euclidean geometry, Hilbert's axioms, ZFC set theory, probability theory, field theory, group theory, propositional logic) each demonstrate this pattern. A small number of axioms, often fewer than ten, generates an enormous body of theorems, applications, and insights. The axioms are sparse. The consequences are infinite.
Whether axioms are perceived truths, game rules, or structural definitions remains unresolved. Perhaps the question admits no single answer. The Peano axioms feel like truths about counting. The Axiom of Choice feels like a pragmatic convention. The group axioms feel like a structural definition. Different axioms may have different ontological statuses, and the search for a unified account of what all axioms are may be misguided.
What is certain is this: every system of knowledge rests on axioms, acknowledged or not. Mathematics has the honesty to state its axioms explicitly. Most human endeavors do not. The political axioms of a society (who counts as a person, what rights they have, who holds legitimate authority) are rarely stated and rarely examined. The axioms of a scientific discipline (what counts as evidence, what methods are valid, what phenomena are real) are often assumed rather than argued. The axioms of a moral framework (what is good, what is right, what is the purpose of human life) are frequently held with passionate certainty and almost never justified.
The axiomatic method's deepest lesson is not about mathematics. It is about intellectual honesty. State your assumptions. Derive your conclusions. Check for consistency. And remember that someone, somewhere, is working with different axioms and getting different results, and that they may be just as consistent as you.