Introduction to Logic (AI)

Lecture 6

Davide Grossi

7 December 2018
Introduction to Logic (AI)
Lecture 6

Davide Grossi

7 December 2018

∀x (Need(you, x ) → Love(x ))
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege
Translations of complex noun phrases
Semantics and the Hintikka game
Reminder: formal proofs, all rules so far
Universal quantifier rules
Universal instantiation and ∀ Elim
Universal generalization and ∀ Intro
Existential quantifier rules
Existential generalization and ∃ Intro
Existential instantiation and ∃ Elim
Proof strategies
Quantifiers in English

I Every
I Many
I Some
I No
I Few
I Most
Quantifiers in English

I Everyone is valuable.
I Every student is at least 18 years old.
I Many students work hard.
I Some exams are doable.
I No book is perfect.
I Few books are dull.
I Most students like logic.
Some complex quantified expressions in English

The following noun phrases in red are called quantified expressions,

and the sentences containing them quantified sentences.

I At least two major national football teams have not

participated in the last World Cup.
I Every Dutch person is deemed to know the law.
I Every natural number has a unique decomposition into prime
factors.
I At least half of the times when Bob asks her a question, Alice
knows the answer.
Extending the language of FOL

Ingredients of the language of FOL up to now:

I constants a, b, c, . . .
Extending the language of FOL

Ingredients of the language of FOL up to now:

I constants a, b, c, . . .
I functions f, g, h, mother, +, . . .
I terms a, f(a), f(f(a)), g(b,c), g(g(f(c),c),f(g(a,b))), . . .
Extending the language of FOL

Ingredients of the language of FOL up to now:

I constants a, b, c, . . .
I functions f, g, h, mother, +, . . .
I terms a, f(a), f(f(a)), g(b,c), g(g(f(c),c),f(g(a,b))), . . .
I predicates P, Q, A, B, . . .
I connectives ¬, ∧, ∨, →, ↔
I sentences P(a), Q(b)→ R(a,b), . . .
Extending the language of FOL

Ingredients of the language of FOL up to now:

I constants a, b, c, . . .
I functions f, g, h, mother, +, . . .
I terms a, f(a), f(f(a)), g(b,c), g(g(f(c),c),f(g(a,b))), . . .
I predicates P, Q, A, B, . . .
I connectives ¬, ∧, ∨, →, ↔
I sentences P(a), Q(b)→ R(a,b), . . .
To be added:
I variables x, y, z, . . .
I quantifiers ∀, ∃
I well-formed formulas (wffs) P(x), A(y,a)∨B(x), . . .
Variables and atomic formulas

Variables: u, v, w, x, y, z (in Tarski’s World)

Variables can also occur with subscripts: t1 , t2 , t3 . . . , x1 , x2 , . . . , y
(not in Tarski’s World; OK for Fitch)
Variables and atomic formulas

Variables: u, v, w, x, y, z (in Tarski’s World)

Variables can also occur with subscripts: t1 , t2 , t3 . . . , x1 , x2 , . . . , y
(not in Tarski’s World; OK for Fitch)

Variables do not refer to objects. They are placeholders for the

arguments of predicates.
Variables and atomic formulas

Variables: u, v, w, x, y, z (in Tarski’s World)

Variables can also occur with subscripts: t1 , t2 , t3 . . . , x1 , x2 , . . . , y
(not in Tarski’s World; OK for Fitch)

Variables do not refer to objects. They are placeholders for the

arguments of predicates.

Atomic formulas (atomic well-formed formulas, wffs) in FOL

B(a), R(c,d), Q(b,f,e,d,a), a=b
These contain no free variables so they are atomic sentences
Variables and atomic formulas

Variables: u, v, w, x, y, z (in Tarski’s World)

Variables can also occur with subscripts: t1 , t2 , t3 . . . , x1 , x2 , . . . , y
(not in Tarski’s World; OK for Fitch)

Variables do not refer to objects. They are placeholders for the

arguments of predicates.

