Incompleteness

and Computability
An Open Introduction to
Gödel’s Theorems

F19
Incompleteness and
Computability
The Open Logic Project

Instigator
Richard Zach, University of Calgary

Editorial Board
Aldo Antonelli,† University of California, Davis
Andrew Arana, Université Paris I Panthénon–Sorbonne
Jeremy Avigad, Carnegie Mellon University
Tim Button, University College London
Walter Dean, University of Warwick
Gillian Russell, University of North Carolina
Nicole Wyatt, University of Calgary
Audrey Yap, University of Victoria

Contributors
Samara Burns, University of Calgary
Dana Hägg, University of Calgary
Zesen Qian, Carnegie Mellon University
Incompleteness and
Computability
An Open Introduction to
Gödel’s Theorems

Remixed by Richard Zach

Fall 2019
The Open Logic Project would like to acknowledge the gener-
ous support of the Taylor Institute of Teaching and Learning of
the University of Calgary, and the Alberta Open Educational Re-
sources (ABOER) Initiative, which is made possible through an
investment from the Alberta government.

Cover illustrations by Matthew Leadbeater, used under a Cre-

ative Commons Attribution-NonCommercial 4.0 International Li-
cense.

Typeset in Baskervald X and Nimbus Sans by LATEX.

This version of Incompleteness and Computability is revision

368c6d0 (2020-04-28), with content generated from Open Logic
Text revision 534a139 (2020-06-25). Free download at:
https://ic.openlogicproject.org/

Incompleteness and Computability by

Richard Zach is licensed under a
Creative Commons Attribution 4.0
International License. It is based
on The Open Logic Text by the Open
Logic Project, used under a Creative
Commons Attribution 4.0 International
License.
Contents
About this Book x

1 Introduction to Incompleteness 1
1.1 Historical Background . . . . . . . . . . . . . . . 1
1.2 Definitions . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Overview of Incompleteness Results . . . . . . . 14
1.4 Undecidability and Incompleteness . . . . . . . . 16
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2 Recursive Functions 20
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . 20
2.2 Primitive Recursion . . . . . . . . . . . . . . . . . 21
2.3 Composition . . . . . . . . . . . . . . . . . . . . . 24
2.4 Primitive Recursion Functions . . . . . . . . . . . 26
2.5 Primitive Recursion Notations . . . . . . . . . . . 30
2.6 Primitive Recursive Functions are Computable . . 30
2.7 Examples of Primitive Recursive Functions . . . . 31
2.8 Primitive Recursive Relations . . . . . . . . . . . 35
2.9 Bounded Minimization . . . . . . . . . . . . . . . 38
2.10 Primes . . . . . . . . . . . . . . . . . . . . . . . . 39
2.11 Sequences . . . . . . . . . . . . . . . . . . . . . . 40
2.12 Trees . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.13 Other Recursions . . . . . . . . . . . . . . . . . . 45

v
vi CONTENTS

2.14 Non-Primitive Recursive Functions . . . . . . . . 47

2.15 Partial Recursive Functions . . . . . . . . . . . . 49
2.16 The Normal Form Theorem . . . . . . . . . . . . 51
2.17 The Halting Problem . . . . . . . . . . . . . . . . 52
2.18 General Recursive Functions . . . . . . . . . . . . 54
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3 Arithmetization of Syntax 58
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . 58
3.2 Coding Symbols . . . . . . . . . . . . . . . . . . . 60
3.3 Coding Terms . . . . . . . . . . . . . . . . . . . . 62
3.4 Coding Formulas . . . . . . . . . . . . . . . . . . 65
3.5 Substitution . . . . . . . . . . . . . . . . . . . . . 66
3.6 Derivations in Natural Deduction . . . . . . . . . 67
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 74

4 Representability in Q 76
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . 76
4.2 Functions Representable in Q are Computable . 79
4.3 The Beta Function Lemma . . . . . . . . . . . . . 80
4.4 Simulating Primitive Recursion . . . . . . . . . . 85
4.5 Basic Functions are Representable in Q . . . . . 86
4.6 Composition is Representable in Q . . . . . . . . 89
4.7 Regular Minimization is Representable in Q . . 91
4.8 Computable Functions are Representable in Q . 96
4.9 Representing Relations . . . . . . . . . . . . . . . 97
4.10 Undecidability . . . . . . . . . . . . . . . . . . . . 98
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 100

5 Incompleteness and Provability 101

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . 101
5.2 The Fixed-Point Lemma . . . . . . . . . . . . . . 103
5.3 The First Incompleteness Theorem . . . . . . . . 106
vii CONTENTS

5.4 Rosser’s Theorem . . . . . . . . . . . . . . . . . . 108

5.5 Comparison with Gödel’s Original Paper . . . . . 110
5.6 The Derivability Conditions for PA . . . . . . . . 111
5.7 The Second Incompleteness Theorem . . . . . . 112
5.8 Löb’s Theorem . . . . . . . . . . . . . . . . . . . 115
5.9 The Undefinability of Truth . . . . . . . . . . . . 118
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 122

6 Models of Arithmetic 123

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . 123
6.2 Reducts and Expansions . . . . . . . . . . . . . . 124
6.3 Isomorphic Structures . . . . . . . . . . . . . . . 125
6.4 The Theory of a Structure . . . . . . . . . . . . . 128
6.5 Standard Models of Arithmetic . . . . . . . . . . 129
6.6 Non-Standard Models . . . . . . . . . . . . . . . 132
6.7 Models of Q . . . . . . . . . . . . . . . . . . . . . 133
6.8 Models of PA . . . . . . . . . . . . . . . . . . . . 136
6.9 Computable Models of Arithmetic . . . . . . . . 140
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 144

