ITL2: Classical Propositional Logic Pablo Cobreros 1

In the previous lecture we introduced the basic concepts about Logic: what is a sentence,
what is a declarative sentence, what a proposition and what an argument; what does (and
does not) mean that an argument is valid in an informal and in a more elaborated way.
This is the elaborated sense of logical consequence: an argument is valid just in case there is
no interpretation of the non-logical vocabulary making all premises true and the conclusion
false. Now according to this definition, in order to specify a relation of logical consequence,
we need to set: (i) the language (explicitly telling which part of it is logical and which
non-logical vocabulary) and (ii) what is an interpretation for the non-logical vocabulary
and how this extends to all formulas of the language.

Non-logical

Vocabulary
Logical
Language

Grammar
Interpretation
Semantics
Extension

In this lecture we will fill in this schema in order to explain the classical propositional
language. We will, in addition, provide a method to decide for any given argument in the
propositional language, whether it is valid according to the previous definitions.

1. The Classical Propositional Language

The non-logical vocabulary of the classical propositional language consists in an infinite
set of propositional variables. We will use the letters: p, q, r, s, . . . to express these
variables. Propositional variables are like the bricks (the simplest kind of sentence) of
our propositional language. Intuitively, they correspond to simple propositions in natural
language like ‘it is raining’ and ‘Peter came from Madrid’. These propositions are simple
in the sense that don’t contain logical connectives and, therefore, they are not made up of
other propositions.
The logical vocabulary of the classical propositional language includes: conjunction (‘∧’),
disjunction (‘∨’), conditional (‘⊃’) and negation (‘¬’). We will, in addition, make use of a
left bracket ‘(’ and a right bracket ‘)’.

Symbol name Informal reading

∧ conjunction ‘and’
∨ disjunction ‘or’
⊃ conditional ‘If... then...’
¬ negation ‘not’

The logical vocabulary is like the cement that allow us to glue the bricks together to form
complex sentences. For example, from simple sentences,
 ITL2: Classical Propositional Logic Pablo Cobreros 2

(1) Socrates is Athenian

(2) Socrates is Greek

we can form the complex sentence,

(3) If Socrates is Athenian, then Socrates is Greek

which in the propositional language will correspond to a sentence of the form,

(4) (p ⊃ q)

It is important to notice that sentences can be more complex than that, like the following,

(5) If it is not the case that if Socrates is Athenian then Socrates is Greek, then ‘Athenian’
does not mean what it seems.

whose corresponding sentence in the propositional language is,

(6) (¬(p ⊃ q) ⊃ ¬r)

Question 1. What kind of expressions can we find in the propositional language?

Question 2. Say which of the following are simple propositions and which are complex:

• Socrates is the husband of Xantipa.

• The husband of Xantipa was a very well known philosopher.
• If Socrates is the husband of Xantipa then the husband of Xantipa was a very well known
philosopher.
• The sum of the internal angles of a triangle in some non-euclidean geometries is sometimes
bigger than 180o .

From all possible strings of signs of the vocabulary, some will be sentences, some other
won’t. For example, the sequences of signs in (4) and (6) above are sentences belonging
to our formal language (it will be possible to give an interpretation for them, as we shall
see), but the following strings of signs are NOT sentences of our language,

(7) ((p ⊃ q)

(8) p ⊃⊃

(9) q¬p
 ITL2: Classical Propositional Logic Pablo Cobreros 3

(7) is not a formula (yes, this is the name we will give to sentences of our language) since
all formulas will have the same number of left and right parenthesis. (8) is not a formula,
because the conditional (‘⊃’) must be accompanied by formulas at each side. (9) is not a
formula because the negation doesn’t relate two different propositions, it rather applies to
a single proposition.
The strings of symbols of our vocabulary are exactly those that can be constructed following
the next set of rules,
Definition (Grammar): Let A and B be string of symbols of the vocabulary,
− If A is a propositional variable, then A is a formula.
− If A and B are formulas, then (A ∧ B), (A ∨ B), (A ⊃ B) and ¬A are also formulas.
− No other sequence of symbols is a formula.

Question 3. Say which of the following strings of symbols are formulas and why:

• (p)
• ¬(p)
• ¬¬(p ∨ q)
• p⊃q∧r
• (p ⊃ (q ∧ r))
• ¬(p ⊃ (q ∧ r))
• p ⊃ (q ∧ r)

This language is designed to allow us to show the logical structure of certain class of
sentences of natural language and, therefore, the logical structure of certain arguments
in natural language. The process of translating an argument from natural language to a
formal language is called ‘formalization’ and it is one of the crucial tasks of Logic that of
finding an adequate formalization for different sort of arguments. Consider the examples
we met before,

• If Socrates is a man, then Socrates is mortal

Socrates is a man
Therefore,
Socrates is mortal

† If Socrates is a man, then Socrates is mortal

Socrates is mortal
Therefore,
Socrates is a man

The formalization of these arguments in the propositional language is the following,

• p⊃q, p ∴ q

† p⊃q, q ∴ p
ITL2: Classical Propositional Logic Pablo Cobreros 4

We will later see whether these arguments are valid and why.

