Department of Computer Science and Engineering, NIT Calicut

Lecture 2: Classical Deduction in Propositional Logic

Prepared by: K. Murali Krishnan

In modeling a real system, we start with a primitive set of variables V and postulate a
set A of defining properties of the system as the axioms or postulates. The axioms must be
adequate so that all true formulas about the the system (theorems) are logical consequences
of the axioms. A desirable property of an axiom system is to have an independent set of
axioms, so that the axioms contain no redundancies.
Given an axiom system A and a formula φ, the problem of determining whether A |= φ
is called the deduction problem. Attempting all truth assignments brute force is impractical.
Instead, we try to identify a formula ψ already known to satisfy A |= ψ (for example ψ
could be an axiom) and show that ψ ⇒ φ. It will follow that A |= φ. This essentially is the
fundamental notion of logical deduction used in mathematical reasoning.
To execute this plan, one needs a good database of “standard” tautological implications.
Some of them are developed in the following exercises. The following section formalizes the
notions of axiomatic systems, deduction, theorem-hood etc.

Example 1. The formulas (p ∨ ¬p), ¬(p ∧ ¬p) (called laws of negation), p ∧ (p → q) ⇒ q

(Modus Ponens), (p → q) ∧ ¬q ⇒ ¬p (Modus Tollens), (p ∨ q) ∧ ¬p ⇒ q (disjunctive
syllogism) (p → q) ∧ (q → r) ⇒ (p → r) (hypothetical syllogism), (p → q) ⇔ (¬q →
¬p) (Law of contrapositive), ¬(p ∨ q) ⇔ (¬p ∧ ¬q), ¬(p ∧ q) ⇔ (¬p ∨ ¬q) (De-Morgan’s
Laws), (p → q) ⇔ (¬p ∨ q), ¬¬p ⇔ p (Law of double negation) etc. are standard examples
of tautologies. ∧, ∨ are associative, commutative and distribute over each other. The
following properties: (p ∧ p) ⇔ (p ∨ p) ⇔ p, (p ∧ q) ⇒ p, (p ∧ q) ⇒ q are collectively called
the absorption properties. p ⇒ (p ∨ q) is called generalization.
As an example, to verify modus ponens, suppose τ (p ∧ (p → q) → q) = 0 Then τ (q) = 0
and τ (p ∧ (p → q)) = 1. The latter requires τ (p) = 1 and τ ((p → q)) = 1. But as
τ (q) = 0, for τ ((p → q)) = 1, we need τ (p) = 0 which is a contradiction. (Another standard
verification method is using truth tables) Other formulas are verified similarly.

These laws are commonly used in mathematical arguments. For instance, the proof of a
mathematical statement of the form “α if and only if β,” is typically split into two parts -
“if α then β” and “if β then α”. (The latter forumla is called the converse of the former).
This is justified because (α ↔ β) ⇔ (α → β) ∧ (β → α). Note that proving “if ¬α then
¬β” (called the inverse of α → β)) is equivalent to proving “if β then α” (why?). Similarly
proving the contrapositive “if not β then not α” proves “if α then β”.

Exercise 1. Identify logically equivalent pairs of statements among the following:

a) If a number is even, its square is even.
b) if a number is odd its square is odd.
c) if the square of a number is even, the number is even.
d) if the square of a number is odd, the number is odd.

2-1
 The law of double negation√is also used in practical reasoning.√ For instance, the only
proof known for showing that 2 is irrational it to assume that 2 is rational and derive
a contradiction.
Another important technique is substitution. Suppose we replace a variable with any
formula uniformly in a tautology φ, then the resultant formula is also a tautology. For
instance in the tautology φ = p ∨ ¬p, if we substitute p everywhere with (p → (q ∨ r))
(written φ[p ← (p → (q ∨ r))]) we get (p → (q ∨ r) ∨ ¬(p → (q ∨ r)) which also is a
tautology. This is because φ evaluates to 1 under any value to the expression. Similarly,
since equivalent formulas evaluate to the same truth value, any formula in an expression can
be substituted by a logically equivalent formula without affecting truth values. Thus, for
instance, the expression (p → q) can be subtituted with (¬p ∨ q) anywhere in the formula
(p → q) → (q ∨ (p → q)) with no change in the set of models satisiying the formula. These
rules are summarized below for easy reference:
Theorem 1 (Laws of Substitution). 1. If φ ∈ FV is a tautology. Let p ∈ V be a variable
in φ. Let ψ ∈ FV . Then formula φ[p ← ψ] obtained by substitution of all occurrence
of p in φ with ψ in φ also is a tautology.

