Conditional & Biconditional Statements

definitions
discrete-math
logic
conditional-statements

Conditional (\(\to\))

\(p\) \(q\) \(p \to q\)
T T T
T F F
F T T
F F T

Only one row makes a conditional false: the row where the hypothesis \(p\) is true and the conclusion \(q\) is false. Everything else is true, including both rows where \(p\) is false.

\(p \to q\) is true when \(p\) is false because that is how it is defined. There is no deeper intuition to recover; \(\to\) is a truth-functional connective fixed by the table above, and arguing with the bottom two rows is arguing with a definition. The standard name for those two rows is vacuous truth. It does real work later: it is why \(\emptyset \subseteq A\) for every set \(A\) (see sets-and-set-notation).

The picture behind the table is containment:

Figure 1: Why the truth table has the shape it does. Left: p → q asserts that the situations making p true all lie inside the situations making q true, so every region carries T. Right: the conditional fails exactly when part of p escapes q, which is the single F row of the table.

Disjunctive form

By definition,

\[p \to q \;\equiv\; \sim p \lor q\]

The right-hand side is the disjunctive form of the conditional, and it is the single most useful fact in the topic: the laws of logical equivalence (see logical-equivalence-laws) are stated only for \(\sim\), \(\land\) and \(\lor\), so they say nothing about \(\to\) directly. To use the laws on a conditional:

Replace \(p \to q\) with \(\sim p \lor q\) and then use the laws.

Every worked proof about conditionals in this course starts and (usually) ends with that substitution.

\(p \to q\) can be phrased: “\(p\) implies \(q\)”, “if \(p\) then \(q\)”, “\(q\) if \(p\)”, “\(p\) is sufficient for \(q\)”, “\(p\) only if \(q\)”, “\(q\) is necessary for \(p\)”.

Warning: \(p \to q \not\equiv q \to p\).

Contrapositive, converse, inverse

Name Form Equivalent to
Conditional \(p \to q\) Contrapositive
Contrapositive \(\sim q \to \sim p\) Conditional
Converse \(q \to p\) Inverse
Inverse \(\sim p \to \sim q\) Converse

The conditional and its contrapositive are always logically equivalent. The converse and inverse are always logically equivalent to each other — but not to the original conditional.

Figure 2: The four forms and the only equivalences among them. The conditional and the contrapositive are one logically equivalent pair, the converse and the inverse are the other, and p → q ≢ q → p. Swapping the two sides gives the converse, negating both gives the inverse, and doing both gives the contrapositive — the only one of the three that preserves meaning.

Both equivalences fall straight out of the disjunctive form. For the first: \(p \to q \equiv \sim p \lor q\) and \(\sim q \to \sim p \equiv \sim(\sim q) \,\lor\sim p \equiv q \,\lor\sim p\), which is the same disjunction commuted. The converse/inverse pair is the same computation with \(p\) and \(q\) interchanged.

Necessary and sufficient conditions

  • \(p\) is a necessary condition for \(q\) means “if \(\sim p\) then \(\sim q\)”, or equivalently “if \(q\) then \(p\)”, or “\(q\) only if \(p\)”. In symbols, \(q \to p\).
  • \(p\) is a sufficient condition for \(q\) means “if \(p\) then \(q\)”, or equivalently “\(q\) if \(p\)”. In symbols, \(p \to q\).

The two phrasings of “necessary” are contrapositives of each other, which is why they say the same thing: \(\sim p \,\to\sim q\) and \(q \to p\).

The reliable reading is: “if” introduces the sufficient condition; “only if” introduces the necessary one.

Worked: necessary but not sufficient

Being a non-negative integer is necessary for a number to be the square of an integer, but it is not sufficient.

  • Necessary: if a number is the square of an integer, then it is non-negative. So \(-9\) cannot be the square of an integer — failing the condition rules the number out.
  • Not sufficient: \(7\) is a non-negative integer but it is not the square of an integer. Meeting the condition does not settle it.

Worked: sufficient but not necessary

Being an integer that is divisible by 4 is sufficient for being an even integer, but it is not necessary.

  • Sufficient: if an integer is divisible by 4, then it is even. \(8\) is divisible by 4, so we know \(8\) is even without checking anything else.
  • Not necessary: \(10\) is an even integer but \(10\) is not divisible by 4. The condition can fail while the conclusion still holds.

The two shapes are exactly opposite in how they are refuted: a claimed sufficient condition is killed by something satisfying the condition but not the conclusion; a claimed necessary condition is killed by something satisfying the conclusion but not the condition.

Biconditional (\(\leftrightarrow\))

\[a \leftrightarrow b \equiv (a \to b) \land (b \to a)\]

A biconditional is the two one-way conditionals conjoined, and each of the standard English phrasings splits along the same seam:

Figure 3: The biconditional decomposed three ways, with each half in its own ringed column. Reading down the left column: a → b, ‘a only if b’, ‘b is necessary for a’ are three names for the same half. Reading down the right: b → a, ‘a if b’, ‘b is sufficient for a’. The whole necessary/sufficient vocabulary is just this correspondence — ‘only if’ is the necessary half, ‘if’ is the sufficient half.

Negating a conditional

\[\sim(p \to q) \;\equiv\; \sim(\sim p \lor q) \;\equiv\; p \land \sim q\]

by the definition of \(\to\), then De Morgan’s law, then the double negative law.

The negation of a conditional is not a conditional — it’s an and. “If it rains, the match is cancelled” is negated by “it rains and the match is not cancelled”, not by “if it rains, the match is not cancelled”.

The classic error is to answer \(\sim p \to \sim q\) (that’s the inverse) or \(q \to p\) (the converse). Neither is the negation, and neither is even equivalent to the original — see Figure 2 above.

Worked: \(\sim(p \to q) \not\equiv \sim p \to \sim q\). Let \(p\) be true and \(q\) be true. Then \(p \to q\) is true, so \(\sim(p \to q)\) is false. But \(\sim p \to \sim q\) is \(\mathrm{F} \to \mathrm{F}\), which is true. The two differ on this row, so they are not equivalent.

Sanity check on the shapes: \(\sim(p \to q)\) is a conjunction and is true on exactly one of the four rows; \(\sim p \to \sim q\) is a conditional and is true on three. They could hardly be equivalent.

Order of operations

  1. \(\sim\)
  2. \(\land, \lor\) (equal)
  3. \(\to, \leftrightarrow\) (equal)

Within a tier, work left to right if no parentheses are given. So \(p \to \sim q \lor r\) means \(p \to \big((\sim q) \lor r\big)\) — but parenthesise explicitly rather than relying on the tie-breaking rule.

Notation, not to be blurred: the biconditional \(\leftrightarrow\) is a connective used to combine statements (like \(+\), \(-\), \(\times\), \(\div\) with numbers). Logical equivalence \(\equiv\) is used when comparing two statements (like \(=\) with numbers). \(a \leftrightarrow b\) is a statement; \(a \equiv b\) is a claim about two statements.

See also