Atomic formulas (atomic well-formed formulas, wffs) in FOL

B(a), R(c,d), Q(b,f,e,d,a), a=b
These contain no free variables so they are atomic sentences

P(x), a=y, Q(b,f,x,y,x), Cube(x)

these are well-formed formulas, but not sentences. They can be
used together with quantifiers to build sentences.
Universal quantifier

∀
Read ∀x as: “For every object x , . . . ”
Universal quantifier

∀
Read ∀x as: “For every object x , . . . ”

∀x Happy(x ): Everyone is happy.

Universal quantification over Happy(x )
Existential quantifier

∃
Read ∃x as: “There is at least one object x such that . . .”
Existential quantifier

∃
Read ∃x as: “There is at least one object x such that . . .”

∃x Tall(x ): Someone is tall.

Existential quantification over Tall(x )
(Well-formed) formulas vs. sentences

Well-formed formulas form a larger class containing the class of all

sentences of FOL.
I A formula may contain variables;
I A sentence is a formula in which every variable is bound by a
quantifier
Inductive definition of well-formed formula of FOL

Well-formed formulas (wffs) can be defined inductively as follows:

1. If A is an n-place predicate symbol and each of t1 , . . . , tn is a
term, then A(t1 , . . . , tn ) is an atomic wff.
2. If P, Q, P1 , . . . , Pn are well-formed formulas, then so are:
I ¬P
I P1 ∧ . . . ∧ Pn
I P1 ∨ . . . ∨ Pn
I P→Q
I P ↔ Q.
3. If P is a well-formed formula and ν is a variable (i.e. one of
t, u, v , w , x , y , z, t1 , t2 . . .), then ∀νP is a well-formed formula,
and all occurrences of ν in ∀νP are said to be bound.
4. If P is a well-formed formula and ν is a variable (i.e. one of
t, u, v , w , x , y , z, t1 , t2 . . .), then ∃νP is a well-formed formula,
and all occurrences of ν in ∃νP are said to be bound.
Free and bound variables

If A is a formula, then ∀x binds all occurrences of x in ∀x A that

were still free in A.
Variables that are not bound are free or unbound.
Free and bound variables

If A is a formula, then ∀x binds all occurrences of x in ∀x A that

were still free in A.
Variables that are not bound are free or unbound.

Example:

P(x ) ∧ ∀x (Q(x ,y ) → R(x ))

Free and bound variables

If A is a formula, then ∀x binds all occurrences of x in ∀x A that

were still free in A.
Variables that are not bound are free or unbound.

Example:

P(x ) ∧ ∀x (Q(x ,y ) → R(x ))

Here (Q(x , y ) → R(x )) is called the scope of ∀x .
Free and bound variables

If A is a formula, then ∀x binds all occurrences of x in ∀x A that

were still free in A.
Variables that are not bound are free or unbound.

Example:

P(x ) ∧ ∀x (Q(x ,y ) → R(x ))

Here (Q(x , y ) → R(x )) is called the scope of ∀x .

Note: One occurrence of a variable cannot be bound by two

quantifiers. For example, ∀x does not bind the last x in the
sentence ∀x ∃xR(x ): the x in ∃xR(x ) is not free, but has already
been bound by ∃x .
Free and bound variables

If A is a formula, then ∀x binds all occurrences of x in ∀x A that

were still free in A.
Variables that are not bound are free or unbound.

Example:

P(x ) ∧ ∀x (Q(x ,y ) → R(x ))

Here (Q(x , y ) → R(x )) is called the scope of ∀x .

Note: One occurrence of a variable cannot be bound by two

quantifiers. For example, ∀x does not bind the last x in the
sentence ∀x ∃xR(x ): the x in ∃xR(x ) is not free, but has already
been bound by ∃x .

A sentence is a formula without free variables.

Some examples of formulas and sentences

I (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a formula with three free

occurrences of x .
You may leave out the outer brackets of
Cube(x ) ∨ Dodec(x ) ∨ Tet(x )
Some examples of formulas and sentences

I (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a formula with three free

occurrences of x .
You may leave out the outer brackets of
Cube(x ) ∨ Dodec(x ) ∨ Tet(x )
I ∀x (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a sentence in which ∀x
binds all three occurrences of x in formula:
(Cube(x ) ∨ Dodec(x ) ∨ Tet(x )).
Some examples of formulas and sentences

