A Problem Course in

Mathematical Logic
Version 1.5

Volume I
Propositional and
First-Order Logic

Stefan Bilaniuk
Author address:
Department of Mathematics
Trent University
Peterborough, Ontario
Canada K9J 7B8
E-mail address: sbilaniuk@trentu.ca
 1991 Mathematics Subject Classification. 03.

A Problem Course In Mathematical Logic

Volume I: Propositional and First-Order Logic
Version 1.5
Copyright 1994–1999
c by Stefan Bilaniuk.

Abstract. This is a text for a problem-oriented undergraduate

course in mathematical logic. It covers the basics of propositional
and first-order logic through the Soundness, Completeness, and
Compactness Theorems. Volume II, Computation, covers the ba-
sics of computability using Turing machines and recursive func-
tions, the Incompleteness Theorems, and complexity theory through
the P and NP.
Information on availabality and the conditions under which
this book may be used and reproduced are given in the preface.
This book was typeset using LATEX, using the AMS-LATEX and
AMSFonts packages of the American Mathematical Society.
 Contents

Preface v
Introduction 1

Propositional Logic 5
Chapter 1. Language 7
Chapter 2. Truth Assignments 11
Chapter 3. Deductions 15
Chapter 4. Soundness and Completeness 19

First-Order Logic 21
Chapter 5. Languages 23
Chapter 6. Structures and Models 33
Chapter 7. Deductions 41
Chapter 8. Soundness and Completeness 47
Chapter 9. Applications of Compactness 51

Hints 57
Chapter 1. Hints 59
Chapter 2. Hints 61
Chapter 3. Hints 63
Chapter 4. Hints 65
Chapter 5. Hints 67
Chapter 6. Hints 69
Chapter 7. Hints 71
iii
iv CONTENTS

Chapter 8. Hints 73
Chapter 9. Hints 75

Appendices 77
Appendix A. A Little Set Theory 79
Appendix B. The Greek Alphabet 81
Appendix C. Logic Limericks 83
Bibliography 85
Index 87
 Preface

This book is intended to be the basis for a problem-oriented full-year

course in mathematical logic for students with a modicum of mathe-
matical sophistication. Volume I covers the basics of propositional and
first-order logic through the Soundness, Completeness, and Compact-
ness Theorems, plus some material on applications of the Compactness
Theorem. It could easily be used for a one-semester course on these
topics. Volume II covers the basics of computability using Turing ma-
chines and recursive functions, the Incompleteness Theorem, and basic
complexity theory; it could also be used as for a one-semester course
on these topics.
In keeping with the modified Moore-method, this book supplies
definitions, problems, and statements of results, along with some ex-
planations, examples, and hints. The intent is for the students, indi-
vidually or in groups, to learn the material by solving the problems
and proving the results for themselves. Besides constructive criticism,
it will probably be necessary for the instructor to supply further hints
or direct the students to other sources from time to time. Just how
this text is used will, of course, depend on the instructor and students
in question. However, it is probably not appropriate for a conventional
lecture-based course nor for a really large class.
The material presented in this volume is somewhat stripped-down.
Various concepts and topics that are often covered in introductory
mathematical logic courses are given very short shrift or omitted en-
tirely, among them normal forms, definability, and model theory.1 In-
structors might consider having students do projects on additional ma-
terial if they wish to to cover it. A diagram giving the dependence of
the chapters in this volume can be found in the Introduction.

Acknowledgements. Various people and institutions deserve the

credit for this text: All the people who developed the subject. My

1
Future versions of both volumes may include more – or less! – material. Feel
free to send suggestions, corrections, criticisms, and the like — I’ll feel free to ignore
them or use them.
v
vi PREFACE

teachers and colleagues, especially Gregory H. Moore, whose math-

ematical logic course convinced me that I wanted to do the stuff.
The students at Trent University who suffered, suffer, and will suf-
fer through assorted versions of this text. Trent University and the
taxpayers of Ontario, who paid my salary. Ohio University, where I
spent my sabbatical in 1995–96. All the people and organizations who
developed the software and hardware with which this book was pre-
pared. Anyone else I’ve missed.
Any blame properly accrues to the author.
Conditions. This book may be freely transmitted, stored, copied,
and used until 31 December, 2000, subject to the following restrictions:2
1. It may not be modified in any way without the express written
permission of the author.
2. It may not be sold at a profit, but only to recover the cost of
reproduction.
3. After 31 December, 2000, it may no longer be reproduced, stored,
or transmitted in any form without the express written permis-
sion of the author, except that printed copies existing as of 31
December, 2000, may be retained and used after this date.
The reason for the time-limit is that I hope to improve this book
and make a new version available.3
Availability. The URL of the home page for A Problem Course
In Mathematical Logic, with links to LATEX, PostScript, and Portable
Document Format (pdf) files of the latest available releases of both
volumes, is:
• http://www.trentu.ca/mathematics/sb/misc/pcml.html
A text-only information file and LATEX, PostScript, and Portable Doc-
ument Format (pdf) files of the latest release of both volumes can be
accessed by anonymous ftp in the directory:
• ftp://ftp.trentu.ca/pub/sbilaniuk/pcml/
Please note that in addition to LATEX you will need the AMS-LATEX
and AMSFonts packages to typeset and print either volume.
If you have any problems, feel free to contact the author at the
addresses given on the title page, preferably by e-mail, for assistance
or even to ask for a paper copy.
Author’s Opinion. It’s not great, but the price is right!
2
If you violate these restrictions, I will be flattered, but you will still be in the
wrong.
3
Who knows, maybe even find a publisher . . .
 Introduction

Mathematical Logic. What sets mathematics aside from other

disciplines is its reliance on proof as the principal technique for de-
termining truth, where science, for example, relies on (carefully ana-
lyzed) experience. So what is a proof? Practically speaking, it is any
reasoned argument accepted as a proof by other mathematicians.4 A
more precise definition is needed, however, if one wishes to discover
what mathematical reasoning can accomplish in principle. This is one
of the reasons for studying mathematical logic, which is also pursued
for its own sake and finding new tools to use in the rest of mathematics
and in related fields.
In any case, mathematical logic is concerned with formalizing and
analyzing the kinds of reasoning used in the rest of mathematics. The
point of mathematical logic is not to try to do mathematics per se
completely formally — the practical problems involved in doing so are
usually such as to make this an exercise in frustration — but to study
formal logical systems as mathematical objects in their own right in
order to (informally!) prove things about them. For this reason, the
formal systems developed in this book are optimized to be easy to prove
things about, rather than to be easy to use. Natural deductive systems
such as those developed by philosophers to formalize logical reasoning
are equally capable in principle and much easier to actually use, but
harder to prove things about.
Part of the problem with formalizing mathematical reasoning is the
necessity of precisely specifying the language(s) in which it is to be
done. The natural languages spoken by humans won’t do: they are
so complex and continually changing as to be impossible to pin down
completely. By contrast, the languages which underly formal logical
systems are, like programming languages, much simpler and less flexible
than natural languages but rigidly defined. A formal logical system also
requires the careful specification of the allowable rules of reasoning, plus
some notion of how to interpret statements in the underlying language

4
If you are not a mathematician, gentle reader, you are hereby temporarily
promoted.
1
2 INTRODUCTION

and determine their truth. The real fun lies in the relationship between
interpretation of statements, truth, and reasoning.
This volume develops the basics of two kinds of formal logical sys-
tems, propositional logic and first-order logic. Propositional logic at-
tempts to make precise the relationships that certain connectives like
not, and , or , and if . . . then are used to express in English. While it
has uses, propositional logic is not powerful enough to formalize most
mathematical discourse. For one thing, it cannot handle the concepts
expressed by all and there is. First-order logic adds all and there is to
those which propositional logic could handle, and suffices, in principle,
to formalize most mathematical reasoning. To be sure, it will not han-
dle concepts which arise outside of mathematics, such as possible and
relevant, among many others. (Trying to incorporate such concepts
into systems extending first-order logic is a substantial industry in phi-
losophy, but of marginal interest in mathematics.) Propositional logic,
which is much simpler, will be dealt with first in order to gain some
experience in dealing with formal systems before tackling first-order
logic. Besides, some of the results about propositional logic carry over
to first-order logic with little change.

Approach. This book supplies definitions and statements of re-

sults, plus some explanations and a number of problems and examples,
but no proofs of the results. The hope is that you, gentle reader, will
learn the material presented here by solving the problems and proving
the results for yourself. Brief hints are supplied for almost all of the
problems and results, but if these do not suffice, you should consult
your peers, your instructor, or other texts.

Prerequisites. In principle, not much is needed by way of prior

mathematical knowledge to define and prove the basic facts about
propositional and first-order logic. Some knowledge of the natural
numbers and a little set theory suffices; the former will be assumed
and the latter is very briefly summarized in Appendix A. What re-
ally is needed to get anywhere with the material developed here is
competence in handling abstraction and proofs, especially proofs by
induction. The experience provided by a rigorous introductory course
in algebra, analysis, or discrete mathematics ought to be sufficient.
Some problems and examples draw on concepts from other parts of
mathematics; students who are not already familiar with these should
consult texts in the appropriate subjects for the necessary definitions.

Other Sources and Further Reading. [4], [8], and [9] are texts
which go over similar ground (and much more), while [1] and [3] are
 INTRODUCTION 3

good references for more advanced material. Entertaining accounts of

some related topics may be found in [7] and [10]. Those interested
in natural deductive systems might try [2], which has a very clean
presentation.
Credit. Almost no attempt has been made to give due credit to
those who developed and refined the ideas, results, and proofs men-
tioned in this work. In mitigation, it would often be difficult to assign
credit fairly because many people were involved, frequently having in-
teracted in complicated ways. (Which really means that I’m too lazy
to do it. I apologize to those who have been hurt by this.) Those
interested in who did what should start by consulting other texts or
reference works covering similar material.
Chapter Dependencies. The following diagram indicates how
the chapters in this volume depend on one another, with the excep-
tion of a few isolated problems or results.

1
H
 HH


 H
j
H
2 3
H 
H 
H 
Hj
H
4

5
H
 H



HH
j?
H
6 7
H 
H 
H 
Hj
H
8

?
9
4 INTRODUCTION
Propositional Logic
 CHAPTER 1

Language

Propositional logic (sometimes called sentential or predicate logic)

attempts to formalize the reasoning that can be done with connectives
like not, and , or , and if . . . then. We will define the formal language
of propositional logic, LP , by specifying its symbols and rules for as-
sembling these symbols into the formulas of the language.
Definition 1.1. The symbols of LP are:
1. Parentheses: ( and ).
2. Connectives: ¬ and →.
3. Atomic formulas: A0, A1, A2, . . . , An , . . .
We still need to specify the ways in which the symbols of LP can
be put together.
Definition 1.2. The formulas of LP are those finite sequences or
strings of the symbols given in Definition 1.1 which satisfy the following
rules:
1. Every atomic formula is a formula.
2. If α is a formula, then (¬α) is a formula.
3. If α and β are formulas, then (α → β) is a formula.
4. No other sequence of symbols is a formula.
We will often use lower-case Greek characters to represent formulas,
as we did in the definition above, and upper-case Greek characters
to represent sets of formulas.1 All formulas in Chapters 1–4 will be
assumed to be formulas of LP unless stated otherwise.
What do these definitions mean? The parentheses are just punc-
tuation: their only purpose is to group other symbols together. (One
could get by without them; see Problem 1.6.) ¬ and → are supposed to
represent the connectives not and if . . . then respectively. The atomic
formulas, A0, A1, . . . , are meant to represent statements that cannot
be broken down any further using our connectives, such as “The moon
is made of cheese.” Thus, one might translate the the English sen-
tence “If the moon is red, it is not made of cheese” into the formula
1
The Greek alphabet is given in Appendix B.
7
8 1. LANGUAGE

(A0 → (¬A1)) of LP by using A0 to represent “The moon is red” and

A1 to represent “The moon is made of cheese.” Note that the truth
of the formula depends on the interpretation of the atomic sentences
which appear in it. Using the interpretations just given of A0 and A1 ,
the formula (A0 → (¬A1)) is true, but if we instead use A0 and A1
to interpret “My telephone is ringing” and “Someone is calling me”,
respectively, (A0 → (¬A1)) is false.
Definition 1.2 says that that every atomic formula is a formula and
every other formula is built from shorter formulas using the connectives
and parentheses in particular ways. For example, A1123, (A2 → (¬A0 )),
and (((¬A1) → (A1 → A7)) → A7) are all formulas, but X3 , (A5 ),
()¬A41, A5 → A7 , and (A2 → (¬A0 ) are not.
Problem 1.1. Why are the following not formulas of LP ? There
might be more than one reason . . .
1. A−56
2. (Y → A)
3. (A7 ← A4)
4. A7 → (¬A5))
5. (A8A9 → A1043998
6. (((¬A1 ) → (A` → A7) → A7 )
Problem 1.2. Show that every formula of LP has the same number
of left parentheses as it has of right parentheses.
Problem 1.3. Suppose α is any formula of LP . Let `(α) be the
length of α as a sequence of symbols and let p(α) be the number of
parentheses (counting both left and right parentheses) in α. What are
the minimum and maximum values of p(α)/`(α)?
Problem 1.4. Suppose α is any formula of LP . Let s(α) be the
number of atomic formulas in α (counting repetitions) and let c(α) be
the number of occurrences of → in α. Show that s(α) = c(α) + 1.
Problem 1.5. What are the possible lengths of formulas of LP ?
Prove it.
Problem 1.6. Find a way for doing without parentheses or other
punctuation symbols in defining a formal language for propositional
logic.
Proposition 1.7. Show that the set of formulas of LP is count-
able.
Informal Conventions. At first glance, LP may not seem capable
of breaking down English sentences with connectives other than not
 1. LANGUAGE 9

and if . . . then. However, the sense of many other connectives can be

captured by these two by using suitable circumlocutions. We will use
the symbols ∧, ∨, and ↔ to represent and , or ,2 and if and only if
respectively. Since they are not among the symbols of LP , we will use
them as abbreviations for certain constructions involving only ¬ and
→. Namely,
• (α ∧ β) is short for (¬(α → (¬β))),
• (α ∨ β) is short for ((¬α) → β), and
• (α ↔ β) is short for ((α → β) ∧ (β → α)).
Interpreting A0 and A1 as before, for example, one could translate the
English sentence “The moon is red and made of cheese” as (A0 ∧ A1 ).
(Of course this is really (¬(A0 → (¬A1 ))), i.e. “It is not the case that
if the moon is green, it is not made of cheese.”) ∧, ∨, and ↔ were not
included among the official symbols of LP partly because we can get
by without them and partly because leaving them out makes it easier
to prove things about LP .
Problem 1.8. Take a couple of English sentences with several con-
nectives and translate them into formulas of LP . You may use ∧, ∨,
and ↔ if appropriate.
Problem 1.9. Write out ((α ∨ β) ∧ (β → α)) using only ¬ and →.
For the sake of readability, we will occasionally use some informal
conventions that let us get away with writing fewer parentheses:
• We will usually drop the outermost parentheses in a formula,
writing α → β instead of (α → β) and ¬α instead of (¬α).
• We will let ¬ take precedence over → when parentheses are miss-
ing, so ¬α → β is short for ((¬α) → β), and fit the informal
connectives into this scheme by letting the order of precedence
be ¬, ∧, ∨, →, and ↔.
• Finally, we will group repetitions of →, ∨, ∧, or ↔ to the right
when parentheses are missing, so α → β → γ is short for (α →
(β → γ)).
Just like formulas using ∨, ∧, or ¬, formulas in which parentheses have
been omitted as above are not official formulas of LP , they are conve-
nient abbreviations for official formulas of LP . Note that a precedent
for the precedence convention can be found in the way that · commonly
takes precedence over + in writing arithmetic formulas.
Problem 1.10. Write out ¬(α ↔ ¬δ) ∧ β → ¬α → γ first with
the missing parentheses included and then as an official formula of LP .
2
We will use or inclusively, so that “A or B” is still true if both of A and B
are true.
10 1. LANGUAGE

The following notion will be needed later on.