7 Second-Order Logic 146

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . 146
7.2 Terms and Formulas . . . . . . . . . . . . . . . . 147
7.3 Satisfaction . . . . . . . . . . . . . . . . . . . . . 149
7.4 Semantic Notions . . . . . . . . . . . . . . . . . . 152
7.5 Expressive Power . . . . . . . . . . . . . . . . . . 153
7.6 Describing Infinite and Countable Domains . . . 154
7.7 Second-order Arithmetic . . . . . . . . . . . . . . 156
7.8 Second-order Logic is not Axiomatizable . . . . . 159
7.9 Second-order Logic is not Compact . . . . . . . . 159
7.10 The Löwenheim-Skolem Theorem Fails for
Second-order Logic . . . . . . . . . . . . . . . . . 160
7.11 Comparing Sets . . . . . . . . . . . . . . . . . . . 161
7.12 Cardinalities of Sets . . . . . . . . . . . . . . . . . 163
viii CONTENTS

7.13 The Power of the Continuum . . . . . . . . . . . 164

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 168

8 The Lambda Calculus 170

8.1 Overview . . . . . . . . . . . . . . . . . . . . . . . 170
8.2 The Syntax of the Lambda Calculus . . . . . . . 172
8.3 Reduction of Lambda Terms . . . . . . . . . . . . 173
8.4 The Church-Rosser Property . . . . . . . . . . . . 174
8.5 Currying . . . . . . . . . . . . . . . . . . . . . . . 175
8.6 Lambda Definability . . . . . . . . . . . . . . . . 176
8.7 𝜆 -Definable Arithmetical Functions . . . . . . . . 178
8.8 Pairs and Predecessor . . . . . . . . . . . . . . . . 180
8.9 Truth Values and Relations . . . . . . . . . . . . 181
8.10 Primitive Recursive Functions are 𝜆 -Definable . . 183
8.11 Fixpoints . . . . . . . . . . . . . . . . . . . . . . . 185
8.12 Minimization . . . . . . . . . . . . . . . . . . . . 189
8.13 Partial Recursive Functions are 𝜆 -Definable . . . 191
8.14 𝜆 -Definable Functions are Recursive . . . . . . . 191
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 192

A Derivations in Arithmetic Theories 194

B First-order Logic 202

B.1 First-Order Languages . . . . . . . . . . . . . . . 202
B.2 Terms and Formulas . . . . . . . . . . . . . . . . 204
B.3 Free Variables and Sentences . . . . . . . . . . . 207
B.4 Substitution . . . . . . . . . . . . . . . . . . . . . 209
B.5 Structures for First-order Languages . . . . . . . 211
B.6 Satisfaction of a Formula in a Structure . . . . . . 213
B.7 Variable Assignments . . . . . . . . . . . . . . . . 218
B.8 Extensionality . . . . . . . . . . . . . . . . . . . . 222
B.9 Semantic Notions . . . . . . . . . . . . . . . . . . 224
B.10 Theories . . . . . . . . . . . . . . . . . . . . . . . 227
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 229
ix CONTENTS

C Natural Deduction 232

C.1 Natural Deduction . . . . . . . . . . . . . . . . . 232
C.2 Rules and Derivations . . . . . . . . . . . . . . . 234
C.3 Propositional Rules . . . . . . . . . . . . . . . . . 235
C.4 Quantifier Rules . . . . . . . . . . . . . . . . . . . 236
C.5 Derivations . . . . . . . . . . . . . . . . . . . . . 238
C.6 Examples of Derivations . . . . . . . . . . . . . . 239
C.7 Derivations with Quantifiers . . . . . . . . . . . . 244
C.8 Derivations with Identity predicate . . . . . . . . 248
C.9 Proof-Theoretic Notions . . . . . . . . . . . . . . 250
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 252
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 253

D Biographies 254
D.1 Alonzo Church . . . . . . . . . . . . . . . . . . . 254
D.2 Kurt Gödel . . . . . . . . . . . . . . . . . . . . . . 255
D.3 Rózsa Péter . . . . . . . . . . . . . . . . . . . . . 257
D.4 Julia Robinson . . . . . . . . . . . . . . . . . . . . 259
D.5 Alfred Tarski . . . . . . . . . . . . . . . . . . . . 261

Photo Credits 264

Bibliography 266

About the Open Logic Project 270

About this Book
This is a textbook on Gödel’s incompleteness theorems and re-
cursive function theory. I use it as the main text when I teach
Philosophy 479 (Logic III) at the University of Calgary. It is
based on material from the Open Logic Project.
As its name suggests, the course is the third in a sequence,
so students (and hence readers of this book) are expected to be
familiar with first-order logic already. (Logic I uses the text forall
x: Calgary, and Logic II another textbook based on the OLP,
Sets, Logic, Computation.) The material assumed from Logic II,
however, is included as appendices B and C.
Logic III is a thirteen-week course, meeting three hours per
week. This is typically enough to cover the material in chapters 1
to 5 and either chapter 6 or chapter 8, depending on student inter-
est. You may want to spend more time on the basics of first-order
logic and especially on natural deduction, if students are not al-
ready familiar with it. Note that when provability in arithmetical
theories (such as Q and PA) is discussed in the main text, the
proofs of provability claims are not given using a specific proof
system. Rather, that certain claims follow from the axioms by
first-order logic is justified intuitively. However, appendix A con-
tains a number of examples of actual natural deduction deriva-
tions from the axioms of Q .

x
xi

Acknowledgments
The material in the OLP used in chapters 1 to 5 and 8 was based
originally on Jeremy Avigad’s lecture notes on “Computability
and Incompleteness,” which he contributed to the OLP. I have
heavily revised and expanded this material. The lecture notes,
e.g., based theories of arithmetic on an axiomatic proof system.
Here, we use Gentzen’s standard natural deduction system (de-
scribed in appendix C), which requires dealing with trees prim-
itive recursively (in section 2.12) and a more complicated ap-
proach to the arithmetization of derivations (in section 3.6). The
material in chapter 8 was also expanded by Zesen Qian during
his stay in Calgary as a Mitacs summer intern.
The material in the OLP on model theory and models of
arithmetic in chapter 6 was originally taken from Aldo Antonelli’s
lecture notes on “The Completeness of Classical Propositional
and Predicate Logic,” which he contributed to the OLP before
his untimely death in 2015.
The biographies of logicians in appendix D and much of the
material in appendix C are originally due to Samara Burns. Dana
Hägg originally worked on the material in appendix B.
CHAPTER 1

