The Problem of Induction

No amount of observations can justify a universal claim.

Induction is traditionally defined as the derivation of general conclusions from particular observations.

The Problem

Induction generalizes from observed instances:

• The sun has risen every day I have observed. Therefore, the sun will rise tomorrow.
• Every swan I have seen is white. Therefore, all swans are white.
• This medicine has worked in all trials so far. Therefore, it will work in the future.

No finite number of observations can guarantee a universal claim. The next observation might be the counterexample.

Every white swan you observe is compatible with the next swan being black. Every successful trial is compatible with the next trial failing. Every sunrise is compatible with the sun not rising tomorrow.

Justification

Attempts to justify induction fail in every direction:

1. Logical justification fails. The step from "all observed X are Y" to "all X are Y" is not deductively valid. The premises can be true and the conclusion false.

2. Inductive justification is circular. "Induction has worked in the past, therefore it will work in the future" is itself an inductive argument. It assumes what it tries to prove.

3. Probabilistic justification shifts the problem. Saying "probably all X are Y" still requires justifying why past frequencies should predict future frequencies, which is the same problem restated.

The Trilemma

The Trilemma of Justification shows that all chains of justification terminate in regress, circularity, or dogma.

The problem of induction is a specific instance:

• Regress: each inductive step requires another justification
• Circularity: induction is justified by induction
• Dogma: "induction just works"

Both problems point to the same conclusion: justification cannot ground knowledge.

Implications

The problem of induction is a constraint that any adequate epistemology must respect.

1. Induction does not produce knowledge. If induction cannot justify universal claims, then observations alone cannot ground explanatory conjectures. Knowledge requires conjecture and criticism, not generalization from instances.

2. Justified true belief cannot account for knowledge of regularities. If justification requires inductive support, and induction is unjustifiable, then JTB cannot ground knowledge.

3. Observations test, not confirm. A theory that survives testing is retained because it has not been refuted, not because it has been justified.

2025-10-04 Aaron Brinton
2026-02-10 overhaul; connected to published articles
2026-02-26 reframed as traditional definition