You are on page 1of 8

5

Propositional Logic: Consistency and


completeness
Reading: Metalogic Part II, 24, 15, 28-31

Contents
5.1 Soundness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2 Consistency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.3 Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.3.1 An Axiomatization of Propositional Logic . . . . . . 63
5.3.2 Kalmar’s Proof: Informal Exposition . . . . . . . . . . . 66
5.3.3 Kalmar’s Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.4 Homework Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.4.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.4.2 Answers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

5.1 Soundness

In this section, we establish the soundness of the system, i.e.,


Theorem 3 (Soundness). Every theorem is a tautology, i.e., If ` A then
|= A.
Proof The proof is by induction the length of the proof of A. For the Basis
step, we show that each of the axioms is a tautology. For the induction step,
we show that if A and A ⊃ B are tautologies, then B is a tautology.

Case 1 (PS1):
A B B ⊃ A (A ⊃ (B ⊃ A))
TT T T
TF T T
FT F T
F F T T
Case 2 (PS2)
62 5 Propositional Logic: Consistency and completeness

X Y Z
A B C B ⊃ C A ⊃ (B ⊃ C) A ⊃ B A ⊃ C Y ⊃ Z X ⊃ (Y ⊃ Z)
TTT T T T T T T
TTF F F T F F T
TFT T T F T T T
TFF T T F F T T
FTT T T T T T T
FTF F T T T T T
F FT T T T T T T
F FF T T T T T T
Case 3 (PS3)
AB ∼B ∼ A ∼ B ⊃∼ A A ⊃ B (∼ B ⊃∼ A) ⊃ (A ⊃ B)
TT F F T T T
TF T F F F T
FT F T T T T
F F T T T T T
Case 4 (MP). If A is a tautology, i.e., true for every assignment of truth
values to the atomic letters, and if A ⊃ B is a tautology, then there is no
assignment which makes A T and B F. Since every assignment makes A T,
every assignment must also make B T; so B must also be a tautology. ¤

5.2 Consistency
From Church (1956):
The notion of consistency is semantically motivated in that one wants one’s
system to be free from absurdity or contradiction. It is, however, a syntactical
notion, and there are several ways of defining it.
Definition 29 A logical system is Consistent with Respect to a partic-
ular transformation by which each sentence or propositional form A is trans-
formed into a sentence or propositional form A0 , if there is no sentence or
propositional form A such that ` A and ` A0 .
Lemma 5.2.1 The Propositional Calculus with respect to the transformation
of A into ∼ A.

Proof. By the definition of tautology, it is not the case both that A and ∼ A
are tautologies. In fact, if A is a tautology, ∼ A is a contradiction. Since every
theorem is a tautology, it cannot be both that ` A and `∼ A. QED.
Definition 30 A logical system is Absolutely Consistent if not all its sen-
tences and propositional forms are theorems.
Lemma 5.2.2 The Propositional Calculus is absolutely consistent.
5.3 Completeness 63

Proof. The wff A∧ ∼ A is not a tautology, and since every theorem is a


tautology, 6` (A∧ ∼ A). So there is at least one wff that is not provable. QED.
Definition 31 A logical system is Consistent in the sense of Post (with
respect to a certain category of primitive symbols designated as ‘propositional
variables’) if a wff consisting of a propositional variable alone is not a theorem.
Lemma 5.2.3 The Propositional Calculus is consistent in the sense of Post
Proof. A wff consisting of a propositional variable alone is not a tautology
because its value for the value F of the variable is F. Since every theorem is
a tautology, a propositional variable is not a theorem. QED.

5.3 Completeness
Historical Note. Before the notion of a structure or a model had become clear,
logicians had attempted to define the notion of completeness syntactically, although,
as we all know, its motivation was semantical. Here (following Church) are some of
the definitions proffered.
Definition 32 A Logical system is complete with respect to a given transforma-
tion by which each sentence or propositional form A is transformed into a sentence
or propositional form A0 if, for every sentence or propositional form B, either ` B or
the system, upon addition of B to it as an axiom, becomes inconsistent with respect
to the given transformation.
Definition 33 A logical system is absolutely complete if, for every sentence or
propositional form B, either ` B or the system, upon addition of B to it as an axiom,
becomes absolutely inconsistent.