Definition 1.3. Suppose ϕ is a formula of LP . The set of subfor-
mulas of ϕ, S(ϕ), is defined as follows.
1. If ϕ is an atomic formula, then S(ϕ) = {ϕ}.
2. If ϕ is (¬α), then S(ϕ) = S(α) ∪ {(¬α)}.
3. If ϕ is (α → β), then S(ϕ) = S(α) ∪ S(β) ∪ {(α → β)}.
For example, if ϕ is (((¬A1) → A7 ) → (A8 → A1)), then S(ϕ)
includes A1 , A7, A8, (¬A1), (A8 → A1), ((¬A1 ) → A7 ), and (((¬A1 ) →
A7) → (A8 → A1 )) itself.
Note that if you write out a formula with all the official parenthe-
ses, then the subformulas are just the parts of the formula enclosed by
matching parentheses, plus the atomic formulas. In particular, every
formula is a subformula of itself. Note that some subformulas of for-
mulas involving our informal abbreviations ∨, ∧, or ↔ will be most
conveniently written using these abbreviations. For example, if ψ is
A4 → A1 ∨ A4, then
S(ψ) = { A1 , A4, (¬A1 ), (A1 ∨ A4), (A4 → (A1 ∨ A4 )) } .
(As an exercise, where did (¬A1 ) come from?)
Problem 1.11. Find all the subformulas of each of the following
formulas.
1. (¬((¬A56) → A56))
2. A9 → A8 → ¬(A78 → ¬¬A0 )
3. ¬A0 ∧ ¬A1 ↔ ¬(A0 ∨ A1)
Unique Readability. The slightly paranoid — er, truly rigorous
— might ask whether Definitions 1.1 and 1.2 actually ensure that the
formulas of LP are unambiguous, i.e. can be read in only one way
according to the rules given in Definition 1.2. To actually prove this
one must add to Definition 1.1 the requirement that all the symbols
of LP are distinct and that no symbol is a subsequence of any other
symbol. With this addition, one can prove the following:
Theorem 1.12 (Unique Readability Theorem). A formula of LP
must satisfy exactly one of conditions 1–3 in Definition 1.2.
 CHAPTER 2

Truth Assignments

Whether a given formula ϕ of LP is true or false usually depends on

how we interpret the atomic formulas which appear in ϕ. For example,
if ϕ is the atomic formula A2 and we interpret it as “2+2 = 4”, it is true,
but if we interpret it as “The moon is made of cheese”, it is false. Since
we don’t want to commit ourselves to a single interpretation — after
all, we’re really interested in general logical relationships — we will
define how any assignment of truth values T (“true”) and F (“false”)
to atomic formulas of LP can be extended to all other formulas. We
will also get a reasonable definition of what it means for a formula of
LP to follow logically from other formulas.
Definition 2.1. A truth assignment is a function v whose domain
is the set of all formulas of LP and whose range is the set {T, F } of
truth values, such that:
1. v(An ) is defined for every atomic formula An .
2. For any formula α,
(
T if v(α) = F
v( (¬α) ) =
F if v(α) = T .
3. For any formulas α and β,
(
F if v(α) = T and v(β) = F
v( (α → β) ) =
T otherwise.

Given interpretations of all the atomic formulas of LP , the corre-

sponding truth assignment would give each atomic formula representing
a true statement the value T and every atomic formula representing a
false statement the value F . Note that we have not defined how to
handle any truth values besides T and F in LP . Logics with other
truth values have uses, but are not relevant in most of mathematics.
For an example of how non-atomic formulas are given truth values
on the basis of the truth values given to their components, suppose
v is a truth assignment such that v(A0) = T and v(A1) = F . Then
v( ((¬A1) → (A0 → A1)) ) is determined from v( (¬A1) ) and v( (A0 →
11
12 2. TRUTH ASSIGNMENTS

A1) ) according to clause 3 of Definition 2.1. In turn, v( (¬A1) ) is deter-

mined from of v(A1) according to clause 2 and v( (A0 → A1 ) ) is deter-
mined from v(A1) and v(A0) according to clause 3. Finally, by clause 1,
our truth assignment must be defined for all atomic formulas to begin
with; in this case, v(A0) = T and v(A1) = F . Thus v( (¬A1) ) = T and
v( (A0 → A1) ) = F , so v( ((¬A1) → (A0 → A1)) ) = F .
A convenient way to write out the determination of the truth value
of a formula on a given truth assignment is to use a truth table: list all
the subformulas of the given formula across the top in order of length
and then fill in their truth values on the bottom from left to right.
Except for the atomic formulas at the extreme left, the truth value of
each subformula will depend on the truth values of the subformulas to
its left. For the example above, one gets something like:
A0 A1 (¬A1) (A0 → A1 ) (¬A1 ) → (A0 → A1))
T F T F F
Problem 2.1. Suppose v is a truth assignment such that v(A0) =
v(A2) = T and v(A1) = v(A3) = F . Find v(α) if α is:
1. ¬A2 → ¬A3
2. ¬A2 → A3
3. ¬(¬A0 → A1 )
4. A0 ∨ A1
5. A0 ∧ A1
The use of finite truth tables to determine what truth value a par-
ticular truth assignment gives a particular formula is justified by the
following proposition, which asserts that only the truth values of the
atomic sentences in the formula matter.
Proposition 2.2. Suppose δ is any formula and u and v are truth
assignments such that u(An ) = v(An ) for all atomic formulas An which
occur in δ. Then u(δ) = v(δ).
Corollary 2.3. Suppose u and v are truth assignments such that
u(An ) = v(An) for every atomic formula An . Then u = v, i.e. u(ϕ) =
v(ϕ) for every formula ϕ.
Proposition 2.4. If α and β are formulas and v is a truth assign-
ment, then:
1. v(¬α) = T if and only if v(α) = F .
2. v(α → β) = T if and only if v(β) = T whenever v(α) = T ;
3. v(α ∧ β) = T if and only if v(α) = T and v(β) = T ;
4. v(α ∨ β) = T if and only if v(α) = T or v(β) = T ; and
5. v(α ↔ β) = T if and only if v(α) = v(β).
 2. TRUTH ASSIGNMENTS 13

Truth tables are often used even when the formula in question is
not broken down all the way into atomic formulas. For example, if α
and β are any formulas and we know that α is true but β is false, then
the truth of (α → (¬β)) can be determined by means of the following
table:
α β (¬β) (α → (¬β))
T F T T
Definition 2.2. If v is a truth assignment and ϕ is a formula, we
will often say that v satisfies ϕ if v(ϕ) = T . Similarly, if Σ is a set
of formulas, we will often say that v satisfies Σ if v(σ) = T for every
σ ∈ Σ. We will say that ϕ (respectively, Σ) is satisfiable if there is
some truth assignment which satisfies it.
Definition 2.3. A formula ϕ is a tautology if it is satisfied by
every truth assignment. A formula ψ is a contradiction if there is no
truth assignment which satisfies it.
For example, (A4 → A4) is a tautology while (¬(A4 → A4)) is a
contradiction, and A4 is a formula which is neither. One can check
whether a given formula is a tautology, contradiction, or neither, by
grinding out a complete truth table for it, with a separate line for each
possible assignment of truth values to the atomic subformulas of the
formula. For A3 → (A4 → A3 ) this gives
A3 A4 A4 → A3 A3 → (A4 → A3)
T T T T
T F T T
F T F T
F F T T
so A3 → (A4 → A3) is a tautology. Note that, by Proposition 2.2, we
need only consider the possible truth values of the atomic sentences
which actually occur in a given formula.
One can often use truth tables to determine whether a given formula
is a tautology or a contradiction even when it is not broken down all
the way into atomic formulas. For example, if α is any formula, then
the table
α (α → α) (¬(α → α))
T T F
F T F
demonstrates that (¬(α → α)) is a contradiction, no matter which
formula of LP α actually is.
14 2. TRUTH ASSIGNMENTS

Proposition 2.5. If α is any formula, then ((¬α) ∨ α) is a tau-

tology and ((¬α) ∧ α) is a contradiction.
Proposition 2.6. A formula β is a tautology if and only if ¬β is
a contradiction.
After all this warmup, we are finally in a position to define what it
means for one formula to follow logically from other formulas.
Definition 2.4. A set of formulas Σ implies a formula ϕ, written
as Σ |= ϕ, if every truth assignment v which satisfies Σ also satisfies ϕ.
We will often write Σ 2 ϕ if it is not the case that Σ |= ϕ. In the case
where Σ is empty, we will usually write |= ϕ instead of ∅ |= ϕ.
Similarly, if ∆ and Γ are sets of formulas, then ∆ implies Γ, written
as ∆ |= Γ, if every truth assignment v which satisfies ∆ also satisfies
Γ.
For example, { A3, (A3 → ¬A7) } |= ¬A7 , but { A8 , (A5 → A8) } 2
A5. (There is a truth assignment which makes A8 and A5 → A8 true,
but A5 false.) Note that a formula ϕ is a tautology if and only if |= ϕ,
and a contradiction if and only if |= (¬ϕ).
Proposition 2.7. If Γ and Σ are sets of formulas such that Γ ⊆ Σ,
then Σ |= Γ.
Problem 2.8. How can one check whether or not Σ |= ϕ for a
formula ϕ and a finite set of formulas Σ?
Proposition 2.9. Suppose Σ is a set of formulas and ψ and ρ are
formulas. Then Σ ∪ {ψ} |= ρ if and only if Σ |= ψ → ρ.
Proposition 2.10. A set of formulas Σ is satisfiable if and only
if there is no contradiction χ such that Σ |= χ.
 CHAPTER 3

Deductions

In this chapter we develop a way of defining logical implication

that does not rely on any notion of truth, but only on manipulating
sequences of formulas, namely formal proofs or deductions. (Of course,
any way of defining logical implication had better be compatible with
that given in Chapter 2.) To define these, we first specify a suitable
set of formulas which we can use freely as premisses in deductions.
Definition 3.1. The three axiom schema of LP are:
A1: (α → (β → α))
A2: ((α → (β → γ)) → ((α → β) → (α → γ)))
A3: (((¬β) → (¬α)) → (((¬β) → α) → β)).
Replacing α, β, and γ by particular formulas of LP in any one of the
schemas A1, A2, or A3 gives an axiom of LP .
For example, (A1 → (A4 → A1 )) is an axiom, being an instance of
axiom schema A1, but (A9 → (¬A0 )) is not an axiom as it is not the
instance of any of the schema. As had better be the case, every axiom
is always true:
Proposition 3.1. Every axiom of LP is a tautology.
Second, we specify our one (and only!) rule of inference.1
Definition 3.2 (Modus Ponens). Given the formulas ϕ and (ϕ →
ψ), one may infer ψ.
We will usually refer to Modus Ponens by its initials, MP. Like any
rule of inference worth its salt, MP preserves truth.
Proposition 3.2. Suppose ϕ and ψ are formulas. Then { ϕ, (ϕ →
ψ) } |= ψ.
With axioms and a rule of inference in hand, we can execute formal
proofs in LP .
1
Natural deductive systems, which are usually more convenient to actually
execute deductions in than the system being developed here, compensate for having
few or no axioms by having many rules of inference.
15
16 3. DEDUCTIONS

Definition 3.3. Let Σ be a set of formulas. A deduction or proof

from Σ in LP is a finite sequence ϕ1ϕ2 . . . ϕn of formulas such that for
each k ≤ n,
1. ϕk is an axiom, or
2. ϕk ∈ Σ, or
3. there are i, j < k such that ϕk follows from ϕi and ϕj by MP.
A formula of Σ appearing in the deduction is called a premiss. Σ proves
a formula α, written as Σ ` α, if α is the last formula of a deduction
from Σ. We’ll usually write ` α for ∅ ` α, and take Σ ` ∆ to mean
that Σ ` δ for every formula δ ∈ ∆.
In order to make it easier to verify that an alleged deduction really
is one, we will number the formulas in a deduction, write them out in
order on separate lines, and give a justification for each formula. Like
the additional connectives and conventions for dropping parentheses in
Chapter 1, this is not officially a part of the definition of a deduction.
Example 3.1. Let us show that ` ϕ → ϕ.
1. (ϕ → ((ϕ → ϕ) → ϕ)) → ((ϕ → (ϕ → ϕ)) → (ϕ → ϕ)) A2
2. ϕ → ((ϕ → ϕ) → ϕ) A1
3. (ϕ → (ϕ → ϕ)) → (ϕ → ϕ) 1,2 MP
4. ϕ → (ϕ → ϕ) A1
5. ϕ → ϕ 3,4 MP
Hence ` ϕ → ϕ, as desired. Note that indication of the formulas from
which formulas 3 and 5 beside the mentions of MP.
Example 3.2. Let us show that { α → β, β → γ } ` α → γ.
1. (β → γ) → (α → (β → γ)) A1
2. β → γ Premiss
3. α → (β → γ) 1,2 MP
4. (α → (β → γ)) → ((α → β) → (α → γ)) A2
5. (α → β) → (α → γ) 4,3 MP
6. α → β Premiss
7. α → γ 5,6 MP
Hence { α → β, β → γ } ` α → γ, as desired.
It is frequently convenient to save time and effort by simply referring
to a deduction one has already done instead of writing it again as part
of another deduction. If you do so, please make sure you appeal only
to deductions that have already been carried out.
Example 3.3. Let us show that ` (¬α → α) → α.
1. (¬α → ¬α) → ((¬α → α) → α) A3
 3. DEDUCTIONS 17