Introduction to
Incompleteness
1.1 Historical Background
In this section, we will briefly discuss historical developments
that will help put the incompleteness theorems in context. In
particular, we will give a very sketchy overview of the history of
mathematical logic; and then say a few words about the history
of the foundations of mathematics.
The phrase “mathematical logic” is ambiguous. One can in-
terpret the word “mathematical” as describing the subject mat-
ter, as in, “the logic of mathematics,” denoting the principles
of mathematical reasoning; or as describing the methods, as in
“the mathematics of logic,” denoting a mathematical study of the
principles of reasoning. The account that follows involves math-
ematical logic in both senses, often at the same time.
The study of logic began, essentially, with Aristotle, who lived
approximately 384–322 bce. His Categories, Prior analytics, and
Posterior analytics include systematic studies of the principles of
scientific reasoning, including a thorough and systematic study
of the syllogism.
Aristotle’s logic dominated scholastic philosophy through the
middle ages; indeed, as late as eighteenth century Kant main-

1
2 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

tained that Aristotle’s logic was perfect and in no need of revision.

But the theory of the syllogism is far too limited to model any-
thing but the most superficial aspects of mathematical reasoning.
A century earlier, Leibniz, a contemporary of Newton’s, imag-
ined a complete “calculus” for logical reasoning, and made some
rudimentary steps towards designing such a calculus, essentially
describing a version of propositional logic.
The nineteenth century was a watershed for logic. In 1854
George Boole wrote The Laws of Thought, with a thorough alge-
braic study of propositional logic that is not far from modern
presentations. In 1879 Gottlob Frege published his Begriffsschrift
(Concept writing) which extends propositional logic with quan-
tifiers and relations, and thus includes first-order logic. In fact,
Frege’s logical systems included higher-order logic as well, and
more. In his Basic Laws of Arithmetic, Frege set out to show that
all of arithmetic could be derived in his Begriffsschrift from purely
logical assumption. Unfortunately, these assumptions turned out
to be inconsistent, as Russell showed in 1902. But setting aside
the inconsistent axiom, Frege more or less invented modern
logic singlehandedly, a startling achievement. Quantificational
logic was also developed independently by algebraically-minded
thinkers after Boole, including Peirce and Schröder.
Let us now turn to developments in the foundations of math-
ematics. Of course, since logic plays an important role in mathe-
matics, there is a good deal of interaction with the developments
just described. For example, Frege developed his logic with the
explicit purpose of showing that all of mathematics could be
based solely on his logical framework; in particular, he wished
to show that mathematics consists of a priori analytic truths in-
stead of, as Kant had maintained, a priori synthetic ones.
Many take the birth of mathematics proper to have occurred
with the Greeks. Euclid’s Elements, written around 300 B.C., is
already a mature representative of Greek mathematics, with its
emphasis on rigor and precision. The definitions and proofs in
Euclid’s Elements survive more or less in tact in high school geom-
etry textbooks today (to the extent that geometry is still taught in
3 1.1. HISTORICAL BACKGROUND

high schools). This model of mathematical reasoning has been

held to be a paradigm for rigorous argumentation not only in
mathematics but in branches of philosophy as well. (Spinoza
even presented moral and religious arguments in the Euclidean
style, which is strange to see!)
Calculus was invented by Newton and Leibniz in the seven-
teenth century. (A fierce priority dispute raged for centuries, but
most scholars today hold that the two developments were for the
most part independent.) Calculus involves reasoning about, for
example, infinite sums of infinitely small quantities; these fea-
tures fueled criticism by Bishop Berkeley, who argued that belief
in God was no less rational than the mathematics of his time. The
methods of calculus were widely used in the eighteenth century,
for example by Leonhard Euler, who used calculations involving
infinite sums with dramatic results.
In the nineteenth century, mathematicians tried to address
Berkeley’s criticisms by putting calculus on a firmer foundation.
Efforts by Cauchy, Weierstrass, Bolzano, and others led to our
contemporary definitions of limits, continuity, differentiation,
and integration in terms of “epsilons and deltas,” in other words,
devoid of any reference to infinitesimals. Later in the century,
mathematicians tried to push further, and explain all aspects of
calculus, including the real numbers themselves, in terms of the
natural numbers. (Kronecker: “God created the whole numbers,
all else is the work of man.”) In 1872, Dedekind wrote “Continu-
ity and the irrational numbers,” where he showed how to “con-
struct” the real numbers as sets of rational numbers (which, as
you know, can be viewed as pairs of natural numbers); in 1888 he
wrote “Was sind und was sollen die Zahlen” (roughly, “What are
the natural numbers, and what should they be?”) which aimed
to explain the natural numbers in purely “logical” terms. In 1887
Kronecker wrote “Über den Zahlbegriff” (“On the concept of
number”) where he spoke of representing all mathematical ob-
ject in terms of the integers; in 1889 Giuseppe Peano gave formal,
symbolic axioms for the natural numbers.
The end of the nineteenth century also brought a new bold-
4 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