Definition 34 A logical system is complete in the sense of Post if, for every sen-
tence or propositional form B, either ` B or the system, upon addition of B to it as
an axiom, becomes inconsistent in the sense of Post.

5.3.1 Kalmar’s Proof: Informal Exposition

So, we want to develop rules of inference for each of our connectives. For
example, suppose our wff is A ⊃ B, wh ere A and are each as complicated as
one wants. And suppose that we have our basic statement Xi . Here are the
possibilities:
Xi ABA⊃B
Xi TT T
Xi TF F
Xi FT T
Xi F F T
64 5 Propositional Logic: Consistency and completeness

So, we have the following rules of inference, corresponding to the four lines
on the table:
Xi ⊃ A, Xi ⊃ B ` Xi ⊃ (A ⊃ B)
Xi ⊃ A, Xi ⊃∼ B ` Xi ⊃∼ (A ⊃ B)
Xi ⊃∼ A, Xi ⊃ B ` Xi ⊃ (A ⊃ B)
Xi ⊃∼ A, Xi ⊃∼ B ` Xi ⊃ (A ⊃ B)
Our job is to provide such rules for each of the connectives: CONJUNC-
TION
Xi A B A∧B
Xi TT T
Xi TF F
Xi FT F
Xi F F F
So, we have the following rules of inference, corresponding to the four lines
on the table:
Xi ⊃ A, Xi ⊃ B ` Xi ⊃ (A ∧ B)
Xi ⊃ A, Xi ⊃∼ B ` Xi ⊃∼ (A ∧ B)
Xi ⊃∼ A, Xi ⊃ B ` Xi ⊃∼ (A ∧ B)
Xi ⊃∼ A, Xi ⊃∼ B ` Xi ⊃∼ (A ∧ B)
DISJUNCTION
Xi A B A∨B
Xi TT T
Xi TF T
Xi FT T
Xi F F F
So, we have the following rules of inference, corresponding to the four lines
on the table:
Xi ⊃ A, Xi ⊃ B ` Xi ⊃ (A ∨ B)
Xi ⊃ A, Xi ⊃∼ B ` Xi ⊃ (A ∨ B)
Xi ⊃∼ A, Xi ⊃ B ` Xi ⊃ (A ∨ B)
Xi ⊃∼ A, Xi ⊃∼ B ` Xi ⊃∼ (A ∨ B)
BICONDITIONAL
Xi ABA≡B
Xi TT T
Xi TF F
Xi FT F
Xi F F T
5.3 Completeness 65

So, we have the following rules of inference, corresponding to the four lines
on the table:
Xi ⊃ A, Xi ⊃ B ` Xi ⊃ (A ≡ B)
Xi ⊃ A, Xi ⊃∼ B ` Xi ⊃∼ (A ≡ B)
Xi ⊃∼ A, Xi ⊃ B ` Xi ⊃∼ (A ≡ B)
Xi ⊃∼ A, Xi ⊃∼ B ` Xi ⊃ (A ≡ B)
NEGATION
Xi A ∼ A ∼∼ A
Xi T F T
So, we have the following rules of inference:

Xi ⊃ A ` Xi ⊃∼∼ A

Next, we put in the following axiom schema:

(X1 ∧ X2 ∧ · · · ∧ Xn ) ⊃ Xi

where 1 ≤ i ≤ n.
Next, to work out tautologies, the second part of our requirement, we
include the following rules:

(A ∧ B) ⊃ C, (A∧ ∼ B) ⊃ C ` A ⊃ C

B ⊃ C, ∼ B ⊃ C ` C
This system is relatively cumbersome, but it can be simplified and then
shown to be sound and complete. That is the heart of Kalmar’s pro of.