I (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a formula with three free

occurrences of x .
You may leave out the outer brackets of
Cube(x ) ∨ Dodec(x ) ∨ Tet(x )
I ∀x (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a sentence in which ∀x
binds all three occurrences of x in formula:
(Cube(x ) ∨ Dodec(x ) ∨ Tet(x )).
Brackets are important here!
Some examples of formulas and sentences

I (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a formula with three free

occurrences of x .
You may leave out the outer brackets of
Cube(x ) ∨ Dodec(x ) ∨ Tet(x )
I ∀x (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a sentence in which ∀x
binds all three occurrences of x in formula:
(Cube(x ) ∨ Dodec(x ) ∨ Tet(x )).
Brackets are important here!
I ∃x Cube(x ) ∧ ∃x Small(x ) is a sentence.
The first ∃x has as scope Cube(x ) and binds the x in it;
the second ∃x has as scope Small(x ) and binds the x in it.
Some examples of formulas and sentences

I (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a formula with three free

occurrences of x .
You may leave out the outer brackets of
Cube(x ) ∨ Dodec(x ) ∨ Tet(x )
I ∀x (Cube(x ) ∨ Dodec(x ) ∨ Tet(x )) is a sentence in which ∀x
binds all three occurrences of x in formula:
(Cube(x ) ∨ Dodec(x ) ∨ Tet(x )).
Brackets are important here!
I ∃x Cube(x ) ∧ ∃x Small(x ) is a sentence.
The first ∃x has as scope Cube(x ) and binds the x in it;
the second ∃x has as scope Small(x ) and binds the x in it.
I ∀x ((Cube(x ) ∧ Small(x )) → ∃y LeftOf (x , y )) is a sentence.
The ∀x binds all three occurrences of x in
((Cube(x ) ∧ Small(x )) → ∃yLeftOf (x , y )).
The ∃y binds the y in LeftOf (x , y ).
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Existential quantifier rules

Proof strategies
Aristotle of Stagira (384–322 v.C.): Syllogisms

From Aristotle (384-322 BC) until the

19th century, most works on logic
were about syllogisms.
Example syllogism
All men are mortal
Socrates is a man
Therefore, Socrates is mortal
Aristotle of Stagira (384–322 v.C.): Syllogisms

From Aristotle (384-322 BC) until the

19th century, most works on logic
were about syllogisms.
Example syllogism
All men are mortal
Socrates is a man
Therefore, Socrates is mortal

Syllogisms are based on 4 types of premises:

I universal affirmative: Every S is a P
I particular affirmative: Some S is a P
I universal negative: No S is a P
I particular negative: Some S is not a P
These are all expressible with quantifiers.
First application of quantifiers: syllogisms

Every S is a P

Some S is a P

No S is a P

Some S is not a P
First application of quantifiers: syllogisms

Every S is a P ∀x (S(x ) → P(x ))

Some S is a P

No S is a P

Some S is not a P
First application of quantifiers: syllogisms

Every S is a P ∀x (S(x ) → P(x ))

Some S is a P ∃x (S(x ) ∧ P(x ))

No S is a P

Some S is not a P
First application of quantifiers: syllogisms

Every S is a P ∀x (S(x ) → P(x ))

Some S is a P ∃x (S(x ) ∧ P(x ))

No S is a P ∀x (S(x ) → ¬P(x ))

Also correct: ¬∃x (S(x ) ∧ P(x ))

Some S is not a P
First application of quantifiers: syllogisms

Every S is a P ∀x (S(x ) → P(x ))

Some S is a P ∃x (S(x ) ∧ P(x ))

No S is a P ∀x (S(x ) → ¬P(x ))

Also correct: ¬∃x (S(x ) ∧ P(x ))

Some S is not a P ∃x (S(x ) ∧ ¬P(x ))

First application of quantifiers: syllogisms

Every S is a P ∀x (S(x ) → P(x ))

Some S is a P ∃x (S(x ) ∧ P(x ))

No S is a P ∀x (S(x ) → ¬P(x ))

Also correct: ¬∃x (S(x ) ∧ P(x ))

Some S is not a P ∃x (S(x ) ∧ ¬P(x ))

These are our modern formulas for the “categorical sentences”

used in syllogisms. At the time of Aristotle, there was no notation
for quantifiers.
Gottlob Frege (1848–1925): inventor of first-order logic
Begriffschrift (1879)
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Existential quantifier rules

Proof strategies
Pay attention when translating “some P’s are Q’s”

∃ x (P(x) → Q(x)) means something different than

∃ x (P(x) ∧ Q(x)).