ness in dealing with the infinite. Before then, infinitary objects

and structures (like the set of natural numbers) were treated gin-
gerly; “infinitely many” was understood as “as many as you want,”
and “approaches in the limit” was understood as “gets as close
as you want.” But Georg Cantor showed that it was possible to
take the infinite at face value. Work by Cantor, Dedekind, and
others help to introduce the general set-theoretic understanding
of mathematics that is now widely accepted.
This brings us to twentieth century developments in logic and
foundations. In 1902 Russell discovered the paradox in Frege’s
logical system. In 1904 Zermelo proved Cantor’s well-ordering
principle, using the so-called “axiom of choice”; the legitimacy
of this axiom prompted a good deal of debate. Between 1910
and 1913 the three volumes of Russell and Whitehead’s Principia
Mathematica appeared, extending the Fregean program of estab-
lishing mathematics on logical grounds. Unfortunately, Russell
and Whitehead were forced to adopt two principles that seemed
hard to justify as purely logical: an axiom of infinity and an ax-
iom of “reducibility.” In the 1900’s Poincaré criticized the use
of “impredicative definitions” in mathematics, and in the 1910’s
Brouwer began proposing to refound all of mathematics in an
“intuitionistic” basis, which avoided the use of the law of the ex-
cluded middle (A ∨ ¬A).
Strange days indeed! The program of reducing all of math-
ematics to logic is now referred to as “logicism,” and is com-
monly viewed as having failed, due to the difficulties mentioned
above. The program of developing mathematics in terms of in-
tuitionistic mental constructions is called “intuitionism,” and is
viewed as posing overly severe restrictions on everyday mathe-
matics. Around the turn of the century, David Hilbert, one of
the most influential mathematicians of all time, was a strong sup-
porter of the new, abstract methods introduced by Cantor and
Dedekind: “no one will drive us from the paradise that Cantor
has created for us.” At the same time, he was sensitive to founda-
tional criticisms of these new methods (oddly enough, now called
“classical”). He proposed a way of having one’s cake and eating
5 1.1. HISTORICAL BACKGROUND

it too:

1. Represent classical methods with formal axioms and rules;

represent mathematical questions as formulas in an ax-
iomatic system.

2. Use safe, “finitary” methods to prove that these formal de-

ductive systems are consistent.

Hilbert’s work went a long way toward accomplishing the first

goal. In 1899, he had done this for geometry in his celebrated
book Foundations of geometry. In subsequent years, he and a num-
ber of his students and collaborators worked on other areas of
mathematics to do what Hilbert had done for geometry. Hilbert
himself gave axiom systems for arithmetic and analysis. Zermelo
gave an axiomatization of set theory, which was expanded on by
Fraenkel, Skolem, von Neumann, and others. By the mid-1920s,
there were two approaches that laid claim to the title of an ax-
iomatization of “all” of mathematics, the Principia mathematica of
Russell and Whitehead, and what came to be known as Zermelo-
Fraenkel set theory.
In 1921, Hilbert set out on a research project to establish the
goal of proving these systems to be consistent. He was aided
in this project by several of his students, in particular Bernays,
Ackermann, and later Gentzen. The basic idea for accomplishing
this goal was to cast the question of the possibility of a derivation
of an inconsistency in mathmatics as a combinatorial problem
about possible sequences of symbols, namely possible sequences
of sentences which meet the criterion of being a correct deriva-
tion of, say, A ∧ ¬A from the axioms of an axiom system for
arithmetic, analysis, or set theory. A proof of the impossibility
of such a sequence of symbols would—since it is itself a math-
ematical proof—be formalizable in these axiomatic systems. In
other words, there would be some sentence Con which states that,
say, arithmetic is consistent. Moreover, this sentence should be
provable in the systems in question, especially if its proof requires
only very restricted, “finitary” means.
6 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

The second aim, that the axiom systems developed would set-
tle every mathematical question, can be made precise in two ways.
In one way, we can formulate it as follows: For any sentence A
in the language of an axiom system for mathematics, either A
or ¬A is provable from the axioms. If this were true, then there
would be no sentences which can neither be proved nor refuted
on the basis of the axioms, no questions which the axioms do not
settle. An axiom system with this property is called complete. Of
course, for any given sentence it might still be a difficult task to
determine which of the two alternatives holds. But in principle
there should be a method to do so. In fact, for the axiom and
derivation systems considered by Hilbert, completeness would
imply that such a method exists—although Hilbert did not real-
ize this. The second way to interpret the question would be this
stronger requirement: that there be a mechanical, computational
method which would determine, for a given sentence A, whether
it is derivable from the axioms or not.
In 1931, Gödel proved the two “incompleteness theorems,”
which showed that this program could not succeed. There is
no axiom system for mathematics which is complete, specifically,
the sentence that expresses the consistency of the axioms is a
sentence which can neither be proved nor refuted.
This struck a lethal blow to Hilbert’s original program. How-
ever, as is so often the case in mathematics, it also opened
up exciting new avenues for research. If there is no one, all-
encompassing formal system of mathematics, it makes sense to
develop more circumscribesd systems and investigate what can
be proved in them. It also makes sense to develop less restricted
methods of proof for establishing the consistency of these sys-
tems, and to find ways to measure how hard it is to prove their
consistency. Since Gödel showed that (almost) every formal sys-
tem has questions it cannot settle, it makes sense to look for
“interesting” questions a given formal system cannot settle, and
to figure out how strong a formal system has to be to settle them.
To the present day, logicians have been pursuing these questions
in a new mathematical discipline, the theory of proofs.
7 1.2. DEFINITIONS

1.2 Definitions
In order to carry out Hilbert’s project of formalizing mathematics
and showing that such a formalization is consistent and complete,
the first order of business would be that of picking a language,
logical framework, and a system of axioms. For our purposes, let
us suppose that mathematics can be formalized in a first-order
language, i.e., that there is some set of constant symbols, func-
tion symbols, and predicate symbols which, together with the
connectives and quatifiers of first-order logic, allow us to express
the claims of mathematics. Most people agree that such a lan-
guage exists: the language of set theory, in which ∈ is the only
non-logical symbol. That such a simple language is so expressive
is of course a very implausible claim at first sight, and it took a
lot of work to establish that practically of all mathematics can be
expressed in this very austere vocabulary. To keep things simple,
for now, let’s restrict our discussion to arithmetic, so the part of
mathematics that just deals with the natural numbers N. The nat-
ural language in which to express facts of arithmetic is LA . LA
contains a single two-place predicate symbol <, a single constant
symbol 0, one one-place function symbol ′, and two two-place
function symbols + and ×.