Question 4. Explain what is the process of formalization and why we will use it in this
course.

2. Semantics
Formulas of our language, like (p∧q), are not sentences (they do not express any particular
proposition) in much the same way the expression n + m does not express any particular
number. Rather, formulas are schemes of propositions and they can express a particu-
lar proposition when we substitute variables by particular propositions (similarly, if we
substitute n and m by particular numbers, the expression n + m will denote a particular
number).
Interpreting a formula like p ∧ q consists in substituting each variable by a declarative
sentence. We can interpret that formula, for example, substituting p by the sentence
‘Socrates was athenian’ and q by the sentence ‘Socrates is Greek’ which would render the
sentence,

(10) Socrates is Athenian and Socrates is Greek

which is a true statement. We could have interpreted variables in a different way, selecting
different sentences to substitute variables with, like in the following interpretation,

(11) Socrates is Greek and Socrates is Theban

In this case the resulting sentence is false.

The intuitive idea of an interpretation, that of substituting variables by declarative sen-
tences, is basically correct. The problem is that it is not clear which is exactly the set of
English declarative sentences and, therefore, it is not clear which are exactly the possible
interpretations for a formula like p ∧ q. For this reason, we will make a huge idealiza-
tion and consider that an interpretation for a given formula is just this: an assignment of
truth-values (true or false and not both) to each variable appearing in the formula. The
motivation for this notion of interpretation is the fact that, in order to evaluate a formula
like p ∧ q in a given interpretation we don’t need to know precisely the meaning of p or
q in that interpretation, but only their truth-value. So, for example, the interpretation of
p ∧ q resulting in (10) will be just the interpretation assigning to p the value ‘true’ and to
q the value ‘true’. The interpretation resulting in (11) will be just that assigning to p the
value ‘true’ and to q the value ‘false’.
From a more mathematical perspective, an interpretation I is a function from the set of
propositional variables V ar to the set of truth values {0, 1}, I : V ar 7→ {1, 0} (where 0 is
to be read as ‘false’ and 1 as ‘true’). We will write I(A) = 1 to mean that the formula A
is assigned value 1 in interpretation I.
Definition (Interpretation). An interpretation I (for a propositional language) is an
assignment of truth-values (true or false and not both) to the propositional variables of
the language.
 ITL2: Classical Propositional Logic Pablo Cobreros 5

Question 5. Explain the main benefit of the sophisticated idea of formalization vs the
intuitive one.
Question 6. Let a formula A contain n distinct propositional variables. Can you tell how
many different interpretations can be found for A?

Interpretations make only reference to propositional variables (the bricks). But once you
give an interpretation for every propositional variable, formulas of the language will receive
a single truth-value according to that interpretation. Consider the following example,

(12) p ∧ q

There are 4 different interpretations for this formula:

I1 (p) = 1 & I1 (q) = 1
I2 (p) = 1 & I2 (q) = 0
I3 (p) = 0 & I3 (q) = 1
I4 (p) = 0 & I4 (q) = 0
The formula in (12) is true in the first interpretation and false in the remaining. The
reason is that a conjunction is true just in case both members of the conjunction are, and
this happens, for the case at hand, just in the first interpretation.
The formula in (12) is relatively simple, since it contains a single logical constant. But
formulas of the language can contain an arbitrarily large (though finite) number of logical
constants. If you give an interpretation for all the variables in a formula, that formula, no
matter how complex, will receive a single truth-value according to that interpretation. The
following definition explains how the truth-value of a formula depends on the truth-value
of its components:
Definition (Extension of an Interpretation).

• I(A ∧ B) = 1 if and only if (that is: exactly when) I(A) = 1 and I(B) = 1.
• I(¬A) = 1 if and only if I(A) = 0.
Question 7. Give the corresponding definitions for the conditional (‘⊃’) and the disjunc-
tion (‘∨’).

3. Logical Consequence
We have all the ingredients in order to specify a logic: the language, what is an interpreta-
tion (for the non-logical vocabulary) and how this extends to all formulas of the language.
Notice that our definition of logical consequence here follows the ideas introduced at the
end of Lecture 1.
Definition (Logical Consequence) A formula B is a logical consequence of a set of
formulas A1 , . . . An , written A1 , . . . An  B, just in case there is no interpretation I such
that I(A) = 1 for all A in A1 , . . . An and I(B) = 0.