2. ¬α → ¬α Example 3.1
3. (¬α → α) → α 1,2 MP
Hence ` (¬α → α) → α, as desired. To be completely formal, one
would have to insert the deduction given in Example 3.1 (with ϕ re-
placed by ¬α throughout) in place of line 2 above and renumber the
old line 3.
Problem 3.3. Show that if α, β, and γ are formulas, then
1. { α → (β → γ), β } ` α → γ
2. ` α ∨ ¬α
Example 3.4. Let us show that ` ¬¬β → β.
1. (¬β → ¬¬β) → ((¬β → ¬β) → β) A3
2. ¬¬β → (¬β → ¬¬β) A1
3. ¬¬β → ((¬β → ¬β) → β) 1,2 Example 3.2
4. ¬β → ¬β Example 3.1
5. ¬¬β → β 3,4 Problem 3.3.1
Hence ` ¬¬β → β, as desired.
Certain general facts are sometimes handy:
Proposition 3.4. If ϕ1ϕ2 . . . ϕn is a deduction of LP , then ϕ1 . . . ϕ`
is also a deduction of LP for any ` such that 1 ≤ ` ≤ n.
Proposition 3.5. If Γ ` δ and Γ ` δ → β, then Γ ` β.
Proposition 3.6. If Γ ⊆ ∆ and Γ ` α, then ∆ ` α.
Proposition 3.7. If Γ ` ∆ and ∆ ` σ, then Γ ` σ.
The following theorem often lets one take substantial shortcuts
when trying to show that certain deductions exist in LP , even though
it doesn’t give us the deductions explicitly.
Theorem 3.8 (Deduction Theorem). If Σ is any set of formulas
and α and β are any formulas, then Σ ` α → β if and only if Σ∪{α} `
β.
Example 3.5. Let us show that ` ϕ → ϕ. By the Deduction
Theorem it is enough to show that {ϕ} ` ϕ, which is trivial:
1. ϕ Premiss
Compare this to the deduction in Example 3.1.
Problem 3.9. Appealing to previous deductions and the Deduction
Theorem if you wish, show that:
1. {δ, ¬δ} ` γ
18 3. DEDUCTIONS

2. ` ϕ → ¬¬ϕ
3. ` (¬β → ¬α) → (α → β)
4. ` (α → β) → (¬β → ¬α)
5. ` (β → ¬α) → (α → ¬β)
6. ` (¬β → α) → (¬α → β)
7. ` σ → (σ ∨ τ )
8. {α ∧ β} ` β
9. {α ∧ β} ` α
 CHAPTER 4

Soundness and Completeness

How are deduction and implication related, given that they were
defined in completely different ways? We have some evidence that they
behave alike; compare, for example, Proposition 2.9 and the Deduction
Theorem. It had better be the case that if there is a deduction of a
formula ϕ from a set of premisses Σ, then ϕ is implied by Σ. (Otherwise,
what’s the point of defining deductions?) It would also be nice for the
converse to hold: whenever ϕ is implied by Σ, there is a deduction of
ϕ from Σ. (So anything which is true can be proved.) The Soundness
and Completeness Theorems say that both ways do hold, so Σ ` ϕ if
and only if Σ |= ϕ, i.e. ` and |= are equivalent for propositional logic.
One direction is relatively straightforward to prove . . .
Theorem 4.1 (Soundness Theorem). If ∆ is a set of formulas and
α is a formula such that ∆ ` α, then ∆ |= α.
. . . but for the other direction we need some additional concepts.
Definition 4.1. A set of formulas Γ is inconsistent if Γ ` ¬(α →
α) for some formula α, and consistent if it is not inconsistent.
For example, {A41} is consistent by Proposition 4.2, but it follows
from Problem 3.9 that {A13, ¬A13} is inconsistent.
Proposition 4.2. If a set of formulas is satisfiable, then it is con-
sistent.
Proposition 4.3. Suppose ∆ is an inconsistent set of formulas.
Then ∆ ` ψ for any formula ψ.
Proposition 4.4. Suppose Σ is an inconsistent set of formulas.
Then there is a finite subset ∆ of Σ such that ∆ is inconsistent.
Corollary 4.5. A set of formulas Γ is consistent if and only if
every finite subset of Γ is consistent.
To obtain the Completeness Theorem requires one more definition.
Definition 4.2. A set of formulas Σ is maximally consistent if Σ
is consistent but Σ ∪ {ϕ} is inconsistent for any ϕ ∈
/ Σ.
19
20 4. SOUNDNESS AND COMPLETENESS

That is, a set of formulas is maximally consistent if it is consistent,

but there is no way to add any other formula to it and keep it consistent.
Problem 4.6. Suppose v is a truth assignment. Show that Σ =
{ ϕ | v(ϕ) = T } is maximally consistent.
We will need some facts concerning maximally consistent theories.
Proposition 4.7. If Σ is a maximally consistent set of formulas,
ϕ is a formula, and Σ ` ϕ, then ϕ ∈ Σ.
Proposition 4.8. Suppose Σ is a maximally consistent set of for-
mulas and ϕ is a formula. Then ¬ϕ ∈ Σ if and only if ϕ ∈
/ Σ.
Proposition 4.9. Suppose Σ is a maximally consistent set of for-
mulas and ϕ and ψ are formulas. Then ϕ → ψ ∈ Σ if and only if
ϕ∈
/ Σ or ψ ∈ Σ.
It is important to know that any consistent set of formulas can be
expanded to a maximally consistent set.
Theorem 4.10. Suppose Γ is a consistent set of formulas. Then
there is a maximally consistent set of formulas Σ such that Γ ⊆ Σ.
Now for the main event!
Theorem 4.11. A set of formulas is consistent if and only if it is
satisfiable.
Theorem 4.11 gives the equivalence between ` and |= in slightly
disguised form.
Theorem 4.12 (Completeness Theorem). If ∆ is a set of formulas
and α is a formula such that ∆ |= α, then ∆ ` α.
It follows that anything provable from a given set of premisses must
be true if the premisses are, and vice versa. The fact that ` and |= are
actually equivalent can be very convenient in situations where one is
easier to use than the other. For example, most parts of Problems 3.3
and 3.9 are much easier to do with truth tables instead of deductions,
even if one makes use of the Deduction Theorem.
Finally, one more consequence of Theorem 4.11.
Theorem 4.13 (Compactness Theorem). A set of formulas Γ is
satisfiable if and only if every finite subset of Γ is satisfiable.
We will not look at any uses of the Compactness Theorem now,
but we will consider a few applications of its counterpart for first-order
logic in Chapter 9.
First-Order Logic
 CHAPTER 5

Languages

As noted in the Introduction, propositional logic has obvious defi-

ciencies as a tool for mathematical reasoning. First-order logic remedies
enough of these to be adequate for formalizing most ordinary mathe-
matics. It does have enough in common with propositional logic to let
us recycle some of the material in Chapters 1–4.
A few informal words about how first-order languages work are in
order. In mathematics one often deals with structures consisting of
a set of elements plus various operations on them or relations among
them. To cite three common examples, a group is a set of elements
plus a binary operation on these elements satisfying certain conditions,
a field is a set of elements plus two binary operations on these elements
satisfying certain conditions, and a graph is a set of elements plus a
binary relation with certain properties. In most such cases, one fre-
quently uses symbols naming the operations or relations in question,
symbols for variables which range over the set of elements, symbols
for logical connectives such as not and for all, plus auxiliary symbols
such as parentheses, to write formulas which express some fact about
the structure in question. For example, if (G, ·) is a group, one might
express the associative law by writing something like
∀x ∀y ∀z x · (y · z) = (x · y) · z ,
it being understood that the variables range over the set G of group
elements. A formal language to do as much will require some or all of
these: symbols for various logical notions and for variables, some for
functions or relations, plus auxiliary symbols. It will also be necessary
to specify rules for putting the symbols together to make formulas, for
interpreting the meaning and determining the truth of these formulas,
and for making inferences in deductions.
For a concrete example, consider elementary number theory. The
set of elements under discussion is the set of natural numbers N =
{ 0, 1, 2, 3, 4, . . . }. One might need symbols or names for certain inter-
esting numbers, say 0 and 1; for variables over N such as n and x; for
functions on N, say · and +; and for relations, say =, <, and |. In
addition, one is likely to need symbols for punctuation, such as ( and
23
24 5. LANGUAGES

); for logical connectives, such as ¬ and →; and for quantifiers, such

as ∀ (“for all”) and ∃ (“there exists”). A statement of mathematical
English such as “For all n and m, if n divides m, then n is less than or
equal to m” can then be written as a cool formula like

∀n∀m (n | m → (n < m ∧ n = m)) .

The extra power of first-order logic comes at a price: greater com-

plexity. First, there are many first-order languages one might wish to
use, practically one for each subject, or even problem, in mathematics.1
We will set up our definitions and general results, however, to apply to
a wide range of them.2
As with LP , our formal language for propositional logic, first-order
languages are defined by specifying their symbols and how these may
be assembled into formulas.

Definition 5.1. The symbols of a first-order language L include:

1. Parentheses: ( and ).
2. Connectives: ¬ and →.
3. Quantifier: ∀.
4. Variables: v0, v1, v2, . . . , vn , . . .
5. Equality: =.
6. A (possibly empty) set of constant symbols.
7. For each k ≥ 1, a (possibly empty) set of k-place function sym-
bols.
8. For each k ≥ 1, a (possibly empty) set of k-place relation (or
predicate) symbols.
The symbols described in parts 1–5 are the logical symbols of L, shared
by every first-order language, and the rest are the non-logical symbols
of L, which usually depend on what the language’s intended use.

Note. It is possible to define first-order languages without =, so

= is considered a non-logical symbol by many authors. While such lan-
guages have some uses, they are uncommon in ordinary mathematics.
Observe that any first-order language L has countably many logical
symbols. It may have uncountably many symbols if it has uncountably
many non-logical symbols. Unless explicitly stated otherwise, we will

1
It is possible to formalize almost all of mathematics in a single first-order
language, like that of set theory or category theory. However, trying to actually do
most mathematics in such a language is so hard as to be pointless.
2
Specifically, to countable one-sorted first-order languages with equality.
 5. LANGUAGES 25

assume that every first-order language we encounter has only count-

ably many non-logical symbols. Most of the results we will prove actu-
ally hold for countable and uncountable first-order languages alike, but
some require heavier machinery to prove for uncountable languages.
Just as in LP , the parentheses are just punctuation while the con-
nectives, ¬ and →, are intended to express not and if . . . then. How-
ever, the rest of the symbols are new and are intended to express ideas
that cannot be handled by LP . The quantifier symbol, ∀, is meant to
represent for all, and is intended to be used with the variable symbols,
e.g. ∀v4. The constant symbols are meant to be names for particular
elements of the structure under discussion. k-place function symbols
are meant to name particular functions which map k-tuples of elements
of the structure to elements of the structure. k-place relation symbols
are intended to name particular k-place relations among elements of
the structure.3 Finally, = is a special binary relation symbol intended
to represent equality.
Example 5.1. Since the logical symbols are always the same, first-
order languages are usually defined by specifying the non-logical sym-
bols. A formal language for elementary number theory like that unof-
ficially described above, call it LN T , can be defined as follows.
• Constant symbols: 0 and 1
• Two 2-place function symbols: + and ·
• Two binary relation symbols: < and |
Each of these symbols is intended to represent the same thing it does
in informal mathematical usage: 0 and 1 are intended to be names
for the numbers zero and one, + and · names for the operations of
addition and multiplications, and < and | names for the relations “less
than” and “divides”. (Note that we could, in principle, interpret things
completely differently – let 0 represent the number forty-one, + the
operation of exponentiation, and so on – or even use the language to
talk about a different structure – say the real numbers, R, with 0,
1, +, ·, and < representing what they usually do and, just for fun,
| interpreted as “is not equal to”. More on this in Chapter 6.) We
will usually use the same symbols in our formal languages that we use
informally for various common mathematical objects. This convention
3
Intuitively, a relation or predicate expresses some (possibly arbitrary) relation-
ship among one or more objects. For example, “n is prime” is a 1-place relation
on the natural numbers, < is a 2-place or binary relation on the rationals, and
~a × (~b × ~c) = ~0 is a 3-place relation on R3 . Formally, a k-place relation on a set X
is just a subset of X k , i.e. the collection of sequences of length k of elements of X
for which the relation is true.
26 5. LANGUAGES

can occasionally cause confusion if it is not clear whether an expression

involving these symbols is supposed to be an expression in a formal
language or not.

Example 5.2. Here are some other first-order languages. Recall

that we need only specify the non-logical symbols in each case and
note that some parts of Definitions 5.2 and 5.3 may be irrelevant for
a given language if it is missing the appropriate sorts of non-logical
symbols.
1. The language of pure equality, L= :
• No non-logical symbols at all.
2. A language for fields, LF :
• Constant symbols: 0, 1
• 2-place function symbols: +, ·
3. A language for set theory, LS :
• 2-place relation symbol: ∈
4. A language for linear orders, LO :
• 2-place relation symbol: <
5. Another language for elementary number theory, LN :
• Constant symbol: 0
• 1-place function symbol: S
• 2-place function symbols: +, ·, E
Here 0 is intended to represent zero, S the successor function, i.e.
S(n) = n+1, and E the exponential function, i.e. E(n, m) = nm .
6. A “worst-case” countable language, L1 :
• Constant symbols: c1, c2 , c3, . . .
• For each k ≥ 1, k-place function symbols: f1k , f2k , f3k , . . .
• For each k ≥ 1, k-place relation symbols: P1k , P2k , P3k , . . .
This language has no use except as an abstract example.

It remains to specify how to form valid formulas from the symbols

of a first-order language L. This will be more complicated than it was
for LP . In fact, we first need to define a type of expression in L which
has no counterpart in propositional logic.

Definition 5.2. The terms of a first-order language L are those

finite sequences of symbols of L which satisfy the following rules:
1. Every variable symbol vn is a term.
2. Every constant symbol c is a term.
3. If f is a k-place function symbol and t1, . . . , tk are terms, then
ft1 . . . tk is also a term.
4. Nothing else is a term.
 5. LANGUAGES 27

That is, a term is an expression which represents some (possibly

indeterminate) element of the structure under discussion. For example,
in LN T or LN , +v0 v1 (informally, v0 + v1 ) is a term, though precisely
which natural number it represents depends on what values are assigned
to the variables v0 and v1.
Problem 5.1. Which of the following are terms of one of the lan-
guages defined in Examples 5.1 and 5.2? If so, which of these lan-
guage(s) are they terms of; if not, why not?
1. ·v2
2. +0 · +v611
3. |1 + v30
4. (< E101 → +11)
5. + + · + 00000
6. f43 f72 c4 v9c1 v4
7. ·v5(+1v8 )
8. < v6v2
9. 1 + 0
Note that in languages with no function symbols all terms have
length one.
Problem 5.2. Choose one of the languages defined in Examples
5.1 and 5.2 which has terms of length greater than one and determine
the possible lengths of terms of this language.
Proposition 5.3. The set of terms of a countable first-order lan-
guage L is countable.
Having defined terms, we can finally define first-order formulas.
Definition 5.3. The formulas of a first-order language L are the
finite sequences of the symbols of L satisfying the following rules:
1. If P is a k-place relation symbol and t1 , . . . , tk are terms, then
P t1 . . . tk is a formula.
2. If t1 and t2 are terms, then = t1 t2 is a formula.
3. If α is a formula, then (¬α) is a formula.
4. If α and β are formulas, then (α → β) is a formula.
5. If ϕ is a formula and vn is a variable, then ∀vn ϕ is a formula.
6. Nothing else is a formula.
Formulas of form 1 or 2 will often be referred to as the atomic formulas
of L.
Note that three of the conditions in Definition 5.3 are borrowed
directy from propositional logic. As before, we will exploit the way
28 5. LANGUAGES

formulas are built up in making definitions and in proving results by

induction on the length of a formula. We will also recycle the use
of lower-case Greek characters to refer to formulas and of upper-case
Greek characters to refer to sets of formulas.
Problem 5.4. Which of the following are formulas of one of the
languages defined in Examples 5.1 and 5.2? If so, which of these lan-
guage(s) are they formulas of; if not, why not?
1. = 0 + v7 · 1v3
2. (¬ = v1v1 )
3. (|v20 → ·01)
4. (¬∀v5(= v5v5 ))
5. < +01|v1 v3
6. (v3 = v3 → ∀v5 v3 = v5)
7. ∀v6(= v60 → ∀v9(¬|v9v6 ))
8. ∀v8 < +11v4
Problem 5.5. Show that every formula of a first-order language
has the same number of left parentheses as of right parentheses.
Problem 5.6. Choose one of the languages defined in Examples
5.1 and 5.2 and determine the possible lengths of formulas of this lan-
guage.
Proposition 5.7. A countable first-order language L has count-
ably many formulas.
In practice, devising a formal language intended to deal with a par-
ticular (kind of) structure isn’t the end of the job: one must also specify
axioms in the language that the structure(s) one wishes to study should
satisfy. Defining satisfaction is officially done in the next chapter, but
it is usually straightforward to unofficially figure out what a formula
in the language is supposed to mean.
Problem 5.8. In each case, write down a formula of the given
language expressing the given informal statement.
1. “Addition is associative” in LF .
2. “There is an empty set” in LS .
3. “Between any two distinct elements there is a third element” in
LO .
4. “n0 = 1 for every n different from 0” in LN .
5. “There is only one thing” in L= .
Problem 5.9. Define first-order languages to deal with the follow-
ing structures and, in each case, an appropriate set of axioms in your
language:
 5. LANGUAGES 29