Definition 1.1. A set of sentences 𝛤 is a theory if it is closed

under entailment, i.e., if 𝛤 = {A : 𝛤 ⊨ A}.

There are two easy ways to specify theories. One is as the

set of sentences true in some structure. For instance, consider
the structure for LA in which the domain is N and all non-logical
symbols are interpreted as you would expect.

Definition 1.2. The standard model of arithmetic is the struc-

ture N defined as follows:

1. |N| = N
8 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

2. 0N = 0

3. ′N (n) = n + 1 for all n ∈ N

4. +N (n,m) = n + m for all n,m ∈ N

5. ×N (n,m) = n · m for all n,m ∈ N

6. <N = {⟨n,m⟩ : n ∈ N,m ∈ N,n < m}

Note the difference between × and ·: × is a symbol in the

language of arithmetic. Of course, we’ve chosen it to remind us
of multiplication, but × is not the multiplication operation but
a two-place function symbol (officially, f12 . By contrast, · is the
ordinary multiplication function. When you see something like
n · m, we mean the product of the numbers n and m; when you
see something like x × y we are talking about a term in the lan-
guage of arithmetic. In the standard model, the function symbol
times is interpreted as the function · on the natural numbers. For
addition, we use + as both the function symbol of the language
of arithmetic, and the addition function on the natural numbers.
Here you have to use the context to determine what is meant.

Definition 1.3. The theory of true arithmetic is the set of sen-

tences satisfied in the standard model of arithmetic, i.e.,

TA = {A : N ⊨ A}.

TA is a theory, for whenever TA ⊨ A, A is satisfied in every

structure which satisfies TA. Since M ⊨ TA, M ⊨ A, and so
A ∈ TA.
The other way to specify a theory 𝛤 is as the set of sentences
entailed by some set of sentences 𝛤0 . In that case, 𝛤 is the “clo-
sure” of 𝛤0 under entailment. Specifying a theory this way is only
interesting if 𝛤0 is explicitly specified, e.g., if the elements of 𝛤0
are listed. At the very least, 𝛤0 has to be decidable, i.e., there
has to be a computable test for when a sentence counts as an
9 1.2. DEFINITIONS

element of 𝛤0 or not. We call the sentences in 𝛤0 axioms for 𝛤,

and 𝛤 axiomatized by 𝛤0 .

Definition 1.4. A theory 𝛤 is axiomatized by 𝛤0 iff

𝛤 = {A : 𝛤0 ⊨ A}

Definition 1.5. The theory Q axiomatized by the following sen-

tences is known as “Robinson’s Q ” and is a very simple theory
of arithmetic.

∀x ∀y (x ′ = y ′ → x = y) (Q 1 )
′
∀x 0 ≠ x (Q 2 )
′
∀x (x = 0 ∨ ∃y x = y ) (Q 3 )
∀x (x + 0) = x (Q 4 )
′ ′
∀x ∀y (x + y ) = (x + y) (Q 5 )
∀x (x × 0) = 0 (Q 6 )
′
∀x ∀y (x × y ) = ((x × y) + x) (Q 7 )
′
∀x ∀y (x < y ↔ ∃z (z + x) = y) (Q 8 )

The set of sentences {Q 1 , . . . ,Q 8 } are the axioms of Q , so Q

consists of all sentences entailed by them:

Q = {A : {Q 1 , . . . ,Q 8 } ⊨ A}.

Definition 1.6. Suppose A(x) is a formula in LA with free vari-

ables x and y 1 , . . . , y n . Then any sentence of the form

∀y 1 . . . ∀y n ((A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x))

is an instance of the induction schema.

10 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

Peano arithmetic PA is the theory axiomatized by the axioms

of Q together with all instances of the induction schema.

Every instance of the induction schema is true in N. This is

easiest to see if the formula A only has one free variable x. Then
A(x) defines a subset XA of N in N. XA is the set of all n ∈ N such
that N,s ⊨ A(x) when s (x) = n. The corresponding instance of
the induction schema is

((A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x)).

If its antecedent is true in N, then 0 ∈ XA and, whenever n ∈ XA ,

so is n + 1. Since 0 ∈ XA , we get 1 ∈ XA . With 1 ∈ XA we get
2 ∈ XA . And so on. So for every n ∈ N, n ∈ XA . But this means
that ∀x A(x) is satisfied in N.
Both Q and PA are axiomatized theories. The big question
is, how strong are they? For instance, can PA prove all the truths
about N that can be expressed in LA ? Specifically, do the axioms
of PA settle all the questions that can be formulated in LA ?
Another way to put this is to ask: Is PA = TA? TA obviously
does prove (i.e., it includes) all the truths about N, and it settles
all the questions that can be formulated in LA , since if A is a sen-
tence in LA , then either N ⊨ A or N ⊨ ¬A, and so either TA ⊨ A
or TA ⊨ ¬A. Call such a theory complete.

Definition 1.7. A theory 𝛤 is complete iff for every sentence A in

its language, either 𝛤 ⊨ A or 𝛤 ⊨ ¬A.

By the Completeness Theorem, 𝛤 ⊨ A iff 𝛤 ⊢ A, so 𝛤 is

complete iff for every sentence A in its language, either 𝛤 ⊢ A or
𝛤 ⊢ ¬A.
Another question we are led to ask is this: Is there a computa-
tional procedure we can use to test if a sentence is in TA, in PA,
or even just in Q ? We can make this more precise by defining
when a set (e.g., a set of sentences) is decidable.
11 1.2. DEFINITIONS

Definition 1.8. A set X is decidable iff there is a computational

procedure which on input x returns 1 if x ∈ X and 0 otherwise.

So our question becomes: Is TA (PA, Q ) decidable?

The answer to all these questions will be: no. None of these
theories are decidable. However, this phenomenon is not spe-
cific to these particular theories. In fact, any theory that satisfies
certain conditions is subject to the same results. One of these
conditions, which Q and PA satisfy, is that they are axiomatized
by a decidable set of axioms.

