Propositional Logic

Thinking ToolsAny step

From the Stoic logic of Chrysippus (3rd century BCE), recast by George Boole (1847) and Gottlob Frege (1879), and given modern form by Alfred North Whitehead and Bertrand Russell (Principia Mathematica, 1910).

Reduces an argument to lettered statements joined by and, or, not, and if-then, then checks whether the conclusion follows from that structure alone.

The argument for why the client left, tested row by row

P is the launch slips; Q is the client walks. The argument says: if P then Q; Q; so P. Each pick of true or false is one row of the truth table. Try all four and find the row where both premises hold and the conclusion fails. Illustrative.

P: the launch slips

Q: the client walks

  • Pick true or false for both statements.The counterexample. The launch was on time and the client walked anyway, for some other reason, and both premises are still true. So the argument is invalid: this error is called affirming the consequent.The case everyone pictures: premises true, conclusion true. One agreeing row proves nothing; an argument is valid only if no row has true premises and a false conclusion.Premise 1 fails here (the launch slipped and the client stayed), so this row cannot test the argument.Premise 2 fails here (the client stayed), so this row cannot test the argument.
  • The counterexample. The launch was on time and the client walked anyway, for some other reason, and both premises are still true. So the argument is invalid: this error is called affirming the consequent. (P: False, Q: True)
  • The case everyone pictures: premises true, conclusion true. One agreeing row proves nothing; an argument is valid only if no row has true premises and a false conclusion. (P: True, Q: True)
  • Premise 1 fails here (the launch slipped and the client stayed), so this row cannot test the argument. (P: True, Q: False)
  • Premise 2 fails here (the client stayed), so this row cannot test the argument. (P: False, Q: False)
Premise 1: if P then Q
Pick every rowFalseTrue
Premise 2: Q
Pick every rowFalseTrue
Conclusion: P
Pick every rowFalseTrue
Premises true, conclusion false?
Pick every rowNoYes: a counterexample

An example

An agency lost its largest client, and the account director's explanation at the partners' meeting was tidy: the client had told them that if the platform launch slipped, they would walk; the launch slipped; they walked. The proposal on the table was to replace the delivery lead.

The finance partner wrote the claim as letters. P, the launch slips; Q, the client walks. The client's warning was if P then Q. But the director's case for blaming the launch was a different argument: if P then Q, and Q, so P was the cause. She drew the four rows and pointed to the one where P is false and Q true: the client walks for some other reason, and the warning is still true. Nothing in the argument ruled it out. Two calls to the client's team found the other reason: a new chief information officer consolidating agencies across the group. The delivery lead stayed, and the partners spent the next meeting on the consolidation risk at their other two clients in the same group.

Why it works

Whether a conclusion follows depends on the shape of the argument, not on what its statements are about. Replacing each statement with a letter strips out the content that makes a conclusion feel right, and leaves only the shape to check.

The truth table checks every case at once. With two statements there are four rows, one for each way the facts could be. An argument is valid only if no row makes every premise true and the conclusion false; one such row is a counterexample, and the argument fails whatever its words mean. Some shapes fail so often in practice that they have names. Affirming the consequent (if P then Q, Q, so P) is the classic one in post-mortems: an outcome a cause would produce is taken as proof of that cause.

The thinking behind it

Chrysippus of Soli Chrysippus of Soli, third head of the Stoic school in the third century BCE, built logic on propositions where Aristotle had built it on terms. He separated simple from compound propositions, defined the connectives that join them, and set out five basic argument forms, among them the ones now called modus ponens and modus tollens. None of his works survive whole; his logic is known from fragments and later reports, chiefly Diogenes Laertius.

George Boole George Boole, an English mathematician, recast logic as an algebra of classes and propositions in The Mathematical Analysis of Logic (1847) and An Investigation of the Laws of Thought (1854). Treating true and false as values that can be calculated with is what makes a truth table a computation, and the algebra that carries his name now underlies both propositional logic and digital circuit design.

Bertrand Russell Bertrand Russell, a founder of analytic philosophy, and Alfred North Whitehead, an English mathematician who later founded process philosophy, set out to derive the whole of mathematics from logic in Principia Mathematica (three volumes, 1910 to 1913). The attempt laid down a formal system for propositional and predicate logic, with its opening part developing the propositional calculus, and gave the subject the modern form in which it is still taught.

How to run it

  1. Write the argument as plain sentences. Premises first, conclusion last, each a statement that is either true or false.
  2. Give each simple statement a letter. P for the launch slips, Q for the client walks; reuse a letter wherever the same statement appears.
  3. Rewrite with the four connectives. And, or, not, and if-then; watch for unless (if not) and only if (which reverses the arrow).
  4. Draw the truth table. Two letters make four rows, three make eight; fill in each premise and the conclusion on every row.
  5. Look for the counterexample row. Any row with every premise true and the conclusion false proves the argument invalid.
  6. Name the shape if it has a name. Affirming the consequent and denying the antecedent are the common invalid ones.
  7. Go back to the content. A counterexample row is a real possibility to investigate, such as another reason the client could have left.

Where it goes wrong

The translation is wrong. Only if gets written as if, or a clause with two ideas gets one letter, and the table then tests a different argument. Read the lettered version back in plain words and ask the person who made the argument whether that is what they meant.

Valid gets taken for true. An argument can have a perfect shape and false premises; the table says the conclusion follows, not that the premises hold. Check the premises separately.

The table is built for an argument that is really about probability. Most business reasoning says likely, not always, and a strict if-then overstates it. Where a premise is a tendency, use the table to find the gap and then weigh the evidence for it.

When to use something else

  • Toulmin Model. When the argument rests on evidence and a warrant that make a claim likely, not certain.
  • Abductive Reasoning. When the job is finding the best explanation for an outcome, such as why the client really left.
  • Bayesian Updating. When the question is how much one piece of evidence should change the odds of a cause.

Where it sits in SOLVE eX

  • Any step: Tools that serve every step: habits of thought and checks on the thinking itself.

Go to the source

  • Stoic propositional logic, 3rd century BCE
  • The Mathematical Analysis of Logic, 1847
  • Begriffsschrift, 1879
  • Principia Mathematica, 1910

Back to the catalog