Let a counterexample for an argument Γ ∴ A be an interpretation making all premises true

 ITL2: Classical Propositional Logic Pablo Cobreros 6

and the conclusion false. Then an argument is valid just in case there’s no counterexample
to it.

Question 8. The definition of logical consequence above corresponds to the idea “it is
impossible for premises to be true and conclusion false”. What would be the definition
corresponding to “necessarily if premises are true, the conclusion is true”?

Question 9. The definition of logical consequence above includes the possibility of multiple
premises but only one conclusion. How would you extend the definition to cover multiple
conclusions?

Let’s have a look now to the arguments we discussed above,

• If Socrates is a man, then Socrates is mortal

Socrates is a man
Therefore,
Socrates is mortal

† If Socrates is a man, then Socrates is mortal

Socrates is mortal
Therefore,
Socrates is a man

The formalization of these arguments is the following,

• p⊃q, p ∴ q

† p⊃q, q ∴ p

Given that these arguments involve the use of just two distinct propositional variables,
there are only 4 different interpretations. We can have a look to these and check whether
any of them is a counterexample for some of the arguments.

p q (•) p⊃q , p ∴ q (†) p⊃q , q ∴ p

I1 : 1 1 1 1 1 1 1 1
I2 : 1 0 0 1 0 0 0 1
I3 : 0 1 1 0 1 1 1 0
I4 : 0 0 1 0 0 1 0 0

• is a valid argument since, among the 4 different interpretations, no one makes all premises
true and the conclusion false. Interpretation I2 , for example, makes the conclusion false,
but does not make all premises true. On the contrary, argument (†) is NOT valid, since
the interpretation I3 makes all premises true and the conclusion false (that is, I3 is a
counterexample for the argument).
ITL2: Classical Propositional Logic Pablo Cobreros 7

Question 10. Consider the following argument and say whether it is valid in the same
way we did in the arguments above,
p ⊃ q, q ⊃ r ∴ p ⊃ r

4. Trees
The number of interpretations for a formula A, (ni(A)), grows exponentially as a function
of the number of variables in A, (nv(A)). More particularly, ni(A) = 2nv(A) . If we want
to check the validity of an argument just by looking at each possible interpretation, we
would have to check 256 interpretations for an argument with 8 variables and 32768 for
an argument involving 15 variables. For this reason we would like to have some method
to check whether an argument is valid that avoids, as much as possible, the necessity of
reviewing all possible interpretations.
A tree is an ordered set with this shape,

 

  

Elements in the tree are called nodes, the topmost node is called the root and each of the
paths from the root to any of the lowest nodes is called a branch.
Our method for checking whether an argument is valid consists on the construction of a
tree out of the argument we want to check. We will write an initial tree and will use a set
of rules to extend the tree. Once we extend the tree as far as possible, some features of
the tree will tell use whether the initial argument is valid. Our method of trees is almost
algorithmic; we explain first the procedure for trees in all its details and will then explain
the general idea behind it.
Method of trees. Let A1 , . . . , An ∴ B be the argument we want to evaluate. First of all,
write a list with all the premises. Add as another element to the list the negation of the
conclusion.

A1
..
.
An
¬B

Next, we should extend the tree attending to the shape of the formulas in this initial list.
There are two kinds of rules: those that extend the tree in a single branch (do not open a
ITL2: Classical Propositional Logic Pablo Cobreros 8

branch) and those that extend the tree with two branches (open a branch). If, for example,
we have a formula of the form (A ∧ B) then we should extend the tree in a single branch
with the two formulas A and B:

A∧B
|
A
B

If, for example, we have a formula of the form (A ∨ B), then we have to extend the tree
with two branches: one in which we write A and another one in which we write B

A∨B

A B

The meaning of opening or not opening a branch is the following. The way the tree extends
records the truth-conditions of the formulas before the extension. A formula of the form
A ∧ B is true just in case both subformulas A and B are true; for this reason we write
both A and B in the same branch. A formula of the form A ∨ B is true just in case some
subformula A or B is true. In this second case, it could be A that is true and so A ∨ B is
true, or it could be B that is true, making A ∨ B true. We know that at least one of A
and B is true, but we don’t know which. For this reason we need to examine separately
what happens when A is true and what happens when B is true (this general form of
reasoning is called reasoning by cases).

Table 1: Non-branching rules

(A ∧ B) ¬(A ∨ B) ¬(A ⊃ B) ¬¬A

A ¬A A A
B ¬B ¬B

Table 2: Branching rules

¬(A ∧ B) (A ∨ B) (A ⊃ B)
¬A ¬B A B ¬A B

Question 11. Explain why some rules branch and others don’t.