∃ x (P(x) → Q(x)) is already true if there is at least one object in

the domain of discourse that does not have property P, and also if
there is at least one object that has property Q.

So ∃ x (Square(x) → Circle(x)) is true for a set of geometrical

figures as soon as there is at least one circle among them.

The correct translation of “some Ps are Qs” is ∃ x (P(x) ∧ Q(x))

Translating complex sentences with quantifiers

At least one cute small kitten was eating

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ E(x))
Translating complex sentences with quantifiers

At least one cute small kitten was eating

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ E(x))

Max likes a cute small kitten

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ L(m,x))
Translating complex sentences with quantifiers

At least one cute small kitten was eating

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ E(x))

Max likes a cute small kitten

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ L(m,x))

All cute small kittens were eating

∀ x ((C(x) ∧ S(x) ∧ K(x)) → E(x))
Translating complex sentences with quantifiers

At least one cute small kitten was eating

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ E(x))

Max likes a cute small kitten

∃ x (C(x) ∧ S(x) ∧ K(x) ∧ L(m,x))

All cute small kittens were eating

∀ x ((C(x) ∧ S(x) ∧ K(x)) → E(x))

In general, show as much logical structure as possible. First

determine the domain of discourse, then determine the translation
key, and finally translate the sentence.

The order of the English sentence does not always correspond to

the order of its FOL translation.
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Existential quantifier rules

Proof strategies
Domain of discourse and objects satisfying a formula

Domain of discourse
The domain of discourse is the (nonempty ) collection of objects
that the quantifiers quantify over.
Domain of discourse and objects satisfying a formula

Domain of discourse
The domain of discourse is the (nonempty ) collection of objects
that the quantifiers quantify over.

Definition of satisfaction
Given a formula P(x ), where x is the only unbound variable that
occurs in P(x ). An object d satisfies P(x ) if d has the property
expressed by P(x ) .
Semantics of ∀, ∃
Now we are ready to define the semantics for quantified sentences:
Truth of universal sentence
A formula ∀x P(x ) is true if and only if
every object in the domain of discourse satisfies P(x ).
Semantics of ∀, ∃
Now we are ready to define the semantics for quantified sentences:
Truth of universal sentence
A formula ∀x P(x ) is true if and only if
every object in the domain of discourse satisfies P(x ).

Truth of existential sentence

A formula ∃x P(x ) is true if and only if
at least one object in the domain of discourse satisfies
P(x ).
Semantics of ∀, ∃
Now we are ready to define the semantics for quantified sentences:
Truth of universal sentence
A formula ∀x P(x ) is true if and only if
every object in the domain of discourse satisfies P(x ).

Truth of existential sentence

A formula ∃x P(x ) is true if and only if
at least one object in the domain of discourse satisfies
P(x ).

Note: If every object in the domain of discourse has a name: a, b,

c, . . . , then
∀x P(x ) corresponds to P(a) ∧ P(b) ∧ P(c) ∧ . . .

∃x P(x ) corresponds to P(a) ∨ P(b) ∨ P(c) ∨ . . .

Notation P(x) for complex formulas

We often refer to possibly complex formulas of FOL as Q(x) or

P(x), e.g. P(x) may stand for:
∃ y (LeftOf(x,y) ∧ FrontOf(y,x))
Then P(b) would stand for the result of replacing all free
occurrences of x by b:
∃ y (LeftOf(b,y) ∧ FrontOf(y,b))
Notation P(x) for complex formulas

We often refer to possibly complex formulas of FOL as Q(x) or

P(x), e.g. P(x) may stand for:
∃ y (LeftOf(x,y) ∧ FrontOf(y,x))
Then P(b) would stand for the result of replacing all free
occurrences of x by b:
∃ y (LeftOf(b,y) ∧ FrontOf(y,b))
Note that such a variable displayed in Q(x) only stands for the free
occurrences of x in Q. Example: suppose Q(x) is
∃ y (LeftOf(x,y) ∧∃ x Cube(x))
Then Q(c) would stand for the result of replacing all free
occurrences of x by c:
∃ y (LeftOf(c,y) ∧∃ x Cube(x))
First-order logic and second-order logic

Where does the name first-order logic come from?