1. Groups.
2. Graphs.
3. Vector spaces.
We will need a few additional concepts and facts about formulas of
first-order logic later on. First, what are the subformulas of a formula?
Problem 5.10. Define the set of subformulas of a formula ϕ of a
first-order language L.
For example, if ϕ is
(((¬∀v1 (¬ = v1 c7)) → P32 v5v8) → ∀v8(= v8 f53 c0 v1v5 → P21 v8))
in the language L1 , then the set of subformulas of ϕ, S(ϕ), ought to
include
• = v1c7 , P32 v5v8, = v8f53 c0 v1v5 , P21 v8 ,
• (¬ = v1c7 ), (= v8f53 c0 v1v5 → P21 v8),
• ∀v1 (¬ = v1 c7), ∀v8(= v8f53 c0 v1v5 → P21 v8),
• (¬∀v1 (¬ = v1 c7)),
• (¬∀v1 (¬ = v1 c7)) → P32 v5v8), and
• (((¬∀v1 (¬ = v1c7 )) → P32 v5 v8) → ∀v8(= v8f53 c0 v1v5 → P21 v8))
itself.
Second, we will need a concept that has no counterpart in proposi-
tional logic.
Definition 5.4. Suppose x is a variable of a first-order language
L. Then x occurs free in a formula ϕ of L is defined as follows:
1. If ϕ is atomic, then x occurs free in ϕ if and only if x occurs in
ϕ.
2. If ϕ is (¬α), then x occurs free in ϕ if and only if x occurs free
in α.
3. If ϕ is (β → δ), then x occurs free in ϕ if and only if x occurs
free in β or in δ.
4. If ϕ is ∀vk ψ, then x occurs free in ϕ if and only if x is different
from vk and x occurs free in ψ.
An occurrence of x in ϕ which is not free is said to be bound . A formula
σ of L in which no variable occurs free is said to be a sentence.
Part 4 is the key: it asserts that an occurrence of a variable x
is bound instead of free if it is in the “scope” of an occurrence of
∀x. For example, v7 is free in ∀v5 = v5v7, but v5 is not. Different
occurences of a given variable in a formula may be free or bound,
depending on where they are; e.g. v6 occurs both free and bound in
∀v0 (= v0f31 v6 → (¬∀v6 P91 v6 )).
30 5. LANGUAGES

Problem 5.11. Give a precise definition of the scope of a quanti-

fier.
Note the distinction between sentences and ordinary formulas intro-
duced in the last part of Definition 5.4. As we shall see, sentences are
often more tractable and useful theoretically than ordinary formulas.
Problem 5.12. Which of the formulas you gave in solving Prob-
lem 5.8 are sentences?
Finally, we will eventually need to consider a relationship between
first-order languages.
Definition 5.5. A first-order language L0 is an extension of a first-
order language L, sometimes written as L ⊆ L0 , if every non-logical
symbol of L is a non-logical symbol of the same kind of L0 .
For example, every first-order language is an extension of L= .
Problem 5.13. Which of the languages given in Example 5.2 are
extensions of other languages given in Example 5.2?
Proposition 5.14. Suppose L is a first-order language and L0 is
an extension of L. Then every formula ϕ of L is a formula of L0 .
Common Conventions. As with propositional logic, we will often
use abbreviations and informal conventions to simplify the writing of
formulas in first-order languages. In particular, we will use the same
additional connectives we used in propositional logic, plus an additional
quantifier, ∃ (“there exists”):
• (α ∧ β) is short for (¬(α → (¬β))).
• (α ∨ β) is short for ((¬α) → β).
• (α ↔ β) is short for ((α → β) ∧ (β → α)).
• ∃vk ϕ is short for (¬∀vk (¬ϕ)).
(∀ is often called the universal quantifier and ∃ is often called the
existential quantifier.)
Parentheses will often be omitted in formulas according to the same
conventions we used in propositional logic, with the modification that
∀ and ∃ take precedence over all the logical connectives:
• We will usually drop the outermost parentheses in a formula,
writing α → β instead of (α → β) and ¬α instead of (¬α).
• We will let ∀ take precedence over ¬, and ¬ take precedence over
→ when parentheses are missing, and fit the informal abbrevia-
tions into this scheme by letting the order of precedence be ∀, ∃,
¬, ∧, ∨, →, and ↔.
 5. LANGUAGES 31

• Finally, we will group repetitions of →, ∨, ∧, or ↔ to the right

when parentheses are missing, so α → β → γ is short for (α →
(β → γ)).
For example, ∃vk ¬α → ∀vn β is short for ((¬∀vk (¬(¬α))) → ∀vnβ).
On the other hand, we will sometimes add parentheses and arrange
things in unofficial ways to make terms and formulas easier to read. In
particular we will often write
1. f(t1 , . . . , tk ) for ft1 . . . tk if f is a k-place function symbol and
t1 , . . . , tk are terms,
2. s ◦ t for ◦st if ◦ is a 2-place function symbol and s and t are
terms,
3. P (t1 , . . . , tk ) for P t1 . . . tk if P is a k-place relation symbol and
t1 , . . . , tk are terms,
4. s•t for •st if • is a 2-place relation symbol and s and t are terms,
and
5. s = t for = st if s and t are terms, and
6. enclose terms in parentheses to group them.
Thus, we could write the formula = +1 · 0v6 · 11 of LN T as 1 + (0 · v6) =
1 · 1.
As was observed in Example 5.1, it is customary in devising a formal
language to recycle the same symbols used informally for the given
objects. In situations where we want to talk about symbols without
committing ourselves to a particular one, such as when talking about
first-order languages in general, we will often use “generic” choices:
• a, b, c, . . . for constant symbols;
• x, y, z, . . . for variable symbols;
• f, g, h, . . . for function symbols;
• P , Q, R, . . . for relation symbols; and
• r, s, t, . . . for generic terms.
These can be thought of as variables in the metalanguage4 ranging over
different kinds objects of first-order logic, much as we’re already using
lower-case Greek characters as variables which range over formulas. (In
fact, we have already used some of these conventions in this chapter
... )

Unique Readability. The slightly paranoid might ask whether

Definitions 5.1, 5.2 and 5.3 actually ensure that the terms and formulas
4
The metalanguage is the language, mathematical English in this case, in which
we talk about a language. The theorems we prove about formal logic are, strictly
speaking, metatheorems, as opposed to the theorems proved within a formal logical
system. For more of this kind of stuff, read some philosophy . . .
32 5. LANGUAGES

of a first-order language L are unambiguous, i.e. cannot be read in

more than one way. As with LP , to actually prove this one must
assume that all the symbols of L are distinct and that no symbol is a
subsequence of any other symbol. It then follows that:
Theorem 5.15. Any term of a first-order language L satisfies ex-
actly one of conditions 1–3 in Definition 5.2.
Theorem 5.16 (Unique Readability Theorem). Any formula of a
first-order language satisfies exactly one of conditions 1–5 in Definition
5.3.
 CHAPTER 6

Structures and Models

Defining truth and implication in first-order logic is a lot harder

than it was in propositional logic. First-order languages are intended
to deal with mathematical objects like groups or linear orders, so it
makes little sense to speak of the truth of a formula without specifying
a context. For example, one can write down a formula expressing the
commutative law in a language for group theory, ∀x ∀y x · y = y · x,
but whether it is true or not depends on which group we’re dealing
with. It follows that we need to make precise which mathematical
objects or structures a given first-order language can be used to discuss
and how, given a suitable structure, formulas in the language are to
be interpreted. Such a structure for a given language should supply
most of the ingredients needed to interpret formulas of the language.
Throughout this chapter, let L be an arbitrary fixed countable first-
order language. All formulas will be assumed to be formulas of L unless
stated otherwise.
Definition 6.1. A structure M for L consists of the following:
1. A non-empty set M, often written as |M|, called the universe of
M.
2. For each constant symbol c of L, an element cM of M.
3. For each k-place function symbol f of L, a function f M : M k →
M, i.e. a k-place function on M.
4. For each k-place relation symbol P of L, a relation P M ⊆ M k ,
i.e. a k-place relation on M.
That is, a structure supplies an underlying set of elements plus in-
terpretations for the various non-logical symbols of the language: con-
stant symbols are interpreted by particular elements of the underlying
set, function symbols by functions on this set, and relation symbols by
relations among elements of this set.
It is customary to use upper-case “gothic” characters such as M
and N for structures.
For example, consider Q = (Q, <), where < is the usual “less than”
relation on the rationals. This is a structure for LO , the language for
linear orders defined in Example 5.2; it supplies a 2-place relation to
33
34 6. STRUCTURES AND MODELS

interpret the language’s 2-place relation symbol. Q is not the only

possible structure for LO : (R, <), ({0}, ∅), and (N, N2) are three others
among infinitely many more. (Note that in these cases the relation
symbol < is interpreted by relations on the universe which are not
linear orders. One can ensure that a structure satisfy various condi-
tions beyond what Definition 6.1 guarantees by requiring appropriate
formulas to be true when interpreted in the structure.) On the other
hand, (R) is not a structure for LO because it lacks a binary relation
to interpret the symbol < by, while (N, 0, 1, +, ·, |, <) is not a structure
for LO because it has two binary relations where LO has a symbol only
for one, plus constants and functions for which LO lacks symbols.
Problem 6.1. The first-order languages referred to below were all
defined in Example 5.2.
1. Is (∅) a structure for L= ?
2. Determine whether Q = (Q, <) is a structure for each of L= ,
LF , and LS .
3. Give three different structures for LF which are not fields.
To determine what it means for a given formula to be true in a
structure for the corresponding language, we will also need to specify
how to interpret the variables when they occur free. (Bound variables
have the associated quantifier to tell us what to do.)
Definition 6.2. Let V = { v0 , v1, v2, . . . } be the set of all variable
symbols of L and suppose M is a structure for L. A function s : V →
|M| is said to be an assignment for M.
Note that these are not truth assignments like those for LP . An
assignment just interprets each variable in the language by an element
of the universe of the structure. Also, as long as the universe of the
structure has more than one element, any variable can be interpreted
in more than one way. Hence there are usually many different possible
assignments for a given structure.
Example 6.1. Consider the structure R = (R, 0, 1, +, ·) for LF .
Each of the following functions V → R is an assignment for R:
1. p(vn ) = π for each n,
2. r(vn ) = en for each n, and
3. s(vn ) = n + 1 for each n.
In fact, every function V → R is an assignment for R.
In order to use assignments to determine whether formulas are true
in a structure, we need to know how to use an assignment to interpret
each term of the language as an element of the universe.
 6. STRUCTURES AND MODELS 35

Definition 6.3. Suppose M is a structure for L and s : V → |M|

is an assignment for M. Let T be the set of all terms of L. Then the
extended assignment s : T → |M| is defined inductively as follows:
1. For each variable x, s(x) = s(x).
2. For each constant symbol c, s(c) = cM .
3. For every k-place function symbol f and terms t1 , . . . , tk ,
s(ft1 . . . tk ) = f M (s(t1), . . . , s(tk )).
Example 6.2. Let R be the structure for LF given in Example
6.1, and let p, r, and s be the extended assignments corresponding to
the assignments p, r, and s defined in Example 6.1. Consider the term
+ · v6v0 + 0v3 of LF . Then:
1. p(+ · v6 v0 + 0v3 ) = π 2 + π,
2. r(+ · v6v0 + 0v3) = e6 + e3 , and
3. s(+ · v6v0 + 0v3 ) = 11.
Here’s why for the last one: since s(v6) = 7, s(v0) = 1, s(v3) = 4,
and s(0) = 0 (by part 2 of Definition 6.3), it follows from part 3 of
Definition 6.3 that s(+ · v6v0 + 0v3) = (7 · 1) + (0 + 4) = 7 + 4 = 11.
Problem 6.2. N = (N, 0, S, +, ·, E) is a structure for LN . Let
s : V → N be the assignment defined by s(vk ) = k + 1. What are
s(E + v19v1 · 0v45 ) and s(SSS + E0v6 v7)?
Proposition 6.3. s is unique, i.e. given an assignment s, no other
function T → |M| satisfies conditions 1–3 in Definition 6.3.
With Definitions 6.2 and 6.3 in hand, we can take our first cut at
defining what it means for a first-order formula to be true.
Definition 6.4. Suppose M is a structure for L, s is an assignment
for M, and ϕ is a formula of L. Then M |= ϕ[s] is defined as follows:
1. If ϕ is t1 = t2 for some terms t1 and t2 , then M |= ϕ[s] if and
only if s(t1) = s(t2).
2. If ϕ is P t1 . . . tk for some k-place relation symbol P and terms
t1 , . . . , tk , then M |= ϕ[s] if and only if (s(t1 ), . . . , s(tk )) ∈ P M ,
i.e. P M is true of (s(t1 ), . . . , s(tk )).
3. If ϕ is (¬ψ) for some formula ψ, then M |= ϕ[s] if and only if it
is not the case that M |= ψ[s].
4. If ϕ is (α → β), then M |= ϕ[s] if and only if M |= β[s] whenever
M |= α[s], i.e. unless M |= α[s] but not M |= β[s].
5. If ϕ is ∀x δ for some variable x, then M |= ϕ[s] if and only if for
all m ∈ |M|, M |= δ[s(x|m)], where s(x|m) is the assignment
36 6. STRUCTURES AND MODELS