We say a branch is closed when it contains a formula and its negation. When a branch
closes we will indicate this writing the symbol ‘×’ under the last node of the branch.
A branch is complete (it cannot be extended further) when it is either closed or there are
no formulas left to analyse.
A tree is closed when all its branches are closed. A tree is complete when all its branches
are. Note that a closed tree is complete, but a complete tree need not be closed.
ITL2: Classical Propositional Logic Pablo Cobreros 9

We will say that there is a proof from A1 , . . . An to B (according to the method of trees),
written A1 , . . . An ` B just in case there is a closed tree for the initial list A1 , . . . An , ¬B.

Example. (p ∧ q) ` (p ∨ q):

(p ∧ q)
¬(p ∨ q)

p
q

¬p
¬q
×

Example. p ⊃ q ` (p ⊃ r) ∨ (s ⊃ q)

p⊃q
¬((p ⊃ r) ∨ (s ⊃ q))
¬(p ⊃ r)
¬(s ⊃ q)
p
¬r
s
¬q
¬p q
× ×

Question 12. Consider the rules for A ∧ B and A ∨ B and looks at their truth-tables. Can
you see a connection between the truth-tables and the shape of the rule?

Meaning of the method of trees. When we write a formula in our tree we are saying
that the formula is true. The first step in the construction of the tree requires writing
the premises together with the negation of the conclusion. This is like saying: “Ok, let’s
suppose that all premises are true and the conclusion false” (note that saying “the conclusion
is false” is the same as saying “the negation of the conclusion is true”). In other words, the
first step is the assumption that there is a counterexample to the argument.
Next we start chopping formulas extending the tree following our rules above. The rules
record the truth-condition of the formulas. For example, in order for ¬(A ∨ B) to be true
(equivalent to: in order for A ∨ B to be false) we need both A and B being false, and so
we write both ¬A and ¬B extending the tree in a single branch. In order for (A ⊃ B) to
be true we need either A being false or B being true, and so we branch the tree writing
¬A in one branch and B in the other.
Each branch of the tree, for a given argument, represents an attempt to find an inter-
pretation making all premises true and the conclusion false (that is, an attempt to find
a counterexample to that argument). When a branch closes, the tree is telling you that
 ITL2: Classical Propositional Logic Pablo Cobreros 10

you won’t find a counterexample that way. The branching of the tree, when appropriate,
guarantees that no counterexample escapes to this analysis and so, if all branches close,
that means that there is no counterexample to the argument (the argument is valid).
Example. p ⊃ (q ∧ r), ¬r ` ¬p

p ⊃ (q ∧ r)
¬r
¬¬p

¬p q∧r
×
q
r
×

A complete open tree (a non-closed complete tree) is telling you that the argument has at
least one counterexample. Any of the open branches of a complete open tree gives you the
information to find a counterexample to the argument. We need to follow the next recipe.
For each propositional variable p in the argument:

• if p appears in the branch, write I(p) = 1,

• if ¬p appears in the branch, write I(p) = 0 and
• if neither p nor ¬p appear in the branch, you can freely choose a truth-value for p.

Example: ((p ⊃ q) ⊃ s) 0 ¬s ⊃ ¬p
Proof. ((p ⊃ q) ⊃ s)
¬(¬s ⊃ ¬p)
¬s
¬¬p Counterexample:
p I(p) = 1
I(q) = 0
¬(p ⊃ q) s I(s) = 0
p ×
¬q

An interpretation constructed out of a complete open branch in a tree is often called the
counterexample induced by that branch. Our method of trees works well in the sense
that if an argument has a closed tree, then there is no counterexample to that argument
and, conversely, if there is a counterexample to an argument, no tree for that argument
closes. These two properties are called soundness and completeness:

Soundness. If A1 , . . . An ` B then A1 , . . . An  B.
Completeness. If A1 , . . . An  B then A1 , . . . An ` B
See (Priest, 2008, 16-18) for more on this.
ITL2: Classical Propositional Logic Pablo Cobreros 11

Question 13. Explain the idea behind the method of trees.

Question 14. Explain why some rules open a branch while other don’t.

Question 15. Consider the following definition of logical consequence (multiple conclu-
sions):
A1 , . . . An  B1 , . . . Bm just in case, for all interpretations I, if I(A) = 1 for all A, then
I(B) = 1 for some B.
Explain how to adapt the tree method to multiple conclusions.

Question 16. Suppose b is a complete open branch in a tree. Let Ib be the counterexample
induced by b. Explain why for any A in b, Ib (A) = 1.

References
Priest, G. (2008). An Introduction to Non-Classical Logic: From If to Is. Cambridge
University Press.