I First-order quantifiers quantify over objects.
I Second-order quantifiers quantify over properties, which can
be seen as sets of objects: ∃PP(a). Second-order logic is an
interesting field, but it is not treated in this introductory logic
course
Hintikka game: Reminder of the rules
You can use the Hintikka game to find out the truth value of
complex sentences in a given situation.
I There are two players: you and the opponent.
I There are two roles: Abelard (commit to false) and Eloise
(commit to true).
Game rules for the connectives ∨, ∧, ¬ and for atomic sentences:
A ∨ B Eloise (commit to true) chooses A or B and the
game continues with that choice.
A ∧ B Abelard (commit-to-false) chooses A or B and the
game continues with that choice.
¬A The players swap roles; the game continues with A.
A → B is treated as abbreviation for ¬A ∨ B
A ↔ B is treated as abbreviation for (¬A ∨ B) ∧ (¬B ∨ A)
atomic sentence For any atomic sentence such as Large(a),
Eloise (commit-to-true) wins if the sentence is true;
Abelard (commit-to-false) wins if the sentence is
false.
Hintikka game: Reminder of winning strategies

It can be proven that:

Truth
A sentence is true if and only if Eloise can win the game, no
matter how Abelard plays.

Falsehood
A sentence is false if and only if Abelard can win the game, no
matter how Eloise plays.
Game rule for the universal quantifier ∀

∀xP(x ) Abelard (commit-to-false) chooses an object (with

the name c) and the game continues with P(c)
Game rule for the universal quantifier ∀

∀xP(x ) Abelard (commit-to-false) chooses an object (with

the name c) and the game continues with P(c)

Mnemonics: ∀BELARD
Checking truth of a sentence with ∀ in a situation
Is the following sentence true or false in the world below?
Find out by playing the Hintikka game.
∀x (¬ Cube(x) ∨ Between(d,e,x) ∨ Between(c,d,x))
Checking truth of a sentence with ∀ in a situation
Answer: Abelard can win the game for the sentence below by
choosing object c, so the sentence is false.
∀x (¬ Cube(x) ∨ Between(d,e,x) ∨ Between(c,d,x))
Game rule for the existential quantifier exists

∃xP(x ) Eloise (commit-to-true) chooses an object (with the

name c) and the game continues with P(c)
Game rule for the existential quantifier exists

∃xP(x ) Eloise (commit-to-true) chooses an object (with the

name c) and the game continues with P(c)

Mnemonics: ∃LOISE
Checking truth of a sentence with ∃ in a situation
Are the following sentences true or false in the world below?
Find out by playing the Hintikka game
∃x ((Large(x) ∨ Medium(x)) ∧ Tet(x) ∧¬(a=b))
∃x ((Large(x) ∨ Medium(x)) ∧ Tet(x) ∧¬(x=e))
Checking truth of a sentence with ∃ in a situation

Abelard wins no matter what Eloise chooses, simply because ¬

(a=b) is false:
∃x ((Large(x) ∨ Medium(x)) ∧ Tet(x) ∧¬(a=b))
Eloise wins the game for the following sentence by choosing d:
∃x ((Large(x) ∨ Medium(x)) ∧ Tet(x) ∧¬(x=e))
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Existential quantifier rules

Proof strategies
Summary of proof rules: Reiteration

For a summary of all proof rules, see also LPL pp.573-577.

..
.
j. P Justification (or premise)
..
.
k. P Reit: j
..
.
= Introduction

..
.
k. a=a = Intro
..
.
= Elimination

..
.
i. P(a) Justification (or premise)
..
.
j. a=b Justification (or premise)
..
.
k. P(b) =Elim: i,j
..
.

