Inference to the Best Explanation

An ampliative reasoning form: from a body of observations, infer the hypothesis that best explains them.

Why does "best explanation" raise credence?

If a hypothesis would explain the data and rivals wouldn’t, that asymmetry raises our Credence in it.

1. We observe phenomenon E.
2. Hypothesis H, if true, would explain E.
3. No other available hypothesis explains E as well.
∴ 4. H is (probably) true.

Also called abduction (Charles S. Peirce). It’s how detectives reason, how doctors diagnose, and how science generates theories from data.

A good explanation typically scores well on:

Explanatory scope: accounts for all the relevant observations
Simplicity / parsimony: fewer ad-hoc assumptions (Occam’s Razor)
Consistency with established background knowledge
Predictive power: generates testable predictions beyond what it was designed to explain
Plausibility: doesn’t require violating well-established laws or facts

"Best" only works among the hypotheses you've considered

The true explanation might not be on the list. Good IBE requires actively brainstorming alternatives, not just defending the first one that comes to mind.

This is also where IBE shades into Affirming the Consequent: data that fit a hypothesis raise Credence only by ruling out competitors, not by deductive force.

Example

Muddy footprints leading to the kitchen, the dog came in from the yard (vs. a burglar in muddy boots, unlikely given other evidence)

The CMB radiation matches Big Bang predictions and not steady-state predictions, Big Bang is the best explanation

A patient has fever, rash, and stiff neck, meningitis is the best explanation among the candidates the doctor considers

How it can fail:

Too few alternatives considered: declaring something “the best” when you only thought of two
Mis-weighting criteria: favoring elegance over fit, or vice versa
Confusing “best available” with “good”: sometimes none of the hypotheses really fits, but you pick the least bad anyway