given by
(
s(vk ) if vk is different from x
s(x|m)(vk ) =
m if vk is x.
If M |= ϕ[s], we shall say that M satisfies ϕ on assignment s or that
ϕ is true in M on assignment s. We will often write M 2 ϕ[s] if it is
not the case that M |= ϕ[s]. Also, if Γ is a set of formulas of L, we
shall take M |= Γ[s] to mean that M |= γ[s] for every formula γ in Γ
and say that M satisfies Γ on assignment s. Similarly, we shall take
M 2 Γ[s] to mean that M 2 γ[s] for some formula γ in Γ.
Clauses 1 and 2 are pretty straightforward and clauses 3 and 4 are
essentially identical to the corresponding parts of Definition 2.1. The
key clause is 5, which says that ∀ should be interpreted as “for all
elements of the universe”.
Example 6.3. Let R be the structure for LF and s the assignment
for R given in Example 6.1, and consider the formula ∀v1 (= v3 ·0v1 →=
v30) of LF . We can verify that R |= ∀v1 (= v3 · 0v1 →= v30) [s] as
follows:
R |= ∀v1 (= v3 · 0v1 →= v3 0) [s]
⇐⇒ for all a ∈ |R|, R |= (= v3 · 0v1 →= v3 0) [s(v1|a)]
⇐⇒ for all a ∈ |R|, if R |== v3 · 0v1 [s(v1|a)],
then R |== v3 0 [s(v1|a)]
⇐⇒ for all a ∈ |R|, if s(v1|a)(v3) = s(v1 |a)(·0v1),
then s(v1|a)(v3) = s(v1|a)(0)
⇐⇒ for all a ∈ |R|, if s(v3) = s(v1|a)(0) · s(v1|a)(v1), then s(v3) = 0
⇐⇒ for all a ∈ |R|, if s(v3) = 0 · a, then s(v3 ) = 0
⇐⇒ for all a ∈ |R|, if 4 = 0 · a, then 4 = 0
⇐⇒ for all a ∈ |R|, if 4 = 0, then 4 = 0
. . . which last is true whether or not 4 = 0 is true or false.
Problem 6.4. Let N be the structure for LN in Problem 6.2. Let
p : V → N be defined by p(v2k ) = k and p(v2k+1 ) = k. Verify that
1. N |= ∀w (¬Sw = 0) [p] and
2. N 2 ∀x∃y x + y = 0 [p].
Proposition 6.5. Suppose M is a structure for L, s is an assign-
ment for M, x is a variable, and ϕ is a formula of a first-order language
L. Then M |= ∃x ϕ[s] if and only if M |= ϕ[s(x|m)] for some m ∈ |M|.
Working with particular assignments is difficult but, while some-
times unavoidable, not always necessary.
 6. STRUCTURES AND MODELS 37

Definition 6.5. Suppose M is a structure for L, and ϕ a formula

of L. Then M |= ϕ if and only if M |= ϕ[s] for every assignment
s : V → |M| for M. M is a model of ϕ or that ϕ is true in M if
M |= ϕ. We will often write M 2 ψ if it is not the case that M |= ψ.
Similarly, if Γ is a set of formulas, we will write M |= Γ if M |= γ
for every formula γ ∈ Γ, and say that M is a model of Γ or that M
satisfies Γ. A formula or set of formulas is satisfiable if there is some
structure M which satisfies it. We will often write M 2 Γ if it is not
the case that M |= Γ.
Note. M 2 ϕ does not mean that for every assignment s : V →
|M|, it is not the case that M |= ϕ[s]. It only means that that there is
some assignment r : V → |M| for which M |= ϕ[r] is not true.
Problem 6.6. Q = (Q, <) is a structure for LO . For each of the
following formulas ϕ of LO , determine whether or not Q |= ϕ.
1. ∀v0 ∃v2 v0 < v2
2. ∃v1 ∀v3 (v1 < v3 → v1 = v3 )
3. ∀v4 ∀v5 ∀v6(v4 < v5 → (v5 < v6 → v4 < v6))
The following facts are counterparts of sorts for Proposition 2.2.
Their point is that what a given assignment does with a given term or
formula depends only on the assignment’s values on the (free) variables
of the term or formula.
Lemma 6.7. Suppose M is a structure for L, t is a term of L, and
r and s are assignments for M such that r(x) = s(x) for every variable
x which occurs in t. Then r(t) = s(t).
Proposition 6.8. Suppose M is a structure for L, ϕ is a formula
of L, and r and s are assignments for M such that r(x) = s(x) for
every variable x which occurs free in ϕ. Then M |= ϕ[r] if and only if
M |= ϕ[s].
Corollary 6.9. Suppose M is a structure for L and σ is a sen-
tence of L. Then M |= σ if and only if there is some assignment
s : V → |M| for M such that M |= σ[s].
Thus sentences are true or false in a structure independently of any
particular assignment. This does not necessarily make life easier when
trying to verify whether a sentence is true in a structure – try doing
Problem 6.6 again with the above results in hand – but it does let us
simplify things on occasion when proving things about sentences rather
than formulas.
We recycle a sense in which we used |= in propositional logic.
38 6. STRUCTURES AND MODELS

Definition 6.6. Suppose Γ is a set of formulas of L and ψ is a

formula of L. Then Γ implies ψ, written as Γ |= ψ, if M |= ψ whenever
M |= Γ for every structure M for L.
Similarly, if Γ and ∆ are sets of formulas of L, then Γ implies ∆,
written as Γ |= ∆, if M |= ∆ whenever M |= Γ for every structure M
for L.
We will usually write |= . . . for ∅ |= . . . .
Proposition 6.10. Suppose α and β are formulas of some first-
order language. Then { (α → β), α } |= β.
Proposition 6.11. Suppose Σ is a set of formulas and ψ and ρ
are formulas of some first-order language. Then Σ ∪ {ψ} |= ρ if and
only if Σ |= (ψ → ρ).
Definition 6.7. A formula ψ of L is a tautology if it is true in
every structure, i.e. if |= ψ. ψ is a contradiction if ¬ψ is a tautology,
i.e. if |= ¬ψ.
For some trivial examples, let ϕ be a formula of L and M a structure
for L. Then M |= {ϕ} if and only if M |= ϕ, so it must be the case
that {ϕ} |= ϕ. It is also easy to check that ϕ → ϕ is a tautology and
¬(ϕ → ϕ) is a contradiction.
Problem 6.12. Show that ∀y y = y is a tautology and that ∃y ¬y =
y is a contradiction.
Problem 6.13. Suppose ϕ is a contradiction. Show that M |= ϕ[s]
is false for every structure M and assignment s : V → |M| for M.
Problem 6.14. Show that a set of formulas Σ is satisfiable if and
only if there is no contradiction χ such that Σ |= χ.
The following fact is a counterpart of Proposition 2.4.
Proposition 6.15. Suppose M is a structure for L and α and β
are sentences of L. Then:
1. M |= ¬α if and only if M 2 α.
2. M |= α → β if and only if M |= β whenever M |= α.
3. M |= α ∨ β if and only if M |= α or M |= β.
4. M |= α ∧ β if and only if M |= α and M |= β.
5. M |= α ↔ β if and only if M |= α exactly when M |= β.
6. M |= ∀x α if and only if M |= α.
7. M |= ∃x α if and only if there is some m ∈ |M| so that M |=
α [s(x|m)] for every assignment s for M.
Problem 6.16. How much of Proposition 6.15 must remain true
if α and β are not sentences?
 6. STRUCTURES AND MODELS 39

Recall that by Proposition 5.14 a formula of a first-order language

is also a formula of any extension of the language. The following rela-
tionship between extension languages and satisfiability will be needed
later on.
Proposition 6.17. Suppose L is a first-order language, L0 is an
extension of L, and Γ is a set of formulas of L. Then Γ is satisfiable
in a structure for L if and only if Γ is satisfiable in a structure for L0 .
One last bit of terminology . . .
Definition 6.8. If M is a structure for L, then the theory of M
is just the set of all sentences of L true in M, i.e.
Th(M) = { τ | τ is a sentence and M |= τ }.
If ∆ is a set of sentences and S is a collection of structures, then ∆ is
a set of (non-logical) axioms for S if for every structure M, M ∈ S if
and only if M |= ∆.
Example 6.4. Consider the sentence ∃x ∃y ((¬x = y) ∧ ∀z (z =
x ∨ z = y)) of L= . Every structure of L= satisfying this sentence must
have exactly two elements in its universe, so { ∃x ∃y ((¬x = y)∧∀z (z =
x ∨ z = y)) } is a set of non-logical axioms for the collection of sets of
cardinality 2:
{ M | M is a structure for L= with exactly 2 elements } .
Problem 6.18. In each case, find a suitable language and a set of
axioms in it for the given collection of structures.
1. Sets of size 3.
2. Bipartite graphs.
3. Commutative groups.
4. Fields of characteristic 5.
40 6. STRUCTURES AND MODELS
 CHAPTER 7

Deductions

Deductions in first-order logic are not unlike deductions in propo-

sitional logic. Of course, some changes are necessary to handle the
various additional features of propositional logic, especially quantifiers.
In particular, one of the new axioms requires a tricky preliminary def-
inition. Roughly, the problem is that we need to know when we can
replace occurrences of a variable in a formula by a term without letting
any variable in the term get captured by a quantifier.
Throughout this chapter, let L be a fixed arbitrary first-order lan-
guage. Unless stated otherwise, all formulas will be assumed to be
formulas of L.

Definition 7.1. Suppose x is a variable, t is a term, and ϕ is a

formula. Then t is substitutable for x in ϕ is defined as follows:

1. If ϕ is atomic, then t is substitutable for x in ϕ.

2. If ϕ is (¬ψ), then t is substitutable for x in ϕ if and only if t is
substitutable for x in ψ.
3. If ϕ is (α → β), then t is substitutable for x in ϕ if and only if t
is substitutable for x in α and t is substitutable for x in β.
4. If ϕ is ∀y δ, then t is substitutable for x in ϕ if and only if either
(a) x does not occur free in ϕ, or
(b) if y does not occur in t and t is substitutable for x in δ.

For example, x is always substitutable for itself in any formula

ϕ and ϕxx is just ϕ (see Problem 7.1). On the other hand, y is not
substitutable for x in ∀y x = y because if x were to be replaced by y,
the new instance of y would be “captured” by the quantifier ∀y. This
makes a difference to the truth of the formula. The truth of ∀y x = y
depends on the structure in which it is interpreted — it’s true if the
universe has only one element and false otherwise — but ∀y y = y is
a tautology by Problem 6.12 so it is true in any structure whatsoever.
This sort of difficulty makes it necessary to be careful when substituting
for variables.

41
42 7. DEDUCTIONS

Definition 7.2. Suppose x is a variable, t is a term, and ϕ is

a formula. If t is substitutable for x in ϕ, then ϕxt (i.e. ϕ with t
substituted for x) is defined as follows:
1. If ϕ is atomic, then ϕxt is the formula obtained by replacing each
occurrence of x in ϕ by t.
2. If ϕ is (¬ψ), then ϕxt is the formula (¬ψtx).
3. If ϕ is (α → β), then ϕxt is the formula (αxt → βtx).
4. If ϕ is ∀y δ, then ϕxt is the formula
(a) ∀y δ if x is y, and
(b) ∀y δtx if x isn’t y.
Problem 7.1. 1. Is x substitutable for z in ψ if ψ is z = x →
∀z z = x? If so, what is ψxz ?
2. Show that if t is any term and σ is a sentence, then t is substi-
tutable in σ for any variable x. What is σtx ?
3. Show that if t is a term in which no variable occurs that occurs
in the formula ϕ, then t is substitutable in ϕ for any variable x.
4. Show that x is substitutable for x in ϕ for any variable x and
any formula ϕ, and that ϕxx is just ϕ.
Along with the notion of substitutability, we need an additional
notion in order to define the logical axioms of L.
Definition 7.3. If ϕ is any formula and x1, . . . , xn are any vari-
ables, then ∀x1 . . . ∀xn ϕ is said to be a generalization of ϕ.
For example, ∀y ∀x (x = y → fx = fy) and ∀z (x = y → fx = fy)
are (different) generalizations of x = y → fx = fy, but ∀x ∃y (x =
y → fx = fy) is not. Note that the variables being quantified don’t
have to occur in the formula being generalized.
Lemma 7.2. Any generalization of a tautology is a tautology.
Definition 7.4. Every first-order language L has eight logical ax-
iom schema:
A1: (α → (β → α))
A2: ((α → (β → γ)) → ((α → β) → (α → γ)))
A3: (((¬β) → (¬α)) → (((¬β) → α) → β))
A4: (∀x α → αxt ), if t is substitutable for x in α.
A5: (∀x (α → β) → (∀x α → ∀x β))
A6: (α → ∀x α), if x does not occur free in α.
A7: x = x
A8: (x = y → (α → β)), if α is atomic and β is obtained from α
by replacing some occurrences (possibly all or none) of x in α by
y.
 7. DEDUCTIONS 43

Plugging in any particular formulas of L for α, β, and γ, and any

particular variables for x and y, in any of A1–A8 gives a logical axiom
of L. In addition, any generalization of a logical axiom of L is also a
logical axiom of L.
The reason for calling the instances of A1–A8 the logical axioms,
instead of just axioms, is to avoid conflict with Definition 6.8.
Problem 7.3. Determine whether or not each of the following for-
mulas is a logical axiom.
1. ∀x ∀z (x = y → (x = c → x = y))
2. x = y → (y = z → z = x)
3. ∀z (x = y → (x = c → y = c))
4. ∀w ∃x (P wx → P ww) → ∃x (P xx → P xx)
5. ∀x (∀x c = fxc → ∀x ∀x c = fxc)
6. (∃x P x → ∃y ∀z Rzfy) → ((∃x P x → ∀y ¬∀z Rzfy) → ∀x ¬P x)
Proposition 7.4. Every logical axiom is a tautology.
Note that we have recycled our axiom schemas A1—A3 from propo-
sitional logic. We will also recycle MP as the sole rule of inference for
first-order logic.
Definition 7.5 (Modus Ponens). Given the formulas ϕ and (ϕ →
ψ), one may infer ψ.
As in propositional logic, we will usually refer to Modus Ponens by
its initials, MP. That MP preserves truth in the sense of Chapter 6
follows from Problem 6.10. Using the logical axioms and MP, we can
execute deductions in first-order logic just as we did in propositional
logic.
Definition 7.6. Let ∆ be a set of formulas of the first-order lan-
guage L. A deduction or proof from ∆ in L is a finite sequence
ϕ1ϕ2 . . . ϕn of formulas of L such that for each k ≤ n,
1. ϕk is a logical axiom, or
2. ϕk ∈ ∆, or
3. there are i, j < k such that ϕk follows from ϕi and ϕj by MP.
A formula of ∆ appearing in the deduction is usually referred to as a
premiss of the deduction. ∆ proves a formula α, written as ∆ ` α, if
α is the last formula of a deduction from ∆. We’ll usually write ` α
instead of ∅ ` α. Finally, if Γ and ∆ are sets of formulas, we’ll take
Γ ` ∆ to mean that Γ ` δ for every formula δ ∈ ∆.
Note. We have reused the axiom schema, the rule of inference, and
the definition of deduction from propositional logic. It follows that any
44 7. DEDUCTIONS

deduction of propositional logic can be converted into a deduction of

first-order logic simply by replacing the formulas of LP occurring in
the deduction by first-order formulas. Feel free to appeal to the deduc-
tions in the exercises and problems of Chapter 3. You should probably
review the Examples and Problems of Chapter 3 before going on, since
most of the rest of this Chapter concentrates on what is different about
deductions in first-order logic.
Example 7.1. We’ll show that {α} ` ∃x α for any first-order for-
mula α and any variable x.
1. (∀x ¬α → ¬α) → (α → ¬∀x ¬α) Problem 3.9.5
2. ∀x ¬α → ¬α A4
3. α → ¬∀x ¬α 1,2 MP
4. α Premiss
5. ¬∀x ¬α 3,4 MP
6. ∃x α Definition of ∃
Strictly speaking, the last line is just for our convenience, like ∃ itself.
Problem 7.5. Show that:
1. ` ∀x ϕ → ∀y ϕxy , if y does not occur at all in ϕ.
2. ` α ∨ ¬α.
3. {c = d} ` ∀z Qazc → Qazd.
4. ` x = y → y = x.
5. {∃x α} ` α if x does not occur free in α.
Many general facts about deductions can be recycled from propo-
sitional logic, including the Deduction Theorem.
Proposition 7.6. If ϕ1 ϕ2 . . . ϕn is a deduction of L, then ϕ1 . . . ϕ`
is also a deduction of L for any ` such that 1 ≤ ` ≤ n.
Proposition 7.7. If Γ ` δ and Γ ` δ → β, then Γ ` β.
Proposition 7.8. If Γ ⊆ ∆ and Γ ` α, then ∆ ` α.
Proposition 7.9. Then if Γ ` ∆ and ∆ ` σ, then Γ ` σ.
Theorem 7.10 (Deduction Theorem). If Σ is any set of formulas
and α and β are any formulas, then Σ ` α → β if and only if Σ∪{α} `
β.
Just as in propositional logic, the Deduction Theorem is useful be-
cause it often lets us take shortcuts when trying to show that deductions
exist. There is also another result about first-order deductions which
often supplies useful shortcuts.
 7. DEDUCTIONS 45