5.3.2 Kalmar’s Proof

In this section, we are going to describe a completeness theorem due to


Kalmar. This theorem is interesting because it shows how the notion of proof
is intimately tied to the truth table idea. Let A1 , . . . , An be the atomic letters
in a wff A, and let v be a truth value assignment to A1 , . . . , An . If v(Ai ) = T
then αi = Ai ; if v(Ai ) = F then αi =∼ Ai Then, we show

α1 , . . . , αn ` α

For example, consider the wff A ⊃ B. We have the following truth table
ABA⊃B
TT T
TF F
FT T
F F T
66 5 Propositional Logic: Consistency and completeness

So, we have to show the following, corresponding to each line of the table:

A, B ` A ⊃ B

A, ∼ B `∼ (A ⊃ B)
∼ A, B ` A ⊃ B
∼ A, ∼ B ` A ⊃ B
Supposing that we have this truth table lemma, we can prove completeness
as follows:
Claim (Completeness). If a wff A is a tautology, then it is a theorem.
Proof. Let A be a tautology, and let A1 , . . . , An be the statement letters in
it. For a given assignment v, we know that

α1 , . . . , αn ` α

And since A is a tautology, we have

α1 , . . . , αn ` A

Now, since the assignment to each letter is independent, we have

α1 , . . . , αn−1 , An ` A

and
α1 , . . . , αn−1 , ∼ An ` A
So, by the Deduction Theorem, we have

α1 , . . . , αn−1 ` An ⊃ A

and
α1 , . . . , αn−1 `∼ An ⊃ A
And, from the lemma and MP, we have

α1 , . . . , αn−1 ` A

Next, we eliminate αn−1 , and after n steps, we get

`A

So, if A is a tautology, i.e., if |= A, then ` A.


So, we only have to prove our lemma
Lemma 5.3.1 Let A1 , . . . , An be the atomic letters in a wff A, and let v
be a truth value assignment to A1 , . . . , An . If v(Ai ) = T then αi = Ai ; if
v(Ai ) = F then αi =∼ Ai Then,

α1 , . . . , αn ` α
5.3 Completeness 67

Proof by induction on the number of occurrences of logical connectives in A.


Case 1 n=0. Then the lemma reduces to showing either
A`A
or
∼ A `∼ A
and this is trivial.
Case 2 A =∼ B, where B has fewer than n connectives.
Case 2a v(B) = T . So, v(A) = F . By the induction hypothesis,
α1 , . . . , αn ` B
And by lemma x and MP, we have
α1 , . . . , αn `∼∼ B
Now, since v(A) = F , α =∼ A =∼∼ B. QED
Case 2b v(B) = F . So, v(A) = T . By the induction hypothesis,
α1 , . . . , αn `∼ B
But since v(A) = T , α = A =∼ B. QED.
Case 3 A = B ⊃ C, where B and C each have fewer than n connectives.
Case 3a v(B) = F . So, v(A) = v(B ⊃ C) = T . By the induction hypothesis,
α1 , . . . , αn `∼ B
But, by lemma x and MP, we have
α1 , . . . , αn ` B ⊃ C
QED
Case 3b v(C) = T . So, v(A) = v(B ⊃ C) = T . By the induction hypothesis,
α1 , . . . , αn ` C
And by lemma x and MP, since C ` B ⊃ C, we have
α1 , . . . , αn ` B ⊃ C
QED
Case 3c v(B) = T and v(C) = F . So, v(A) = v(B ⊃ C) = F . Then we have
α1 , . . . , αn ` B
and
α1 , . . . , αn `∼ C
So, by lemma x and MP, we have
α1 , . . . , αn `∼ (B ⊃ C)
But α =∼ (B ⊃ C). QED.
68 5 Propositional Logic: Consistency and completeness

5.4 Homework Exercises

5.4.1 Questions

5.4.2 Answers

You might also like