Note that you may replace a by b in one place in P(a), or in more

places if it occurs more times. The choice is yours.
Note that in the justification, lines i, j appear in that order.
∧ Introduction (general)

..
.
i. P1 Justification (or premise)
⇓
j. Pn Justification (or premise)
..
.
k. P1 ∧ . . . ∧ Pn ∧ Intro: i,. . . , j
..
.
∧ Elimination (general)

..
.
i. P1 ∧ . . . ∧ Pn Justification (or premise)
..
.
k. Pj ∧ Elim: i
..
.
¬ Introduction

..
.

..
.

i. P
..
.

j. ⊥ (Justification)
k. ¬P ¬ Intro: i–j
..
.
¬ Elimination

..
.
i. ¬¬P Justification (or premise)
..
.
k. P ¬ Elim: i
..
.
⊥ Introduction

..
.
i. P Justification (or premise)
..
.
j. ¬P Justification (or premise)
..
.
k. ⊥ ⊥ Intro: i, j
..
.
⊥ Elimination

..
.
i. ⊥ Justification (or premise)
..
.
k. P ⊥ Elim: i
..
.
∨ Introduction (general)

..
.
i. Pj Justification (or premise)
..
.
k. P1 ∨ . . . ∨ Pn ∨ Intro: i
..
.
∨ Elimination (general)
..
.

i. P1 ∨ . . . ∨ Pn Justification (or premise)

j. P1
..
.

l. R (Justification)
⇓
m. Pn
..
.

n. R (Justification)
k. R ∨ Elim: i, j–l,. . . ,m–n
..
.
→ Introduction
..
.

..
.

i. P
..
.

j. Q (Justification)
..
.

k. P → Q → Intro: i–j
..
.
→ Elimination

..
.
i. P→Q Justification (or premise)
..
.
j. P Justification (or premise)
..
.
k. Q → Elim: i, j
..
.
↔ Introduction
..
.

..
.

i. P
..
.

j. Q (Justification)
m. Q
..
.

n. P (Justification)
k. P ↔ Q ↔ Intro: i–j,m–n
..
↔ Elimination

..
.
i. P ↔ Q or: Q ↔ P Justification (or premise)
..
.
j. P Justification (or premise)
..
.
k. Q ↔ Elim: i,j
..
.
About subproofs

When a subproof has ended:

I you may cite the subproof as a whole in a ¬ Introduction or in
an ∨ elimination . . .
About subproofs

When a subproof has ended:

I you may cite the subproof as a whole in a ¬ Introduction or in
an ∨ elimination . . .
I but you may not use individual steps from it!
About subproofs

When a subproof has ended:

I you may cite the subproof as a whole in a ¬ Introduction or in
an ∨ elimination . . .
I but you may not use individual steps from it!
So you may use only steps from (sub)proofs that are still open (i.e.
not ended), and if you still are “in” that subproof.
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Universal instantiation and ∀ Elim
Universal generalization and ∀ Intro

Existential quantifier rules

Informal proof idea: Universal instantiation
Everyone has overslept at least once.
Carlo has overslept at least once.

Translation key (this time without representing all the logical

structure in the sentence):

Domain: all people

S(x ): x has overslept at least once.

c: Carlo
Informal proof idea: Universal instantiation
Everyone has overslept at least once.
Carlo has overslept at least once.

Translation key (this time without representing all the logical

structure in the sentence):

Domain: all people

S(x ): x has overslept at least once.

c: Carlo
Informal rule of universal instantiation
Suppose we have ∀xS(x ).
Let c be an individual constant, that is, a name of an object in the
universe of discourse.
Informal proof idea: Universal instantiation
Everyone has overslept at least once.
Carlo has overslept at least once.

Translation key (this time without representing all the logical

structure in the sentence):

Domain: all people

S(x ): x has overslept at least once.

c: Carlo
Informal rule of universal instantiation
Suppose we have ∀xS(x ).
Let c be an individual constant, that is, a name of an object in the
universe of discourse.
Then we may conclude S(c).
Formal proof rule:
Universal quantifier elimination / ∀ Elim

..
.
i. ∀xP(x ) Justification (or premise)
..
.
k. P(c) ∀ Elim: i
..
.

Compare this with general ∧-Elimination:

P(a) ∧ P(b) ∧ P(c) ∧ P(d) ∧ . . .
P(c) ∧ Elim
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Universal instantiation and ∀ Elim
Universal generalization and ∀ Intro

Existential quantifier rules

Informal proof idea: Universal generalization

Informal rule of universal generalization

How to prove ∀xS(x ) ?
Informal proof idea: Universal generalization

Informal rule of universal generalization

How to prove ∀xS(x ) ?

