# 2.4 Independence Phenomena

## 2.4 Independence Phenomena

A statement is independent of an axiom system when the system proves neither the statement nor its negation. Independence is a precise form of undecidability.

```text
T ⊬ φ
T ⊬ ¬φ
```

Here `T` is a theory and `φ` is a sentence in the language of that theory. The notation says that `φ` cannot be derived from `T`, and `¬φ` cannot be derived from `T` either.

Independence does not mean that the statement is vague. It means the chosen axioms do not contain enough information to decide it.

A common way to prove independence is model construction. To show that `φ` is independent of `T`, one builds two models:

```text
M ⊨ T + φ
N ⊨ T + ¬φ
```

The first model satisfies the axioms together with `φ`. The second model satisfies the axioms together with `¬φ`. If both models exist, then `T` alone cannot decide `φ`, assuming the background reasoning is sound.

| Goal              | Method                     |
| ----------------- | -------------------------- |
| Show `T ⊬ φ`      | Build a model of `T + ¬φ`  |
| Show `T ⊬ ¬φ`     | Build a model of `T + φ`   |
| Show independence | Build both kinds of models |

The parallel postulate is the classical geometric example. Euclidean geometry accepts it. Hyperbolic geometry rejects it and replaces it with a different behavior of parallel lines. The remaining axioms do not force either choice.

| Geometry         | Parallel behavior                                   |
| ---------------- | --------------------------------------------------- |
| Euclidean        | Exactly one parallel through a point outside a line |
| Hyperbolic       | More than one parallel through such a point         |
| Neutral geometry | Does not decide the question                        |

This shows that a statement may look geometrically obvious because of a familiar model, while still being independent from a smaller axiom base.

Set theory gives deeper examples. The continuum hypothesis concerns the size of infinite sets. It asks whether there is a set whose cardinality lies strictly between the natural numbers and the real numbers. Within the standard ZFC axioms, the continuum hypothesis can neither be proved nor refuted, assuming ZFC is consistent.

This has an important consequence. Both extensions are possible as formal theories:

```text
ZFC + CH
ZFC + ¬CH
```

Each can be studied as a legitimate mathematical universe, relative to consistency assumptions.

Independence also appears in arithmetic. Gödel-style statements are constructed to say, in effect, that they are not provable in a given system. If the system is consistent and strong enough to express arithmetic, such a statement may be true in the intended natural numbers but unprovable inside the system.

This creates a distinction between:

| Status             | Meaning                                                     |
| ------------------ | ----------------------------------------------------------- |
| False              | Refuted in the intended interpretation                      |
| Unproved           | No proof currently known                                    |
| Unprovable in `T`  | No proof exists from the axioms of `T`                      |
| Independent of `T` | Neither the statement nor its negation is provable from `T` |

The distinction between “unproved” and “unprovable” is especially important. A conjecture may be unproved today but provable tomorrow. An independent statement cannot be settled without changing the axioms or the formal setting.

Independence is always relative to a theory. A statement independent of one theory may be decided by a stronger theory.

```text
T ⊬ φ
S ⊢ φ
```

where `S` extends `T`.

For example, adding an axiom can decide a previously independent statement. The cost is that the stronger theory depends on the acceptability and consistency strength of the new axiom.

| Action                         | Effect                                   |
| ------------------------------ | ---------------------------------------- |
| Add `φ` as an axiom            | Forces `φ` to be true in the new theory  |
| Add `¬φ` as an axiom           | Forces `φ` to be false in the new theory |
| Add stronger structural axioms | May decide `φ` indirectly                |
| Change logic or foundation     | May change what counts as proof          |

Independence phenomena show that axioms are not merely starting points. They shape the mathematical universe being studied.

In practice, independence has several uses.

First, it clarifies the limits of a theory. If a statement is independent, further work inside the same axiom system will not settle it.

Second, it separates theorem proving from axiom selection. Once independence is known, the question shifts from “Can we prove this?” to “Which extension gives the right theory for the intended purpose?”

Third, it exposes hidden assumptions. A proof may appear to use only elementary reasoning but actually depend on a stronger principle.

The axiom of choice provides a useful example. Many results in algebra, topology, and analysis use it, sometimes implicitly. Some theorems are equivalent to weak or strong forms of choice. When choice is omitted, those theorems may fail or become independent.

| Statement or principle         | Relation to choice                      |
| ------------------------------ | --------------------------------------- |
| Every vector space has a basis | Uses choice                             |
| Tychonoff theorem              | Equivalent to choice in common settings |
| Zorn's lemma                   | Equivalent to choice                    |
| Well-ordering theorem          | Equivalent to choice                    |

Independence therefore affects mathematical style. A constructive proof may avoid stronger axioms. A classical proof may use them freely. A foundationally careful proof states which principles are required.

Independence also affects classification. A theory with independent statements admits multiple non-equivalent extensions. Instead of one final universe, there may be a family of consistent mathematical worlds.

This is common in set theory, where forcing and inner model methods build models with different properties. The same base axioms can support different answers to questions about cardinality, definability, and combinatorial principles.

A useful workflow for independence questions is:

| Step | Question                                          |
| ---- | ------------------------------------------------- |
| 1    | What theory `T` is being used?                    |
| 2    | What exact statement `φ` is being tested?         |
| 3    | Can `T + φ` be modeled?                           |
| 4    | Can `T + ¬φ` be modeled?                          |
| 5    | What additional axioms decide the statement?      |
| 6    | What mathematical cost do those axioms introduce? |

Independence does not weaken mathematics. It makes the role of axioms explicit. It shows that some questions are not settled by logic alone. They require a choice of framework.

A mature theory handles independence by naming its assumptions, comparing extensions, and tracking which results depend on which axioms.