PHIL 109 · Introduction to Formal Reasoning and Decision Making

← Contents

Formula Sheet

The definitions, axioms and theorems of probability from Part II. Each one links to the place in the book where it is stated (and, for the theorems, proved).

Definitions

Mutually exclusive. Two propositions φ\varphi and ψ\psi are mutually exclusive iff there is no row of their truth-table in which both are true. (So φ∧ψ\varphi \land \psi is a contradiction, and P⁡(φ∧ψ)=0\Prb(\varphi \land \psi) = 0.) (Definition 4.1)

Conditional probability. For any φ\varphi and ψ\psi, if P⁡(ψ)≠0\Prb(\psi) \neq 0, then P⁡(φ∣ψ)=P⁡(φ∧ψ)/P⁡(ψ)\Prb(\varphi \given \psi) = \Prb(\varphi \land \psi) / \Prb(\psi). (Definition 4.2)

Independence. φ\varphi is independent of ψ\psi iff P⁡(ψ)≠0\Prb(\psi) \neq 0 and P⁡(φ∣ψ)=P⁡(φ)\Prb(\varphi \given \psi) = \Prb(\varphi). φ\varphi is dependent on ψ\psi iff P⁡(ψ)≠0\Prb(\psi) \neq 0 and P⁡(φ∣ψ)≠P⁡(φ)\Prb(\varphi \given \psi) \neq \Prb(\varphi). φ\varphi and ψ\psi are independent of each other iff each is independent of the other. (Definition 5.2)

Conditional independence. φ\varphi and χ\chi are independent conditional on ψ\psi iff P⁡(φ∧χ∣ψ)=P⁡(φ∣ψ)×P⁡(χ∣ψ)\Prb(\varphi \land \chi \given \psi) = \Prb(\varphi \given \psi) \times \Prb(\chi \given \psi). (Chapter 6)

Axioms

Non-negativity. For any φ\varphi, P⁡(φ)≥0\Prb(\varphi) \geq 0. (Axiom 1)

Normality. If τ\tau is a tautology, then P⁡(τ)=1\Prb(\tau) = 1. (Axiom 2)

Additivity. If φ\varphi and ψ\psi are mutually exclusive, then P⁡(φ∨ψ)=P⁡(φ)+P⁡(ψ)\Prb(\varphi \lor \psi) = \Prb(\varphi) + \Prb(\psi). (Axiom 3)

Theorems

Contradiction. If φ\varphi is a contradiction, then P⁡(φ)=0\Prb(\varphi) = 0. (Theorem 5.1)

Complement. For any φ\varphi, P⁡(φ)+P⁡(¬φ)=1\Prb(\varphi) + \Prb(\neg\varphi) = 1. (Theorem 5.5)

Equivalence. If φ\varphi and ψ\psi are tautologically equivalent, then P⁡(φ)=P⁡(ψ)\Prb(\varphi) = \Prb(\psi). (Theorem 5.6)

Bounds. For any φ\varphi, 0≤P⁡(φ)≤10 \leq \Prb(\varphi) \leq 1. (Theorem 5.7)

Entailment. If φ⊨ψ\varphi \vDash \psi, then P⁡(φ)≤P⁡(ψ)\Prb(\varphi) \leq \Prb(\psi). (Theorem 5.8)

Extended additivity. If φ1,…,φn\varphi_{1}, \ldots, \varphi_{n} are pairwise mutually exclusive, then P⁡(φ1∨⋯∨φn)=P⁡(φ1)+⋯+P⁡(φn)\Prb(\varphi_{1} \lor \cdots \lor \varphi_{n}) = \Prb(\varphi_{1}) + \cdots + \Prb(\varphi_{n}). (Theorem 5.9)

Inclusion-exclusion. For any φ\varphi and ψ\psi, P⁡(φ∨ψ)=P⁡(φ)+P⁡(ψ)−P⁡(φ∧ψ)\Prb(\varphi \lor \psi) = \Prb(\varphi) + \Prb(\psi) - \Prb(\varphi \land \psi). (Theorem 5.10)

Conjunction. For any φ\varphi and ψ\psi: (a) if P⁡(ψ)≠0\Prb(\psi) \neq 0, then P⁡(φ∧ψ)=P⁡(φ∣ψ)×P⁡(ψ)\Prb(\varphi \land \psi) = \Prb(\varphi \given \psi) \times \Prb(\psi); (b) if P⁡(ψ)=0\Prb(\psi) = 0, then P⁡(φ∧ψ)=0\Prb(\varphi \land \psi) = 0. (Theorem 5.11)

Conjunction of independents. If φ\varphi is independent of ψ\psi, then P⁡(φ∧ψ)=P⁡(φ)×P⁡(ψ)\Prb(\varphi \land \psi) = \Prb(\varphi) \times \Prb(\psi). (Theorem 5.12)

Product test. If P⁡(ψ)≠0\Prb(\psi) \neq 0, then φ\varphi is independent of ψ\psi iff P⁡(φ∧ψ)=P⁡(φ)×P⁡(ψ)\Prb(\varphi \land \psi) = \Prb(\varphi) \times \Prb(\psi). (†)

Symmetry of independence. If P⁡(φ)≠0\Prb(\varphi) \neq 0 and P⁡(ψ)≠0\Prb(\psi) \neq 0, then φ\varphi is independent of ψ\psi iff ψ\psi is independent of φ\varphi. (Theorem 5.13)

Conditional complement. If P⁡(ψ)≠0\Prb(\psi) \neq 0, then P⁡(φ∣ψ)+P⁡(¬φ∣ψ)=1\Prb(\varphi \given \psi) + \Prb(\neg\varphi \given \psi) = 1. (Theorem 5.14)

Bayes' theorem. If P⁡(φ)≠0\Prb(\varphi) \neq 0 and P⁡(ψ)≠0\Prb(\psi) \neq 0, then P⁡(φ∣ψ)=P⁡(ψ∣φ)×P⁡(φ)/P⁡(ψ)\Prb(\varphi \given \psi) = \Prb(\psi \given \varphi) \times \Prb(\varphi) / \Prb(\psi). (Theorem 6.3)

Total probability. If 0<P⁡(ψ)<10 < \Prb(\psi) < 1, then P⁡(φ)=P⁡(φ∣ψ)×P⁡(ψ)+P⁡(φ∣¬ψ)×P⁡(¬ψ)\Prb(\varphi) = \Prb(\varphi \given \psi) \times \Prb(\psi) + \Prb(\varphi \given \neg\psi) \times \Prb(\neg\psi). (Theorem 6.4)

Bayes' theorem (LTP). If 0<P⁡(φ)<10 < \Prb(\varphi) < 1 and P⁡(ψ)≠0\Prb(\psi) \neq 0, then P⁡(φ∣ψ)=P⁡(ψ∣φ)×P⁡(φ)P⁡(ψ∣φ)×P⁡(φ)+P⁡(ψ∣¬φ)×P⁡(¬φ).\Prb(\varphi \given \psi) = \frac{\Prb(\psi \given \varphi) \times \Prb(\varphi)}{\Prb(\psi \given \varphi) \times \Prb(\varphi) + \Prb(\psi \given \neg\varphi) \times \Prb(\neg\varphi)}. (Theorem 6.5)

† Not proved in the text; it follows quickly from the definitions and theorems above.