Let c be the name of an abitrarily chosen object of the universe of

discourse.

Then . . . [some reasoning] . . . S(c).

Since c was chosen arbitrarily, we conclude ∀xS(x ).

Informal proof idea: Universal generalization

Informal rule of universal generalization

How to prove ∀xS(x ) ?

Let c be the name of an abitrarily chosen object of the universe of

discourse.

Then . . . [some reasoning] . . . S(c).

Since c was chosen arbitrarily, we conclude ∀xS(x ).

‘Object c is arbitrarily chosen’ means:

We do not make any assumptions about the properties of c
(only that it belongs to the domain of discourse).
Informal proof idea: Universal generalization

Informal rule of universal generalization

How to prove ∀xS(x ) ?

Let c be the name of an abitrarily chosen object of the universe of

discourse.

Then . . . [some reasoning] . . . S(c).

Since c was chosen arbitrarily, we conclude ∀xS(x ).

‘Object c is arbitrarily chosen’ means:

We do not make any assumptions about the properties of c
(only that it belongs to the domain of discourse).
Example:
I Take an arbitrary object. Let’s call it c.
I c is identical to itself.
I So every object is identical to itself.
Formal proof rule:
Universal quantifier introduction / ∀ Introduction

..
.

i. c
..
.

j. P(c)
∀xP(x ) ∀ Intro: i–j

c does not occur outside the subproof where it is introduced.

This warrants that c is indeed chosen arbitrarily.
Example: The informal argumentation formalized

I Take an arbitrary object. Let’s call it c.

I c is identical to itself.
I So every object is identical to itself.
In a proof an object is said to be arbitrary if we do not make any
assumptions about the properties of the object.
Example: The informal argumentation formalized

I Take an arbitrary object. Let’s call it c.

I c is identical to itself.
I So every object is identical to itself.
In a proof an object is said to be arbitrary if we do not make any
assumptions about the properties of the object.

Formalizing the above argument:

c
..
.

c=c
∀x (x = x )
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Existential quantifier rules

Existential generalization and ∃ Intro
Existential instantiation and ∃ Elim
Informal proof idea: Existential generalization

Informal rule of existential generalization

Suppose we have S(c), where c is an individual constant, i.e. a
name of an object in the universe of discourse.
Informal proof idea: Existential generalization

Informal rule of existential generalization

Suppose we have S(c), where c is an individual constant, i.e. a
name of an object in the universe of discourse.

Then we may conclude ∃xS(x ).

Informal proof idea: Existential generalization

Informal rule of existential generalization

Suppose we have S(c), where c is an individual constant, i.e. a
name of an object in the universe of discourse.

Then we may conclude ∃xS(x ).

Example
a is a small cube
There exists a small cube.

Cube(a) ∧ Small(a)
∃x (Cube(x ) ∧ Small(x ))
Formal proof rule:
Existential quantifier introduction / ∃ Intro

..
.
i. P(c) Justification (or premise)
..
.
k. ∃xP(x ) ∃ Intro: i
..
.
Formal proof rule:
Existential quantifier introduction / ∃ Intro

..
.
i. P(c) Justification (or premise)
..
.
k. ∃xP(x ) ∃ Intro: i
..
.

Compare with
P(c)
P(a) ∨ P(b) ∨ P(c) ∨ P(d) ∨ . . . ∨ Intro
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege

Translations of complex noun phrases

Semantics and the Hintikka game

Reminder: formal proofs, all rules so far

Universal quantifier rules

Existential quantifier rules

Existential generalization and ∃ Intro
Existential instantiation and ∃ Elim
Informal proof idea: Existential instantiation

Informal rule of existential instantiation

Suppose we have ∃xS(x ). How can we use this in a proof?
Informal proof idea: Existential instantiation

Informal rule of existential instantiation

Suppose we have ∃xS(x ). How can we use this in a proof?

Let c be the name of an arbitrarily chosen object of the universe of

discourse. Assume that S(c) holds.

Then . . . [some reasoning] . . . P.