Definition 1.9. A theory is axiomatizable if it is axiomatized by

a decidable set of axioms.

Example 1.10. Any theory axiomatized by a finite set of sen-

tences is axiomatizable, since any finite set is decidable. Thus,
Q , for instance, is axiomatizable.
Schematically axiomatized theories like PA are also axiomati-
zable. For to test if B is among the axioms of PA, i.e., to compute
the function 𝜒 X where 𝜒 X (B) = 1 if B is an axiom of PA and
= 0 otherwise, we can do the following: First, check if B is one
of the axioms of Q . If it is, the answer is “yes” and the value of
𝜒 X (B) = 1. If not, test if it is an instance of the induction schema.
This can be done systematically; in this case, perhaps it’s easiest
to see that it can be done as follows: Any instance of the induc-
tion schema begins with a number of universal quantifiers, and
then a sub-formula that is a conditional. The consequent of that
conditional is ∀x A(x, y 1 , . . . , y n ) where x and y 1 , . . . , y n are all
the free variables of A and the initial quantifiers of B bind the
variables y 1 , . . . , y n . Once we have extracted this A and checked
that its free variables match the variables bound by the univer-
sal qauntifiers at the front and ∀x, we go on to check that the
antecedent of the conditional matches

A(0, y 1 , . . . , y n ) ∧ ∀x (A(x, y 1 , . . . , y n ) → A(x ′, y 1 , . . . , y n ))

12 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

Again, if it does, B is an instance of the induction schema, and

if it doesn’t, B isn’t.

In answering this question—and the more general question

of which theories are complete or decidable—it will be useful
to consider also the following definition. Recall that a set X is
countable iff it is empty or if there is a surjective function f : N →
X . Such a function is called an enumeration of X .

Definition 1.11. A set X is called computably enumerable (c.e. for

short) iff it is empty or it has a computable enumeration.

In addition to axiomatizability, another condition on theories

to which the incompleteness theorems apply will be that they are
strong enough to prove basic facts about computable functions
and decidable relations. By “basic facts,” we mean sentences
which express what the values of computable functions are for
each of their arguments. And by “strong enough” we mean that
the theories in question count these sentences among its theo-
rems. For instance, consider a prototypical computable function:
addition. The value of + for arguments 2 and 3 is 5, i.e., 2 + 3 = 5.
A sentence in the language of arithmetic that expresses that the
value of + for arguments 2 and 3 is 5 is: (2 + 3) = 5. And, e.g.,
Q proves this sentence. More generally, we would like there to
be, for each computable function f (x 1 ,x 2 ) a formula A f (x 1 ,x 2 , y)
in LA such that Q ⊢ A f (n 1 ,n 2 ,m) whenever f (n1 ,n 2 ) = m. In this
way, Q proves that the value of f for arguments n 1 , n 2 is m. In
fact, we require that it proves a bit more, namely that no other
number is the value of f for arguments n 1 , n2 . And the same goes
for decidable relations. This is made precise in the following two
definitions.
13 1.2. DEFINITIONS

Definition 1.12. A formula A(x 1 , . . . ,x k , y) represents the func-

tion f : Nk → N in 𝛤 iff whenever f (n 1 , . . . ,nk ) = m, then

1. 𝛤 ⊢ A(n 1 , . . . ,nk ,m), and

2. 𝛤 ⊢ ∀y (A(n 1 , . . . ,nk , y) → y = m).

Definition 1.13. A formula A(x 1 , . . . ,x k ) represents the relation

R ⊆ Nk iff,

1. whenever R (n 1 , . . . ,nk ), 𝛤 ⊢ A(n 1 , . . . ,nk ), and

2. whenever not R (n 1 , . . . ,nk ), 𝛤 ⊢ ¬A(n 1 , . . . ,nk ).

A theory is “strong enough” for the incompleteness theorems

to apply if it represents all computable functions and all decid-
able relations. Q and its extensions satisfy this condition, but it
will take us a while to establish this—it’s a non-trivial fact about
the kinds of things Q can prove, and it’s hard to show because
Q has only a few axioms from which we’ll have to prove all these
facts. However, Q is a very weak theory. So although it’s hard to
prove that Q represents all computable functions, most interest-
ing theories are stronger than Q , i.e., prove more than Q does.
And if Q proves something, any stronger theory does; since Q
represents all computable functions, every stronger theory does.
This means that many interesting theories meet this condition
of the incompleteness theorems. So our hard work will pay off,
since it shows that the incompletess theorems apply to a wide
range of theories. Certainly, any theory aiming to formalize “all
of mathematics” must prove everything that Q proves, since it
should at the very least be able to capture the results of elemen-
tary computations. So any theory that is a candidate for a theory
of “all of mathematics” will be one to which the incompleteness
theorems apply.
14 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

1.3 Overview of Incompleteness Results

Hilbert expected that mathematics could be formalized in an ax-
iomatizable theory which it would be possible to prove complete
and decidable. Moreover, he aimed to prove the consistency of
this theory with very weak, “finitary,” means, which would de-
fend classical mathematics agianst the challenges of intuitionism.
Gödel’s incompleteness theorems showed that these goals cannot
be achieved.
Gödel’s first incompleteness theorem showed that a version of
Russell and Whitehead’s Principia Mathematica is not complete.
But the proof was actually very general and applies to a wide
variety of theories. This means that it wasn’t just that Principia
Mathematica did not manage to completely capture mathematics,
but that no acceptable theory does. It took a while to isolate the
features of theories that suffice for the incompleteness theorems
to apply, and to generalize Gödel’s proof to apply make it depend
only on these features. But we are now in a position to state a very
general version of the first incompleteness theorem for theories
in the language LA of arithmetic.

Theorem 1.14. If 𝛤 is a consistent and axiomatizable theory in LA

which represents all computable functions and decidable relations, then
𝛤 is not complete.

