1 Rules of Inference

Rules of inference are used to deduce true formulas from other true formulas.
Notation. Upper case letters P, Q, R, S will be used to denote propositional formulas (rather than single
variables).
The rules of inference are stated using the following notation
Hypothesis
Conclusion
which mean given the Hypothesis, we can infer (or deduce) the Conclusion. If there is more than one
hypothesis, it is written
Hypothesis1
Hypothesis2
..
.
Hypothesisk
Conclusion

Rule Name Book name

P ∧Q
P ∧E Simplification
P
Q ∧I Conjunction
P ∧Q
P
— The rules given in the book
..
. avoids this rule by using
Q →I the equivalence of
P →Q p → q and ¬p ∨ q
P
P →Q →E Modus Ponens
Q
P
P ∨Q ∨I Addition
P ∨Q
P →R
Q→R ∨E Disjunctive syllogism
R (not exactly the same)
P →F Modus Tollens
¬P ¬I (not exactly the same)
P
¬P ¬E
F
¬¬P
P ¬¬E
F
P FE

These are the inference rules of a natural deduction system. We will be using these rules but there are
other set of rules that we could have used.
1 Made using Paul Taylor’s boxuser LATEX macros

1
Notation. The “I” in the table means “introduction” and the “E” in the table means “elimination”. So,
for example, the name of the first rule is “and elimination”.
The first rule can be read as “From P ∧ Q, we can infer P ”.
The rule →I states that if taking P as an assumption, after a number of steps we arrive at Q then we
can infer P → Q while losing our assumption P (and any formula derived from it).
Notation. We could have also written a rule
Hypothesis
Conclusion
in a “linear form” as
Hypothesis  Conclusion
and the case with multiple hypotheses as

Hypothesis1 , Hypothesis2 , . . . , Hypothesisk  Conclusion

This can be read as “Given Hypotheses1 ,. . .,Hypothesesk , we can infer the Conclusion”.
Example 1. We will now prove p ∧ q → p (using no premise, thus showing that it is a tautology)

1 p∧q assumption
2 p 1, ∧E
3 p∧q →p 1 − 2, →I
Note that we have proven p ∧ q → p without any hypothesis. Thus we can write  p ∧ q → p.
Definition 1. Any formula which can be infered using no premise is called a theorem.
Example 2. We will now prove what is called “Modus Tollens” in the book. The statement is ¬q, p → q  ¬p.

1 p→q premise
2 ¬q premise
3 p assumption
4 q 1, 3, →E
5 F 2, 4, ¬E
6 p→F 3 − 5, →I
7 ¬p 6, ¬I
Note. Each line of a proof should either be a premise, an assumption or a formula which is true given all
formulas in previous lines are true.
There should not be any assumptions left at the end of a proof (only premises).
Example 3. Here is an example of a mathematical proof which has been formatted differently than usual
so that it ressembles a proof in logic.
√
Theorem 1. If 2 > 32 then 2 > 94 .

2
Proof.
√ 3
1 2> assumption
2
√ 2 √ 3 √
2 2 > 2 1, multiply both sides by 2
 22
√ 3 3
3 2 > 1, multiply both sides by 32
2 2
 2
√ 2 3
4 2 > 2, 3, transitivity of >
2
9
5 2> evaluation of the left-hand side and right-hand side
4

√
Of course, the premise 2 > 32 is false. But this does not make the implication and thus our theorem
false.
Here is what the proof may look like if it were normally formatted.
√ √ 2 √  2
Proof. Suppose 2 > 32 . Then 2 > 2 32 > 32 . Therefore, 2 > 94 .
Definition 2. We call Hypothesis  Conclusion an argument. An argument is valid if we can infer the
Conclusion given Hypotheses1 ,. . .,Hypothesesk and invalid otherwise.
Example 4. In this example, we prove  (p → q) → ((q → r) → (p → r) (and thus, show that it is a valid
argument).
1 p→q assumption
2 q→r assumption
3 p assumption
4 q 1, 3, →E
5 r 2, 4, →E
6 p→r 3 − 5, →I
7 (q → r) → (p → r) 2 − 6, →I
8 (p → q) → ((q → r) → (p → r)) 1 − 7, →I
Exercise 1. Show that (p ∧ q) → r  p → (q → r).
At first, there is no reason to believe that if we can prove P from no premise using rules of inference
that P should be a tautology. And perhaps even less believable is the fact that all tautologies can be proven
using these rules. These concepts are refered to as soundness and completeness respectively. We formalise
these ideas with the |= symbol.
Definition 3. Hypothesis1 ,. . .,Hypothesisk |= Conclusion if for any value of the variables which makes
Hypothesis1 ,. . .,Hypothesisk true, the Conclusion is true.
P |= Q is read as “P models Q”.
Example 5. The following table shows that ¬q, p → q |= ¬p.

p q ¬q p→q ¬p
T T F T F
T F T F F
F T F T T
F F T T T

3
 The hypothesis are all true only when p and q are both false and in that case, ¬p is indeed true. Thus,
¬q, p → q |= ¬p.
We can now write what we would like to be true.
Theorem 2 (Soundness).

P1 , . . . , Pk  Q ⇒ P1 , . . . , Pk |= Q
Theorem 3 (Completeness).

P1 , . . . , Pk |= Q ⇒ P1 , . . . , Pk  Q