Implications, Converses and Contrapositives

An implication is simply an "if this, then that" sentence, and the big question is when such a sentence counts as a lie. The name comes from "imply": $p \to q$ says that $p$ being true forces $q$ to be true too. Every theorem has this shape ("if a function is differentiable, then it's continuous"), so it's worth getting exactly right. The formula $p \to q \equiv \neg p \vee q \equiv \neg q \to \neg p$ looks dense, but it's three ways of saying one thing: $p \to q$ (if p happens, q happens), $\neg p \vee q$ (either p doesn't happen, or q does) and $\neg q \to \neg p$ (if q didn't happen, p didn't either). It's just one promise said three ways.

So far our connectives have been symmetric: $p \wedge q$ and $q \wedge p$ mean the same thing. "If ... then" is different, and it's the connective every theorem is built on.

Think of $p \to q$ as a promise: "if you finish your homework ($p$), you can play ($q$)." When is the promise broken? Only if you finish and still can't play. If you don't finish, nobody can accuse the promise of lying, whether you end up playing or not. It just wasn't tested. So in those rows the implication counts as true, and we call it vacuously true:

Notice that the only F row has $p$ true and $q$ false. That gives a second way to write the implication: either $p$ fails or $q$ holds, which is $\neg p \vee q$.

By shuffling and negating the two parts of $p \to q$ you can form three related statements:

• the converse $q \to p$,
• the inverse $\neg p \to \neg q$,
• the contrapositive $\neg q \to \neg p$.

Why is the contrapositive the same as the original? Think about the promise again. If you can't play, then you must not have finished your homework, because finishing would have guaranteed playing. Both statements rule out exactly the same situation: $p$ true and $q$ false. The converse and inverse rule out a different one ($q$ true, $p$ false), so they agree with each other but not with the original.

A common mistake. It's tempting to think that if $p \to q$ is true, then $q \to p$ must be too, because they sound so similar. It isn't. "If a number is divisible by 4, it's even" is true, but its converse "if a number is even, it's divisible by 4" fails for 6. Proving the converse when you meant to prove the original is one of the most common errors in proofs.

When both directions do hold, we write the biconditional $p \leftrightarrow q$ ("p if and only if q"). It means $p \to q$ and $q \to p$ together, so the two statements always agree.

If-statements in code, database constraints and security policies are implications, and testers often check the contrapositive: if the output is wrong, which condition failed?