To say that 𝛤 is not complete is to say that for at least one sen-
tence A, 𝛤 ⊬ A and 𝛤 ⊬ ¬A. Such a sentence is called independent
(of 𝛤). We can in fact relatively quickly prove that there must
be independent sentences. But the power of Gödel’s proof of the
theorem lies in the fact that it exhibits a specific example of such
an independent sentence. The intriguing construction produces
a sentence G 𝛤 , called a Gödel sentence for 𝛤, which is unprovable
because in 𝛤, G 𝛤 is equivalent to the claim that G 𝛤 is unprovable
in 𝛤. It does so constructively, i.e., given an axiomatization of 𝛤
and a description of the proof system, the proof gives a method
for actually writing down G 𝛤 .
15 1.3. OVERVIEW OF INCOMPLETENESS RESULTS

The construction in Gödel’s proof requires that we find a way

to express in LA the properties of and operations on terms and
formulas of LA itself. These include properties such as “A is
a sentence,” “𝛿 is a derivation of A,” and operations such as
A[t /x]. This way must (a) express these properties and relations
via a “coding” of symbols and sequences thereof (which is what
terms, formulas, derivations, etc. are) as natural numbers (which
is what LA can talk about). It must (b) do this in such a way that
𝛤 will prove the relevant facts, so we must show that these prop-
erties are coded by decidable properties of natural numbers and
the operations correspond to computable functions on natural
numbers. This is called “arithmetization of syntax.”
Before we investigate how syntax can be arithmetized, how-
ever, we will consider the condition that 𝛤 is “strong enough,”
i.e., represents all computable functions and decidable relations.
This requires that we give a precise definition of “computable.”
This can be done in a number of ways, e.g., via the model of
Turing machines, or as those functions computable by programs
in some general-purpose programming language. Since our aim
is to represent these functions and relations in a theory in the
language LA , however, it is best to pick a simple definition of
computability of just numerical functions. This is the notion of
recursive function. So we will first discuss the recursive functions.
We will then show that Q already represents all recursive func-
tions and relations. This will allow us to apply the incompleteness
theorem to specific theories such as Q and PA, since we will have
established that these are examples of theories that are “strong
enough.”
The end result of the arithmetization of syntax is a formula
Prov𝛤 (x) which, via the coding of formulas as numbers, expresses
provability from the axioms of 𝛤. Specifically, if A is coded by
the number n, and 𝛤 ⊢ A, then 𝛤 ⊢ Prov𝛤 (n). This “provability
predicate” for 𝛤 allows us also to express, in a certain sense, the
consistency of 𝛤 as a sentence of LA : let the “consistency state-
ment” for 𝛤 be the sentence ¬Prov𝛤 (n), where we take n to be the
code of a contradiction, e.g., of ⊥. The second incompleteness
16 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

theorem states that consistent axiomatizable theories also do not

prove their own consistency statements. The conditions required
for this theorem to apply are a bit more stringent than just that
the theory represents all computable functions and decidable re-
lations, but we will show that PA satisifes them.

1.4 Undecidability and Incompleteness

Gödel’s proof of the incompleteness theorems require arithmeti-
zation of syntax. But even without that we can obtain some nice
results just on the assumtion that a theory represents all decid-
able relations. The proof is a diagonal argument similar to the
proof of the undecidability of the halting problem.

Theorem 1.15. If 𝛤 is a consistent theory that represents every decid-

able relation, then 𝛤 is not decidable.

Proof. Suppose 𝛤 were decidable. We show that if 𝛤 represents

every decidable relation, it must be inconsistent.
Decidable properties (one-place relations) are represented by
formulas with one free variable. Let A0 (x), A1 (x), . . . , be a com-
putable enumeration of all such formulas. Now consider the fol-
lowing set D ⊆ N:

D = {n : 𝛤 ⊢ ¬An (n)}

The set D is decidable, since we can test if n ∈ D by first com-

puting An (x), and from this ¬An (n). Obviously, substituting the
term n for every free occurrence of x in An (x) and prefixing A(n)
by ¬ is a mechanical matter. By assumption, 𝛤 is decidable, so
we can test if ¬A(n) ∈ 𝛤. If it is, n ∈ D, and if it isn’t, n ∉ D. So
D is likewise decidable.
Since 𝛤 represents all decidable properties, it represents D.
And the formulas which represent D in 𝛤 are all among A0 (x),
A1 (x), . . . . So let d be a number such that Ad (x) represents D
in 𝛤. If d ∉ D, then, since Ad (x) represents D, 𝛤 ⊢ ¬Ad (d ).
17 1.4. UNDECIDABILITY AND INCOMPLETENESS

But that means that d meets the defining condition of D, and so

d ∈ D. This contradicts d ∉ D. So by indirect proof, d ∈ D.
Since d ∈ D, by the definition of D, 𝛤 ⊢ ¬Ad (d ). On the
other hand, since Ad (x) represents D in 𝛤, 𝛤 ⊢ Ad (d ). Hence, 𝛤
is inconsistent. □

The preceding theorem shows that no theory that represents

all decidable relations can be decidable. We will show that Q
does represent all decidable relations; this means that all theories
that include Q , such as PA and TA, also do, and hence also are
not decidable.
We can also use this result to obtain a weak version of the first
incompleteness theorem. Any theory that is axiomatizable and
complete is decidable. Consistent theories that are axiomatizable
and represent all decidable properties then cannot be complete.

Theorem 1.16. If 𝛤 is axiomatizable and complete it is decidable.