So we have S(c) → P for an arbitrarily chosen object c.

But we know that ∃xS(x ), and we conclude: P.

Example argumentation based on existential instantiation

I Someone put the empty milk carton back in the fridge,

finished the cheese and didn’t put the lid back on the butter.
Example argumentation based on existential instantiation

I Someone put the empty milk carton back in the fridge,

finished the cheese and didn’t put the lid back on the butter.
I Whoever did that should go to the store to buy new milk,
cheese and butter.
Example argumentation based on existential instantiation

I Someone put the empty milk carton back in the fridge,

finished the cheese and didn’t put the lid back on the butter.
I Whoever did that should go to the store to buy new milk,
cheese and butter.
I For simplicity, let’s call this person “the slob”.
Example argumentation based on existential instantiation

I Someone put the empty milk carton back in the fridge,

finished the cheese and didn’t put the lid back on the butter.
I Whoever did that should go to the store to buy new milk,
cheese and butter.
I For simplicity, let’s call this person “the slob”.
I So the slob has to go to the store to buy new milk, cheese
and butter.
Example argumentation based on existential instantiation

I Someone put the empty milk carton back in the fridge,

finished the cheese and didn’t put the lid back on the butter.
I Whoever did that should go to the store to buy new milk,
cheese and butter.
I For simplicity, let’s call this person “the slob”.
I So the slob has to go to the store to buy new milk, cheese
and butter.
I Therefore, someone has to go to the store to buy new milk,
cheese and butter.
Here, “the slob” is only used for simplicity, as a temporary marker
(just like “Jack the Ripper”).
Formal proof rule:
Existential quantifier elimination / ∃ Elim

..
.

i. ∃xP(x )
j. c P(c)
..
.

k. Q
Q ∃ Elim: i, j–k
c does not occur outside the subproof where it is introduced.
(So in particular, c does not occur in Q.)
Formal proof rule:
Existential quantifier elimination / ∃ Elim

..
.

i. ∃xP(x )
j. c P(c)
..
.

k. Q
Q ∃ Elim: i, j–k
c does not occur outside the subproof where it is introduced.
(So in particular, c does not occur in Q.)
This warrants that c is indeed chosen arbitrarily.
∃ Elimination reminds of ∨ Elimination

.. ..
. .

i. ∃xP(x ) i. P(a) ∨ P(b)

j. c P(c) j. P(a)
.. ..
. .

k. Q k. Q
Q m. P(b)
..
.

n. Q
Q
Overview

Quantifiers in English vs. FOL

History of quantifiers in logic: from Aristotle to Frege
Translations of complex noun phrases
Semantics and the Hintikka game
Reminder: formal proofs, all rules so far
Universal quantifier rules
Universal instantiation and ∀ Elim
Universal generalization and ∀ Intro
Existential quantifier rules
Existential generalization and ∃ Intro
Existential instantiation and ∃ Elim
Proof strategies
General tips for making formal proofs

I Keep the goal in sight;

I Determine which rules you can apply to the premises and
assumptions you have made; some “easy” rules are ∧
elimination, ⊥ introduction, ∀ elimination, ∃ introduction;
I Develop your intuition by thinking of informal proofs.
A step-by-step guide to planning your proofs

1. Is →, ↔ or ∀x the main operator of the conclusion? Use the

rule that introduces this operator. If not, continue to step 2.
2. Is ∨ or ∃x the main operator of one of the premises? Use the
rule that Eliminates this operator. If not, continue to step 3.
3. Does the conclusion have a main operator? If so, try to
introduce it. Continue to step 4 otherwise.
4. Try to infer the conclusion from the premises informally. Then
try to translate your informal proof into a formal proof.
5. If everything else fails: Prove the conclusion (B) by
contradiction.
Start a new subproof and suppose that ¬B is true. Try to
infer ⊥. You can now end the subproof and introduce a ¬.
You now have ¬¬B. Eliminate these two negations, and voilà.
Next time

Tuesday, December 11th

Questions and Answers session for the midterm exam of
Wednesday December 12th (9:00-11:00) on all the material of
lectures 1-5 up to sets (so no functions and quantifiers).

Friday, December 14th

Formal semantics: First-order structures and the truth definition.