2. Let φ, α.β ∈ FV , if α is a subformula of φ and α ⇔ β, Let φ0 be the formula obtained

by replacing of α with β in φ. Then for every τ ∈ TV , τ (φ) = τ (φ0 ).

Classical Deduction
Definition 1. Let A ⊆ FV . Define the set T h(A) = {φ ∈ FV : A |= φ}. This set is the
collection of all formulas which are logical consequences of A.
Axiomatic deduction methods were known right from the time of the ancient Greeks
and an axiomatic treatment of geometry can be found in Euclid’s book “The Elements”.
The assumptions about the system under study were postulated as axioms and logical
consequences of these axioms were derived using a set of “standard” deduction rules. The
derived consequences are called theorems. The notion of algorithmic (automated) deduction
did not exist in those times and deductions had to be done manually. In this section, we
shall discuss a few “manual” deduction methods well known from from the ancient times.
Theorem 2 (Laws of Deduction). Let A ⊆ FV and φ, ψ ∈ FV . Then:
• (Deduction Theorem): A |= φ and (φ ⇒ ψ) then A |= ψ.

• (Law of Implication): A |= (φ → ψ) if and only if A ∪ {φ} |= ψ

• (Method of Contradiction): A ∪ {¬φ} is inconsistent if and only if A |= φ.

Proof. The first statement is proved here and the rest are left as exercises. Suppose A |= φ.
Suppose τ ∈ TV satisfies τ |= A. Then, by hypothesis, τ |= φ. As φ ⇒ ψ is a tautology,
τ |= ψ (why?). This proves the first part. The other parts are proved similarly.

Example 2. Let A = {p → q, q → ¬(r → p), ¬r ∨ ¬q}. Here is a deduction for ¬p:

1. q → ¬(r → p) (axiom)

2-2
 2. q → ¬(¬r ∨ p) (Substitution: (r → p) ⇔ (¬r ∨ p)).

3. q → (r ∧ ¬p) (Substitution: De-Morgan’s equivalence).

4. p → q (axiom)

5. p → (r ∧ ¬p) (Hypothetical Syllogism from 4,3)

6. ¬p ∨ (r ∧ ¬p) (Substitution)

7. (¬p ∨ r) ∧ (¬p ∨ ¬p) (Distributive law).

8. ¬p ∨ ¬p (Absorption law)

9. ¬p. (Absorption law)

Exercise 2. Let V = {p, q, r}. Consider A = {p → (p → q), q → (r ∨ p), (¬p ∨ ¬q) ∧ ¬r}.
Is A categorical? Is p a logical consequence of the axioms? If so, give a deductive proof.

Exercise 3. It is true that Mike has a bike. If Mike has a bike, then Mike can’t have a car.
If mike does not have a car, then Mike can’t travel long distance. Either Mike travels long
distance or Mike is a sportsman. Formulate the above statements in propositional logic. Is
the statement “Mike is a sportsman” a valid consequence of these statements? If so, find a
deduction for the statement based on the above laws. Are these statements consistent? Do
they form a categorical set?

Exercise 4. There are 3 mess houses - A, B, and C - in a college. One fine day, a notice
appeared in front of the notice board stating the following regarding processing of mess change
applications from students.

• “All except those who have not applied for change of mess from A mess to B mess
or A mess to C mess are not permitted to change their mess. Further, no applicants
applying to A is not permitted to change the mess”

Formulate the above conditions in the framework of propositional logic. You may assume
that a person can make only one mess change request. There are 6 possible mess change
options that a student may exercise - A to B, A to C, B to A, B to C, C to A and C to B. In
each case, formulate the problem of determining whether the mess change is permissible/not
permissible as a problem of checking satisfiability/unsatisfiability of a set of propositional
formulas. Note that, a particular mess change request can have three possibilities - a) It is
sure to be accepted, as the given conditions mandate its acceptance b) It may or may not
be accepted as the stated conditions permit both possibilities c) It is sure to be rejected. For
each of the six transfer requests, logically argue to which category the request belongs to.
Hint: Let V = {pab , pac ...} be propositional variables for various possibilities. Note that
there are some hidden assumptions to take care when one models a real world problem like
this. For example a person cannot simultaneously apply for a mess change from A to B and
B to C, as a person can have only one “current mess.”

2-3