Theorem 7.11 (Generalization Theorem). Suppose x is a variable,

Γ is a set of formulas in which x does not occur free, and ϕ is a formula
such that Γ ` ϕ. Then Γ ` ∀x ϕ.
Theorem 7.12 (Generalization On Constants). Suppose that c is
a constant symbol, Γ is a set of formulas in which c does not occur, and
ϕ is a formula such that Γ ` ϕ. Then there is a variable x which does
not occur in ϕ such that Γ ` ∀x ϕcx.1 Moreover, there is a deduction of
∀x ϕcx from Γ in which c does not occur.
Example 7.2. We’ll show that if ϕ and ψ are any formulas, x is
any variable, and ` ϕ → ψ, then ` ∀x ϕ → ∀x ψ.
Since x does not occur free in any formula of ∅, it follows from
` ϕ → ψ by the Generalization Theorem that ` ∀x (ϕ → ψ). But then
1. ∀x (ϕ → ψ) above
2. ∀x (ϕ → ψ) → (∀x ϕ → ∀x ψ) A5
3. ∀x ϕ → ∀x ψ 1,2 MP
is the tail end of a deduction of ∀x ϕ → ∀x ψ from ∅.
Problem 7.13. Show that:
1. ` ∀x ∀y ∀z (x = y → (y = z → x = z)).
2. ` ∀x α → ∃x α.
3. ` ∃x γ → ∀x γ if x does not occur free in γ.
We conclude with a bit of terminology.
Definition 7.7. If Σ is a set of sentences, then the theory of Σ is
Th(Σ) = { τ | τ is a sentence and Σ ` τ }.
That is, the theory of Σ is just the collection of all sentences which
can be proved from Σ.

1 c
ϕx is ϕ with every occurence of the constant c replaced by x.
46 7. DEDUCTIONS
 CHAPTER 8

Soundness and Completeness

As with propositional logic, first-order logic had better satisfy the

Soundness Theorem and it is desirable that it satisfy the Completeness
Theorem. These theorems do hold for first-order logic. The Soundness
Theorem is proved in a way similar to its counterpart for propositional
logic, but the Completeness Theorem will require a fair bit of additional
work.1 It is in this extra work that the distinction between formulas
and sentences becomes useful.
Let L be a fixed countable first-order language throughout this
chapter. All formulas will be assumed to be formulas of L unless stated
otherwise.
First, we rehash many of the definitions and facts we proved for
propositional logic in Chapter 4 for first-order logic.
Theorem 8.1 (Soundness Theorem). If α is a sentence and ∆ is
a set of sentences such that ∆ ` α, then ∆ |= α.
Definition 8.1. A set of sentences Γ is inconsistent if Γ ` ¬(ψ →
ψ) for some formula ψ, and is consistent if it is not inconsistent.
Recall that a set of sentences Γ is satisfiable if M |= Γ for some
structure M.
Proposition 8.2. If a set of sentences Γ is satisfiable, then it is
consistent.
Proposition 8.3. Suppose ∆ is an inconsistent set of sentences.
Then ∆ ` ψ for any formula ψ.
Proposition 8.4. Suppose Σ is an inconsistent set of sentences.
Then there is a finite subset ∆ of Σ such that ∆ is inconsistent.
Corollary 8.5. A set of sentences Γ is consistent if and only if
every finite subset of Γ is consistent.
1
This is not too surprising because of the greater complexity of first-order logic.
Also, it turns out that first-order logic is about as powerful as a logic can get and
still have the Completeness Theorem hold.
47
48 8. SOUNDNESS AND COMPLETENESS

Definition 8.2. A set of sentences Σ is maximally consistent if Σ

is consistent but Σ ∪ {τ } is inconsistent whenever τ is a sentence such
that τ ∈/ Σ.
One quick way of finding examples of maximally consistent sets is
given by the following proposition.
Proposition 8.6. If M is a structure, then Th(M) is a maximally
consistent set of sentences.
Example 8.1. M = ({5}) is a structure for L= , so Th(M) is a
maximally consistent set of sentences. Since it turns out that Th(M) =
Th ({ ∀x ∀y x = y }), this also gives us an example of a set of sentences
Σ = { ∀x ∀y x = y } such that Th(Σ) is maximally consistent.
Proposition 8.7. If Σ is a maximally consistent set of sentences,
τ is a sentence, and Σ ` τ , then τ ∈ Σ.
Proposition 8.8. Suppose Σ is a maximally consistent set of sen-
tences and τ is a sentence. Then ¬τ ∈ Σ if and only if τ ∈
/ Σ.
Proposition 8.9. Suppose Σ is a maximally consistent set of sen-
tences and ϕ and ψ are any sentences. Then ϕ → ψ ∈ Σ if and only if
ϕ∈/ Σ or ψ ∈ Σ.
Theorem 8.10. Suppose Γ is a consistent set of sentences. Then
there is a maximally consistent set of sentences Σ with Γ ⊆ Σ.
The counterparts of these notions and facts for propositional logic
sufficed to prove the Completeness Theorem, but here we will need
some additional tools. The basic problem is that instead of defining a
suitable truth assignment from a maximally consistent set of formulas,
we need to construct a suitable structure from a maximally consistent
set of sentences. Unfortunately, structures for first-order languages are
usually more complex than truth assignments for propositional logic.
The following definition supplies the key new idea we will use to prove
the Completeness Theorem.
Definition 8.3. Suppose Σ is a set of sentences and C is a set of
(some of the) constant symbols of L. Then C is a set of witnesses for
Σ in L if for every formula ϕ of L with at most one free variable x,
there is a constant symbol c ∈ C such that Σ ` ∃x ϕ → ϕxc.
The idea is that every element of the universe which Σ proves must
exist is named, or “witnessed”, by a constant symbol in C. Note that
if Σ ` ¬∃x ϕ, then Σ ` ∃x ϕ → ϕxc for any constant symbol c.
 8. SOUNDNESS AND COMPLETENESS 49

Proposition 8.11. Suppose Γ and Σ are sets of sentences of L,

Γ ⊆ Σ, and C is a set of witnesses for Γ in L. Then C is a set of
witnesses for Σ in L.
Example 8.2. Let L0O be the first-order language with a single 2-
place relation symbol, <, and countably many constant symbols, cq for
each q ∈ Q. Let Σ include all the sentences
1. cp < cq , for every p, q ∈ Q such that p < q,
2. ∀x (¬x < x),
3. ∀x ∀y (x < y ∨ x = y ∨ y < x),
4. ∀x ∀y ∀z (x < y → (y < z → x < z)),
5. ∀x ∀y (x < y → ∃z (x < z ∧ z < y)),
6. ∀x ∃y (x < y), and
7. ∀x ∃y (y < x).
In effect, Σ asserts that < is a linear order on the universe (2–4) which
is dense (5) and has no endpoints (6–7), and which has a suborder
isomorphic to Q (1). Then C = { cq | q ∈ Q } is a set of witnesses for
Σ in L0O .
In the example above, one can “reverse-engineer” a model for the set
of sentences in question from the set of witnesses simply by letting the
universe of the structure be the set of witnesses. One can also define the
necessary relation interpreting < in a pretty obvious way from Σ.2 This
example is obviously contrived: there are no constant symbols around
which are not witnesses, Σ proves that distinct constant symbols aren’t
equal to to each other, there is little by way of non-logical symbols
needing interpretation, and Σ explicitly includes everything we need to
know about <.
In general, trying to build a model for a set of sentences Σ in this
way runs into a number of problems. First, how do we know whether
Σ has a set of witnesses at all? Many first-order languages have few or
no constant symbols, after all. Second, if Σ has a set of witnesses C,
it’s unlikely that we’ll be able to get away with just letting the universe
of the model be C. What if Σ ` c = d for some distinct witnesses c
and d? Third, how do we handle interpreting constant symbols which
are not in C? Fourth, what if Σ doesn’t prove enough about whatever
relation and function symbols exist to let us define interpretations of
them in the structure under construction? (Imagine, if you like, that
someone hands you a copy of Joyce’s Ulysses and asks you to produce a
complete road map of Dublin on the basis of the book. Even if it has no
2
Note, however, that an isomorphic copy of Q is not the only structure for L0O
satisfying Σ. For example, R = (R, <, q +π : q ∈ Q) will also satisfy Σ if we intepret
cq by q + π.
50 8. SOUNDNESS AND COMPLETENESS

geographic contradictions, you are unlikely to find all the information

in the novel needed to do the job.) Finally, even if Σ does prove all we
need to define functions and relations on the universe to interpret the
function and relation symbols, just how do we do it? Getting around
all these difficulties requires a fair bit of work. One can get around
many by sticking to maximally consistent sets of sentences in suitable
languages.
Lemma 8.12. Suppose Σ is a set of sentences, ϕ is any formula,
and x is any variable. Then Σ ` ϕ if and only if Σ ` ∀x ϕ.
Theorem 8.13. Suppose Γ is a consistent set of sentences of L.
Let C be an infinite countable set of constant symbols which are not
symbols of L, and let L0 = L∪C be the language obtained by adding the
constant symbols in C to the symbols of L. Then there is a maximally
consistent set Σ of sentences of L0 such that Γ ⊆ Σ and C is a set of
witnesses for Σ.
This theorem allows one to use a certain measure of brute force:
No set of witnesses? Just add one! The set of sentences doesn’t decide
enough? Decide everything one way or the other!
Theorem 8.14. Suppose Σ is a maximally consistent set of sen-
tences and C is a set of witnesses for Σ. Then there is a structure M
such that M |= Σ.
The important part here is to define M — proving that M |= Σ
is tedious but fairly straightforward if you have the right definition.
Proposition 6.17 now lets us deduce the fact we really need.
Corollary 8.15. Suppose Γ is a consistent set of sentences of a
first-order language L. Then there is a structure M for L satisfying Γ.
With the above facts in hand, we can rejoin our proof of Soundness
and Completeness, already in progress:
Theorem 8.16. A set of sentences Σ in L is consistent if and only
if it is satisfiable.
The rest works just like it did for propositional logic.
Theorem 8.17 (Completeness Theorem). If α is a sentence and
∆ is a set of sentences such that ∆ |= α, then ∆ ` α.
It follows that in a first-order logic, as in propositional logic, a
sentence is implied by some set of premisses if and only if it has a proof
from those premisses.
Theorem 8.18 (Compactness Theorem). A set of sentences ∆ is
satisfiable if and only if every finite subset of ∆ is satisfiable.
 CHAPTER 9

Applications of Compactness

After wading through the preceding chapters, it should be obvious

that first-order logic is, in principle, adequate for the job it was origi-
nally developed for: the essentially philosophical exercise of formalizing
most of mathematics. As something of a bonus, first-order logic can
supply useful tools for doing “real” mathematics. The Compactness
Theorem is the simplest of these tools and glimpses of two ways of
using it are provided below.

From the finite to the infinite. Perhaps the simplest use of the
Compactness Theorem is to show that if there exist arbitrarily large
finite objects of some type, then there must also be an infinite object
of this type.
Example 9.1. We will use the Compactness Theorem to show that
there is an infinite commutative group in which every element is of order
2, i.e. such that g · g = e for every element g.
Let LG be the first-order language with just two non-logical sym-
bols:
• Constant symbol: e
• 2-place function symbol: ·
Here e is intended to name the group’s identity element and · the group
operation. Let Σ be the set of sentences of LG including:
1. The axioms for a commutative group:
• ∀x x · e = x
• ∀x ∃y x · y = e
• ∀x ∀y ∀z x · (y · z) = (x · y) · z
• ∀x ∀y y · x = x · y
2. A sentence which asserts that every element of the universe is of
order 2:
• ∀x x · x = e
3. For each n ≥ 2, a sentence, σn , which asserts that there are at
least n different elements in the universe:
• ∃x1 . . . ∃xn ((¬x1 = x2) ∧ (¬x1 = x3 ) ∧ · · · ∧ (¬xn−1 = xn ))
51
52 9. APPLICATIONS OF COMPACTNESS

We claim that every finite subset of Σ is satisfiable. The most

direct way to verify this is to show how, given a finite subset ∆ of Σ,
to produce a model M of ∆. Let n be the largest integer such that
σn ∈ ∆ ∪ {σ2} (Why is there such an n?) and choose an integer k such
that 2k ≥ n. Define a structure (G, ◦) for LG as follows:
• G = { ha` | 1 ≤ ` ≤ ki | a` = 0 or 1 }
• ha` | 1 ≤ ` ≤ ki ◦ hb` | 1 ≤ ` ≤ ki = ha` + b` (mod 2) | 1 ≤ ` ≤ ki
That is, G is the set of binary sequences of length k and ◦ is coordi-
natewise addition modulo 2 of these sequences. It is easy to check that
(G, ◦) is a commutative group with 2k elements in which every element
has order 2. Hence (G, ◦) |= ∆, so ∆ is satisfiable.
Since every finite subset of Σ is satisfiable, it follows by the Com-
pactness Theorem that Σ is satisfiable. A model of Σ, however, must
be an infinite commutative group in which every element is of order
2. (To be sure, it is quite easy to build such a group directly; for ex-
ample, by using coordinatewise addition modulo 2 of infinite binary
sequences.)
Problem 9.1. Use the Compactness Theorem to show that there
is an infinite
1. bipartite graph,
2. non-commutative group, and
3. field of characteristic 3,
and also give concrete examples of such objects.
Most applications of this method, including the ones above, are
not really interesting: it is usually more valuable, and often easier, to
directly construct examples of the infinite objects in question rather
than just show such must exist. Sometimes, though, the technique
can be used to obtain a non-trivial result more easily than by direct
methods. We’ll use it to prove an important result from graph theory,
Ramsey’s Theorem. Some definitions first:
Definition 9.1. If X is a set, let the set of unordered pairs of
elements of X be [X]2 = { {a, b} | a, b ∈ X and a 6= b }. (See Defini-
tion A.1.)
1. A graph is a pair (V, E) such that V is a non-empty set and
E ⊆ [V ]2. Elements of V are called vertices of the graph and
elements of E are called edges.
2. A subgraph of (V, E) is a pair (U, F ), where U ⊂ V and F =
E ∩ [U]2.
3. A subgraph (U, F ) of (V, E) is a clique if F = [U]2.
4. A subgraph (U, F ) of (V, E) is an independent set if F = ∅.
 9. APPLICATIONS OF COMPACTNESS 53

