Conditionals: What's in an If?

Core idea

A conditional is a sentence of the form if a then c — call a the antecedent, c the consequent. Conditionals are everywhere in reasoning, yet they are deeply puzzling. The standard logical treatment identifies if a then c with the material conditional, written a → c, which is false exactly when a is true and c is false, and true in every other case.

That truth table delivers two unsettling results. First, whenever c is true, a → c comes out true — so "If Canberra is not the capital of Australia, Canberra is the capital of Australia" is counted true. Second, whenever a is false, a → c comes out true — so "If Sydney is the capital of Australia, then Brisbane is the capital" is counted true. Both conditionals are plainly false.

Priest's argument: These cases show that the ordinary conditional is not a truth function at all — the truth value of if a then c is not fixed by the truth values of a and c alone.

Why it matters

The paradoxes of material implication

"Rome is in France" and "Beijing is in France" are both false. Yet "If Italy is part of France, Rome is in France" is true while "If Italy is part of France, Beijing is in France" is false. Same truth values for the parts, different truth values for the whole. A purely truth-functional conditional cannot tell these apart, so it misrepresents how if actually behaves.

A possible-worlds repair

Truth Functions — or Not?'s machinery offers a fix. Say if a then c is true in a situation s just if c is true in every possible situation associated with s in which a is true. Had Italy been incorporated into France, Rome would have been in France — but China would have been untouched, so Beijing still would not. This account validates modus ponens (since s itself counts among its associated situations) and blocks both paradoxes.

Key takeaways

Mental model

Copy sourceZoom

Practical application

When someone asserts a conditional, do not treat it as a bare claim about truth values. Treat it as a claim that a connection holds.

This matters in argument analysis: an inference can look valid because we silently import a connection the formal conditional never guaranteed.

Example

Consider a chained inference of the standard "transitive" form: If Smith dies before the election, Jones will win; if Jones wins, Smith will retire and draw her pension; therefore if Smith dies before the election, she will retire and draw her pension. Each conditional is plausible on its own. The conclusion is absurd — a dead candidate cannot draw a pension. Or a strengthened one: if Smith jumps off a cliff she will die from the fall; therefore if Smith jumps off a cliff wearing a parachute she will die from the fall. The parachute breaks the inference. Both forms are valid on the possible-worlds account, yet both have transparently bad instances. That is the crack the topic leaves open: even the repaired conditional is not the last word.

Related lessons