That is, a graph is some collection of vertices, some of which are

joined to one another. A subgraph is just a subset of the vertices,
together with all edges joining vertices of this subset in the whole graph.
It is a clique if it happens that the original graph joined every vertex in
the subgraph to all other vertices in the subgraph, and an independent
set if it happens that the original graph joined none of the vertices in
the subgraph to each other. The question of when a graph must have
a clique or independent set of a given size is of some interest in many
applications, especially in dealing with colouring problems.
Theorem 9.2 (Ramsey’s Theorem). For every n ≥ 1 there is an
integer Rn such that any graph with at least Rn vertices has a clique
with n vertices or an independent set with n vertices.
Rn is the nth Ramsey number . It is easy to see that R1 = 1 and
R2 = 2, but R3 is already 6, and Rn grows very quickly as a function
of n thereafter. Ramsey’s Theorem is fairly hard to prove directly, but
the corresponding result for infinite graphs is comparatively straight-
forward.
Lemma 9.3. If (V, E) is a graph with infinitely many vertices, then
it has an infinite clique or an infinite independent set.
A relatively quick way to prove Ramsey’s Theorem is to first prove
its infinite counterpart, Lemma 9.3, and then get Ramsey’s Theorem
out of it by way of the Compactness Theorem. (If you’re an ambitious
minimalist, you can try to do this using the Compactness Theorem for
propositional logic instead!)
Elementary equivalence and non-standard models. One of
the common uses for the Compactness Theorem is to construct “non-
standard” models of the theories satisfied by various standard math-
ematical structures. Such a model satisfies all the same first-order
sentences as the standard model, but differs from it in some way not
expressible in the first-order language in question. This brings home
one of the intrinsic limitations of first-order logic: it can’t always tell
essentially different structures apart. Of course, we need to define just
what constitutes essential difference.
Definition 9.2. Suppose L is a first-order language and N and M
are two structures for L. Then N and M are:
1. isomorphic, written as N ∼ = M, if there is a function F : |N| →
|M| such that
(a) F is 1 − 1 and onto,
(b) F (cN) = cM for every constant symbol c of L,
54 9. APPLICATIONS OF COMPACTNESS

(c) F (f N (a1, . . . , ak ) = f M (F (a1), . . . , F (ak )) for every k-place

function symbol f of L and elements a1, . . . , ak ∈ |N|, and
(d) P N (a1, . . . , ak ) holds if and only if P N (F (a1), . . . , F (ak ))
for every k-place relation symbol of L and elements a1 ,
. . . , ak of |N|;
and
2. elementarily equivalent, written as N ≡ M, if Th(N) = Th(M),
i.e. if N |= σ if and only if M |= σ for every sentence σ of L.
That is, two structures for a given language are isomorphic if they
are structurally identical and elementarily equivalent if no statement
in the language can distinguish between them. Isomorphic structures
are elementarily equivalent:
Proposition 9.4. Suppose L is a first-order language and N and
M are structures for L such that N ∼
= M. Then N ≡ M.
However, as the following application of the Compactness Theorem
shows, elementarily equivalent structures need not be isomorphic:
Example 9.2. Note that C = (N) is an infinite structure for L= .
Expand L= to LR by adding a constant symbol cr for every real number
r, and let Σ be the set of sentences of L= including
• every sentence τ of Th(C), i.e. such that C |= τ , and
• ¬cr = cs for every pair of real numbers r and s such that r 6= s.
Every finite subset of Σ is satisfiable. (Why?) Thus, by the Compact-
ness Theorem, there is a structure U0 for LR satisfying Σ, and hence
Th(C). The structure U obtained by dropping the interpretations of
all the constant symbols cr from U0 is then a structure for L= which
satisfies Th(C). Note that |U| = |U0 | is at least large as the set of all
real numbers R, since U0 requires a distinct element of the universe to
interpret each constant symbol cr of LR .
Since Th(C) is a maximally consistent set of sentences of L= by
Problem 8.6, it follows from the above that C ≡ U. On the other hand,
C cannot be isomorphic to U because there cannot be an onto map
between a countable set, such as N = |C|, and a set which is at least
as large as R, such as |U|.
In general, the method used above can be used to show that if a
set of sentences in a first-order language has an infinite model, it has
many different ones. In L= that is essentially all that can happen:
Proposition 9.5. Two structures for L= are elementarily equiva-
lent if and only if they are isomorphic or infinite.
 9. APPLICATIONS OF COMPACTNESS 55

Problem 9.6. Let N = (N, 0, 1, S, +, ·, E) be the standard struc-

ture for LN . Use the Compactness Theorem to show there is a structure
M for LN such that N ≡ N but not N ∼ = M.
Note that because N and M both satisfy Th(N), which is maximally
consistent by Problem 8.6, there is absolutely no way of telling them
apart in LN .
Proposition 9.7. Every model of Th(N) which is not isomorphic
to N has
1. an isomorphic copy of N embedded in it,
2. an infinite number, i.e. one larger than all of those in the copy
of N, and
3. an infinite decreasing sequence.
The apparent limitation of first-order logic that non-isomorphic
structures may be elementarily equivalent can actually be useful. A
non-standard model may have features that make it easier to work
with than the standard model one is really interested in. Since both
structures satisfy exactly the same sentences, if one uses these features
to prove that some statement expressible in the given first-order lan-
guage is true about the non-standard structure, one gets for free that
it must be true of the standard structure as well. A prime example of
this idea is the use of non-standard models of the real numbers con-
taining infinitesimals (numbers which are infinitely small but different
from zero) in some areas of analysis.
Theorem 9.8. Let R = (R, 0, 1, +, ·) be the field of real numbers,
considered as a structure for LF . Then there is a model of Th(R) which
contains a copy of R and in which there is an infinitesimal.
The non-standard models of the real numbers actually used in anal-
ysis are usually obtained in more sophisticated ways in order to have
more information about their internal structure. It is interesting to
note that infinitesimals were the intuition behind calculus for Leibniz
when it was first invented, but no one was able to put their use on a
rigourous footing until Abraham Robinson did so in 1950.
56 9. APPLICATIONS OF COMPACTNESS
Hints
 CHAPTER 1

Hints

1.1. Symbols not in the language, unbalanced parentheses, lack of

connectives . . .
1.2. Proceed by induction on the length of the formula or on the
number of connectives in the formula.
1.3. Compute p(α)/`(α) for a number of examples and look for
patterns. Getting a minimum value should be pretty easy.
1.4. Proceed by induction on the length of or on the number of
connectives in the formula.
1.5. Construct examples of formulas of all the short lengths that
you can, and then see how you can make longer formulas out of short
ones.
1.6. Hewlett-Packard sells calculators that use such a trick. A sim-
ilar one is used in Definition 5.2.
1.7. Observe that LP has countably many symbols and that every
formula is a finite sequence of symbols. The relevant facts from set
theory are given in Appendix A.
1.8. Stick several simple statements together with suitable connec-
tives.
1.9. This should be straightforward.
1.10. Ditto.
1.11. To make sure you get all the subformulas, write out the for-
mula in official form with all the parentheses.
1.12. Proceed by induction on the length or number of connectives
of the formula.

59
60 1. HINTS
 CHAPTER 2

Hints

2.1. Use truth tables.

2.2. Proceed by induction on the length of δ or on the number of
connectives in δ.
2.3. Use Proposition 2.2.
2.4. In each case, unwind Definition 2.1 and the definitions of the
abbreviations.
2.5. Use truth tables.
2.6. Use Definition 2.3 and Proposition 2.4.
2.7. If a truth assignment satisfies every formula in Σ and every
formula in Γ is also in Σ, then . . .
2.8. Grinding out an appropriate truth table will do the job. Why
is it important that Σ be finite here?
2.9. Use Definition 2.4 and Proposition 2.4.
2.10. Use Definitions 2.3 and 2.4. If you have trouble trying to
prove one of the two directions directly, try proving its contrapositive
instead.

61
62 2. HINTS
 CHAPTER 3

Hints

3.1. Truth tables are probably the best way to do this.

3.2. Look up Proposition 2.4.
3.3. There are usually many different deductions with a given con-
clusion, so you shouldn’t take the following hints as gospel.
1. Use A2 and A1.
2. Recall what ∨ abbreviates.
3.4. You need to check that ϕ1 . . . ϕ` satisfies the three conditions
of Definition 3.3; you know ϕ1 . . . ϕn does.
3.5. Put together a deduction of β from Γ from the deductions of
δ and δ → β from Γ.
3.6. Examine Definition 3.3 carefully.
3.7. The key idea is similar to that for proving Proposition 3.5.
3.8. One direction follows from Proposition 3.5. For the other di-
rection, proceed by induction on the length of the shortest proof of β
from Σ ∪ {α}.
3.9. Again, don’t take these hints as gospel. Try using the Deduc-
tion Theorem in each case, plus
1. A3.
2. A3 and Problem 3.3.
3. A3.
4. A3, Problem 3.3, and Example 3.2.
5. Some of the above parts and Problem 3.3.
6. Ditto.
7. Use the definition of ∨ and one of the above parts.
8. Use the definition of ∧ and one of the above parts.
9. Aim for ¬α → (α → ¬β) as an intermediate step.

63
64 3. HINTS
 CHAPTER 4

Hints

4.1. Use induction on the length of the deduction and Proposition

3.2.
4.2. Assume, by way of contradiction, that the given set of formulas
is inconsistent. Use the Soundness Theorem to show that it can’t be
satisfiable.
4.3. First show that {¬(α → α)} ` ψ.
4.4. Note that deductions are finite sequences of formulas.
4.5. Use Proposition 4.4.
4.6. Use Proposition 4.2, the definition of Σ, and Proposition 2.4.
4.7. Assume, by way of contradiction, that ϕ ∈
/ Σ. Use Definition
4.2 and the Deduction Theorem to show that Σ must be inconsistent.
4.8. Use Definition 4.2 and Problem 3.9.
4.9. Use Definition 4.2 and Proposition 4.8.
4.10. Use Proposition 1.7 and induction on a list of all the formulas
of LP .
4.11. One direction is just Proposition 4.2. For the other, expand
the set of formulas in question to a maximally consistent set of formulas
Σ using Theorem 4.10, and define a truth assignment v by setting
v(An) = T if and only if An ∈ Σ. Now use induction on the length of
ϕ to show that ϕ ∈ Σ if and only if v satisfies ϕ.
4.12. Prove the contrapositive using Theorem 4.11.
4.13. Put Corollary 4.5 together with Theorem 4.11.

65
66 4. HINTS
 CHAPTER 5

Hints

5.1. Try to disassemble each string using Definition 5.2. Note that
some might be valid terms of more than one of the given languages.
5.2. This is similar to Problem 1.5.
5.3. This is similar to Proposition 1.7.
5.4. Try to disassemble each string using Definitions 5.2 and 5.3.
Note that some might be valid formulas of more than one of the given
languages.
5.5. This is just like Problem 1.2.
5.6. This is similar to Problem 1.5. You may wish to use your
solution to Problem 5.2.
5.7. This is similar to Proposition 1.7.
5.8. You might want to rephrase some of the given statements to
make them easier to formalize.
1. Look up associativity if you need to.
2. “There is an object such that every object is not in it.”
3. This should be easy.
4. Ditto.
5. “Any two things must be the same thing.”
5.9. If necessary, don’t hesitate to look up the definitions of the
given structures.
1. Read the discussion at the beginning of the chapter.
2. You really need only one non-logical symbol.
3. There are two sorts of objects in a vector space, the vectors
themselves and the scalars of the field, which you need to be
able to tell apart.
5.10. Use Definition 5.3 in the same way that Definition 1.2 was
used in Definition 1.3.
5.11. The scope of a quantifier ought to be a certain subformula of
the formula in which the quantifier occurs.
67
68 5. HINTS

5.12. Check to see whether they satisfy Definition 5.4.

5.13. Check to see which pairs satisfy Definition 5.5.
5.14. Proceed by induction on the length of ϕ using Definition 5.3.
5.15. This is similar to Theorem 1.12.
5.16. This is similar to Theorem 1.12 and uses Theorem 5.15.
 CHAPTER 6

Hints

6.1. In each case, apply Definition 6.1.

1. This should be easy.
2. Ditto.
3. Invent objects which are completely different except that they
happen to have the right number of the right kind of components.
6.2. Figure out the relevant values of s(vn ) and apply Definition
6.3.
6.3. Suppose s and r both extend the assignment s. Show that
s(t) = r(t) by induction on the length of the term t.
6.4. Unwind the formulas using Definition 6.4 to get informal state-
ments whose truth you can determine.
6.5. Unwind the abbreviation ∃ and use Definition 6.4.
6.6. Unwind each of the formulas using Definitions 6.4 and 6.5 to
get informal statements whose truth you can determine.
6.7. This is much like Proposition 6.3.
6.8. Proceed by induction on the length of the formula using Defi-
nition 6.4 and Lemma 6.7.
6.9. How many free variables does a sentence have?
6.10. Use Definition 6.4.
6.12. Unwind the sentences in question using Definition 6.4.
6.11. Use Definitions 6.4 and 6.5; the proof is similar in form to
the proof of Proposition 2.9.
6.14. Use Definitions 6.4 and 6.5; the proof is similar in form to
the proof for Problem 2.10.
6.15. Use Definitions 6.4 and 6.5 in each case, plus the meanings
of our abbreviations.
69
70 6. HINTS

6.17. In one direction, you need to add appropriate objects to a

structure; in the other, delete them. In both cases, you still have to
verify that Γ is still satisfied.
6.18. Here are some appropriate languages.
1. L=
2. Modify your language for graph theory from Problem 5.9 by
adding a 1-place relation symbol.
3. Use your language for group theory from Problem 5.9.
4. LF
 CHAPTER 7

Hints

7.1. 1. Use Definition 7.1.

2. Ditto.
3. Ditto.
4. Proceed by induction on the length of the formula ϕ.
7.2. Use the definitions and facts about |= from Chapter 6.
7.3. Check each case against the schema in Definition 7.4. Don’t
forget that any generalization of a logical axiom is also a logical axiom.
7.4. You need to show that any instance of the schemas A1–A8 is
a tautology and then apply Lemma 7.2. That each instance of schemas
A1–A3 is a tautology follows from Proposition 6.15. For A4–A8 you’ll
have to use the definitions and facts about |= from Chapter 6.
7.5. You may wish to appeal to the deductions that you made or
were given in Chapter 3.
1. Try using A4 and A6.
2. You don’t need A4–A8 here.
3. Try using A4 and A8.
4. A8 is the key; you may need it more than once.
5. This is just A6 in disguise.
7.6. This is just like its counterpart for propositional logic.
7.7. Ditto.
7.8. Ditto.
7.9. Ditto.
7.10. Ditto.
7.11. Proceed by induction on the length of the shortest proof of
ϕ from Γ.
7.12. Ditto.
7.13. As usual, don’t take the following suggestions as gospel.
1. Try using A8.
71
72 7. HINTS

2. Start with Example 7.1.

3. Start with part of Problem 7.5.
 CHAPTER 8

Hints

8.1. This is similar to the proof of the Soundness Theorem for

propositional logic, using Proposition 6.10 in place of Proposition 3.2.
8.2. This is similar to its counterpart for prpositional logic, Propo-
sition 4.2. Use Proposition 6.10 instead of Proposition 3.2.
8.3. This is just like its counterpart for propositional logic.
8.4. Ditto.
8.5. Ditto.
8.6. This is a counterpart to Problem 4.6; use Proposition 8.2 in-
stead of Proposition 4.2 and Proposition 6.15 instead of Proposition
2.4.
8.7. This is just like its counterpart for propositional logic.
8.8. Ditto
8.9. Ditto.
8.10. This is much like its counterpart for propositional logic, The-
orem 4.10.
8.11. Use Proposition 7.8.
8.12. Use the Generalization Theorem for the hard direction.
8.13. This is essentially a souped-up version of Theorem 8.10. To
ensure that C is a set of witnesses of the maximally consistent set of
sentences, enumerate all the formulas ϕ of L0 with one free variable
and take care of one at each step in the inductive construction.
8.14. To construct the required structure, M, proceed as follows.
Define an equivalence relation ∼ on C by setting c ∼ d if and only if
c = d ∈ Σ, and let [c] = { a ∈ C | a ∼ c } be the equivalence class of
c ∈ C. The universe of M will be M = { [c] | c ∈ C }. For each k-place
function symbol f define f M by setting f M ([a1], . . . , [ak ]) = [b] if and
only if fa1 . . . ak = b is in Σ. Define the interpretations of constant
73
74 8. HINTS

symbols and relation symbols in a similar way. You need to show that
all these things are well-defined, and then show that M |= Σ.
8.15. Expand Γ to a maximally consistent set of sentences with a
set of witnesses in a suitable extension of L, apply Theorem 8.14, and
then cut down the resulting structure to one for L.
8.16. One direction is just Proposition 8.2. For the other, use
Corollary 8.15.
8.17. This follows from Theorem 8.16 in the same way that the
Completeness Theorem for propositional logic followed from Theorem
4.11.
8.18. This follows from Theorem 8.16 in the same way that the
Compactness Theorem for propositional logic followed from Theorem
4.11.
 CHAPTER 9

Hints

9.1. In each case, apply the trick used in Example 9.1. For defi-
nitions and the concrete examples, consult texts on combinatorics and
abstract algebra.
9.2. Suppose Ramsey’s Theorem fails for some n. Use the Com-
pactness Theorem to get a contradiction to Lemma 9.3 by showing
there must be an infnite graph with no clique or independent set of
size n.
9.3. Inductively define a sequence a0, a1 , . . . , of vertices so that for
every n, either it is the case that for all k ≥ n there is an edge joining
an to ak or it is the case that for all k ≥ n there is no edge joining an
to ak . There will then be a subsequence of the sequence which is an
infinite clique or a subsequence which is an infinite independent set.
9.4. The key is to figure out how, given an assignment for one
structure, one should define the corresponding assignment in the other
structure. After that, proceed by induction using the definition of
satisfaction.
9.5. When are two finite structures for L= elementarily equivalent?
9.6. In a suitable expanded language, consider Th(N) together with
the sentences ∃x 0 + x = c, ∃x S0 + x = c, ∃x SS0 + x = c, . . .
9.7. Suppose M |= Th(N) but is not isomorphic to N.
1. Consider the subset of |M| given by 0M , S M (0M ), S M (S M (0M )),
...
2. If it didn’t have one, it would be a copy of N.
3. Start with a infinite number and work down.
9.8. Expand LF by throwing in a constant symbol for every real
number, plus an extra one, and take it from there.

75
76 9. HINTS
Appendices
 APPENDIX A

A Little Set Theory

This apppendix is meant to provide an informal summary of the

notation, definitions, and facts about sets needed in Chapters 1–9. For
a proper introduction to elementary set theory, try [5] or [6].
Definition A.1. Suppose X and Y are sets. Then
1. a ∈ X means that a is an element of (i.e. a thing in) the set X.
2. X is a subset of Y , written as X ⊆ Y , if a ∈ Y for every a ∈ X.
3. The union of X and Y is X ∪ Y = { a | a ∈ X or a ∈ Y }.
4. The intersection of X and Y is X ∩Y = { a | a ∈ X and a ∈ Y }.
5. The complement of Y relative to X is X \ Y = { a | a ∈
X and a ∈/ Y }.
6. The cross product of X and Y is X ×Y = { (a, b) | a ∈ X and b ∈
Y }.
7. The power set of X is P(X) = { Z | Z ⊆ X }.
8. [X]k = { Z | Z ⊆ X and |Z| = k } is the set of subsets of X of
size k.
If all the sets being dealt with are all subsets of some fixed set Z,
the complement of Y , Ȳ , is usually taken to mean the complement
of Y relative to Z. It may sometimes be necessary to take unions,
intersections, and cross products of more than two sets.
Definition A.2. Suppose A is a set and X = { Xa | a ∈ A } is a
family of sets indexed by A. Then
S
1. The union of X is the set X =T{ z | ∃a ∈ A : z ∈ Xa }.
2. The intersection of X is the set X = { z | ∀a ∈ A : z ∈ Xa }.
3. Q
The cross
Q product of X is the set of sequences (indexed by A)
X = a∈A Xa = { ( za | a ∈ A ) | ∀a ∈ A : za ∈ Xa }.
We will denote the cross product of a set X with itself taken n times
(i.e. the set of all sequences of length n of elements of X) by X n .
Definition A.3. If X is any set, a k-place relation on X is a subset
R ⊆ X k.
For example, the set E = { 0, 2, 3, . . . } of even natural numbers is
a 1-place relation on N, D = { (x, y) ∈ N2 | x divides y } is a 2-place
79
80 A. A LITTLE SET THEORY

relation on N, and S = { (a, b, c) ∈ N3 | a + b = c } is a 3-place relation

on N. 2-place relations are usually called binary relations.
Definition A.4. A set X is finite if there is some n ∈ N such that
X has n elements, and is infinite otherwise. X is countable if it is
infinite and there is a 1-1 onto function f : N → X, and uncountable if
it is infinite but not countable.
Various infinite sets occur frequently in mathematics, such as N (the
natural numbers), Q (the rational numbers), and R (the real numbers).
Many of these are uncountable, such as R. The basic facts about
countable sets needed to do the problems are the following.
Proposition A.1. 1. If X is a countable set and Y ⊆ X, then
Y is either finite or a countable.
2. Suppose X = { Xn | n ∈ N } is a finite or countable family S of
sets such that each Xn is either finite or countable. Then X
is also finite or countable.
3. If X is a non-empty finite or countable set, then X n is finite or
countable for each n ≥ 1.
4. If X is a non-empty finite or countable set,S then the set of all
finite sequences of elements of X, X <ω = n∈N X n is countable.
The properly sceptical reader will note that setting up propositional
or first-order logic formally requires that we have some set theory in
hand, but formalizing set theory itself requires one to have first-order
logic.1

1
Which came first, the chicken or the egg? Since, biblically speaking, “In the
beginning was the Word”, maybe we ought to plump for alphabetical order. Which
begs the question: In which alphabet?
 APPENDIX B

The Greek Alphabet

A α alpha
B β beta
Γ γ gamma
∆ δ delta
E  ε epsilon
Z ζ zeta
H η eta
Θ θ ϑ theta
I ι iota
K κ kappa
Λ λ lambda
M µ mu
N ν nu
O o omicron
Ξ ξ xi
Π π $ pi
P ρ % rho
Σ σ ς sigma
T τ tau
Υ υ upsilon
Φ φ ϕ phi
X χ chi
Ψ ψ psi
Ω ω omega

81
82 B. THE GREEK ALPHABET
 APPENDIX C

Logic Limericks

Deduction Theorem

A Theorem fine is Deduction,

For it allows work-reduction:
To show “A implies B”,
Assume A and prove B;
Quite often a simpler production.

Generalization Theorem

When in premiss the variable’s bound,

To get a “for all” without wound,
Generalization.
Not globalization;
Stay away from that management sound!

Soundness Theorem

It’s a critical logical creed:

Always check that it’s safe to proceed.
To tell us deductions
Are truthful productions,
It’s the Soundness of logic we need.

Completeness Theorem

The Completeness of logics is Gödel’s.

’Tis advice for looking for mödels:
They’re always existent
For statements consistent,
Most helpful for logical labörs.

83
84 C. LOGIC LIMERICKS
 Bibliography

[1] Jon Barwise (ed.), Handbook of Mathematical Logic, North Holland, Amster-
dam, 1977, ISBN 0-7204-2285-X.
[2] Merrie Bergman, James Moor, and Jack Nelson, The Logic Book, Random
House, NY, 1980, ISBN 0-394-32323-8.
[3] C.C. Chang and H.J. Keisler, Model Theory, third ed., North Holland, Ams-
terdam, 1990.
[4] Herbert B. Enderton, A Mathematical Introduction to Logic, Academic Press,
New York, 1972.
[5] Paul R. Halmos, Naive Set Theory, Undergraduate Texts in Mathematics,
Springer-Verlag, New York, 1974, ISBN 0-387-90092-6.
[6] James M. Henle, An Outline of Set Theory, Problem Books in Mathematics,
Springer-Verlag, New York, 1986, ISBN 0-387-96368-5.
[7] Douglas R. Hofstadter, Gödel, Escher, Bach, Random House, New York, 1979,
ISBN 0-394-74502-7.
[8] Jerome Malitz, Introduction to Mathematical Logic, Springer-Verlag, New
York, 1979, ISBN 0-387-90346-1.
[9] Yu.I. Manin, A Course in Mathematical Logic, Graduate Texts in Mathemat-
ics 53, Springer-Verlag, New York, 1977, ISBN 0-387-90243-0.
[10] Roger Penrose, The Emperor’s New Mind, Oxford University Press, Oxford,
1989.

85
86 BIBLIOGRAPHY
 Index

(, 7, 24 R, 80
), 7, 24 Rn , 53
=, 24, 25 S, 10
∩, 79 T , 11
∪, 79 Th, 39, 45
∃, 30 vn , 24
∀, 24, 25, 30 X n , 79
↔, 9, 30 [X]k , 79
∈, 79 Ȳ , 79
∧, 9, 30
¬, 7, 24, 25 abbreviations, 9, 30
∨, 9, 30 all, 2
|=, 14, 35, 37 and, 2, 9
2, assignment, 34, 35
Q 14, 36, 37 extended, 35
, 79
`, 16, 43 truth, 11
\, 79 atomic formulas, 7, 27
⊆, 79 axiom, 15, 28, 39
×, 79 logical, 43
→, 7, 24, 25 schema, 15, 42
A1, 15, 42
An , 7 A2, 15, 42
F , 11 A3, 15, 42
ϕxt , 42 A4, 42
L, 24 A5, 42
L= , 26 A6, 42
L1 , 26 A7, 42
LF , 26 A8, 42
LG, 51 bound variable, 29
LN , 26
LNT , 25 chicken, 80
LO , 26 clique, 52
LP , 7 Compactness Theorem, 20, 50
LS , 26 applications of, 51
M, 33 complement, 79
N, 80 Completeness Theorem, 20, 50, 83
N, 33 connectives, 7, 9, 24
P, 79 consistent, 19, 47
Q, 80 maximally, 19, 48
87
88 INDEX

constant, 24, 25, 31, 33, 35 isomorphism of structures, 53

contradiction, 13, 38
convention John, 80
common symbols, 25
language, 26, 31
parentheses, 9, 30
extension of, 30
countable, 80
first-order, 23
cross product, 79
formal, 1
deduction, 16, 43 natural, 1
Deduction Theorem, 17, 44, 83 propositional, 7
limericks, 83
edge, 52 logic
egg, 80 first-order, 2, 23
element, 79 mathematical, 1
elementary equivalence, 54 natural deductive, 1
equality, 24, 25 predicate, 7
equivalence propositional, 2, 7
elementary, 54 sentential, 7
existential quantifier, 30 logical axiom, 43
extension of a language, 30
mathematical logic, 1
finite, 80 maximally consistent, 19, 48
first-order metalanguage, 31
languages, 23 metatheorem, 31
logic, 2, 23 model, 37
for all, 25 Modus Ponens, 15, 43
formula, 7, 27 MP, 15, 43
atomic, 7, 27
natural deductive logic, 1
unique readability, 10, 32
non-standard model, 53, 55
free variable, 29
of the real numbers, 55
function, 24, 31, 33, 35
not, 2, 7, 25
k-place, 24, 25
or, 2, 9
generalization, 42
Generalization Theorem, 45, 83 parentheses, 7, 24
on Constants, 45 conventions, 9, 30
gothic characters, 33 doing without, 8
graph, 52 power set, 79
Greek characters, 7, 28, 81 predicate, 24, 25
predicate logic, 7
if and only if, 9 premiss, 16, 43
if . . . then, 2, 7, 25 proof, 16, 43
implies, 14, 38 propositional logic, 2, 7
inconsistent, 19, 47 proves, 16, 43
independent set, 52 punctuation, 7, 25
inference rule, 15
infinite, 80 quantifier
Infinite Ramsey’s Theorem, 53 existential, 30
infinitesimal, 55 scope of, 30
intersection, 79 universal, 24, 25, 30
 INDEX 89

Ramsey number, 53 witnesses, 48

Ramsey’s Theorem, 53 Word, 80
Infinite, 53
relation, 24, 31, 33
k-place, 24, 25, 79
binary, 25, 80
rule of inference, 15, 43

satisfiable, 13, 37
satisfies, 13, 36, 37
scope of a quantifier, 30
sentence, 29
sentential logic, 7
set theory, 79
Soundness Theorem, 19, 47, 83
structure, 33
subformula, 10, 29
subgraph, 52
subset, 79
substitutable, 41
substitution, 41
symbols, 7, 24
logical, 24
non-logical, 24

tautology, 13, 38
term, 26, 31, 35
unique readability, 32
theorem, 31
theory, 39, 45
there is, 2
truth
assignment, 11
in a structure, 36, 37
table, 12, 13
values, 11

uncountable, 80
union, 79
unique readability
of formulas, 10, 32
of terms, 32
Unique Readability Theorem, 10, 32
universal quantifier, 30
universe, 33

variable, 24, 31, 34, 35

bound, 29
free, 29
vertex, 52