Proof. Any inconsistent theory is decidable, since inconsistent

theories contain all sentences, so the answer to the question “is
A ∈ 𝛤” is always “yes,” i.e., can be decided.
So suppose 𝛤 is consistent, and furthermore is axiomatizable,
and complete. Since 𝛤 is axiomatizable, it is computably enumer-
able. For we can enumerate all the correct derivations from the
axioms of 𝛤 by a computable function. From a correct derivation
we can compute the sentence it derives, and so together there is
a computable function that enumerates all theorems of 𝛤. A
sentence is a theorem of 𝛤 iff ¬A is not a theorem, since 𝛤 is
consistent and complete. We can therefore decide if A ∈ 𝛤 as
follows. Enumerate all theorems of 𝛤. When A appears on this
list, we know that 𝛤 ⊢ A. When ¬A appears on this list, we know
that 𝛤 ⊬ A. Since 𝛤 is complete, one of these cases eventually
obtains, so the procedure eventually produces and answer. □
18 CHAPTER 1. INTRODUCTION TO INCOMPLETENESS

Corollary 1.17. If 𝛤 is consistent, axiomatizable, and represents ev-

ery decidable property, it is not complete.

Proof. If 𝛤 were complete, it would be decidable by the previous

theorem (since it is axiomatizable and consistent). But since 𝛤
represents every decidable property, it is not decidable, by the
first theorem. □

Once we have established that, e.g., Q , represents all decid-

able properties, the corollary tells us that Q must be incomplete.
However, its proof does not provide an example of an indepen-
dent sentence; it merely shows that such a sentence must exist.
For this, we have to arithmetize syntax and follow Gödel’s origi-
nal proof idea. And of course, we still have to show the first claim,
namely that Q does, in fact, represent all decidable properties.
It should be noted that not every interesting theory is incom-
plete or undecidable. There are many theories that are suf-
ficiently strong to describe interesting mathematical facts that
do not satisify the conditions of Gödel’s result. For instance,
Pres = {A ∈ LA+ : N ⊨ A}, the set of sentences of the language
of arithmetic without × true in the standard model, is both com-
plete and decidable. This theory is called Presburger arithmetic,
and proves all the truths about natural numbers that can be for-
mulated just with 0, ′, and +.

Summary
Hilbert’s program aimed to show that all of mathematics could be
formalized in an axiomatized theory in a formal language, such
as the language of arithmetic or of set theory. He believed that
such a theory would be complete. That is, for every sentence A,
either T ⊢ A or T ⊢ ¬A. In this sense then, T would have settled
every mathematical question: it would either prove that it’s true
or that it’s false. If Hilbert had been right, it would also have
turned out that mathematics is decidable. That’s because any
axiomatizable theory is computably enumerable, i.e., there is
19 1.4. UNDECIDABILITY AND INCOMPLETENESS

a computable function that lists all its theorems. We can test if

a sentence A is a theorem by listing all of them until we find A
(in which it is a theorem) or ¬A (in which case it isn’t). Alas,
Hilbert was wrong. Gödel proved that no axiomatizable, consis-
tent theory that is “strong enough” is complete. That’s the first
incompleteness theorem. The requirement that the theory be
“strong enough” amounts to it representing all computable func-
tions and relations. Specifically, the very weak theory Q satisfies
this property, and any theory that is at least as strong as Q also
does. He also showed—that is the second incompleteness the-
orem—that the sentence that expresses the consistency of the
theory is itself undecidable in it, i.e., the theory proves neither
it nor its negation. So Hilbert’s further aim of finding “finitary”
consistency proof of all of mathematics cannot be realized. For
any finitary consistency proof would, presumably, be formalizable
in a theory that captures all of mathematics. Finally, we estab-
lished that theories that represent all computable functions and
relations are not decidable. Note that although axomatizabil-
ity and completeness implies decidability, incompleteness does
not imply undecidability. So this result shows that the second of
Hilbert’s goals, namely that there be a procedure that decides if
T ⊢ A or not, can also not be achieved, at least not for theories
at least as strong as Q .

Problems
Problem 1.1. Show that TA = {A : N ⊨ A} is not axiomatizable.
You may assume that TA represents all decidable properties.
CHAPTER 2

Recursive
Functions
2.1 Introduction
In order to develop a mathematical theory of computability, one
has to, first of all, develop a model of computability. We now
think of computability as the kind of thing that computers do,
and computers work with symbols. But at the beginning of the
development of theories of computability, the paradigmatic ex-
ample of computation was numerical computation. Mathemati-
cians were always interested in number-theoretic functions, i.e.,
functions f : Nn → N that can be computed. So it is not surpris-
ing that at the beginning of the theory of computability, it was
such functions that were studied. The most familiar examples
of computable numerical functions, such as addition, multipli-
cation, exponentiation (of natural numbers) share an interesting
feature: they can be defined recursively. It is thus quite natural
to attempt a general definition of computable function on the basis
of recursive definitions. Among the many possible ways to de-
fine number-theoretic functions recursively, one particulalry sim-
ple pattern of definition here becomes central: so-called primitive
recursion.
In addition to computable functions, we might be interested

20
21 2.2. PRIMITIVE RECURSION

in computable sets and relations. A set is computable if we can

compute the answer to whether or not a given number is an ele-
ment of the set, and a relation is computable iff we can compute
whether or not a tuple ⟨n 1 , . . . ,nk ⟩ is an element of the relation.
By considering the characteristic function of a set or relation, dis-
cussion of computable sets and relations can be subsumed under
that of computable functions. Thus we can define primitive re-
cursive relations as well, e.g., the relation “n evenly divides m” is
a primitive recursive relation.
Primitive recursive functions—those that can be defined using
just primitive recursion—are not, however, the only computable
number-theoretic functions. Many generalizations of primitive re-
cursion have been considered, but the most powerful and widely-
accepted additional way of computing functions is by unbounded
search. This leads to the definition of partial recursive functions,
and a related definition to general recursive functions. General re-
cursive functions are computable and total, and the definition
characterizes exactly the partial recursive functions that happen
to be total. Recursive functions can simulate every other model
of computation (Turing machines, lambda calculus, etc.) and so
represent one of the many accepted models of computation.

2.2 Primitive Recursion

A characteristic of the natural numbers is that every natural num-
ber can be reached from 0 by applying the successor operation +1
finitely many times—any natural number is either 0 or the suc-
cessor of . . . the successor of 0. One way to specify a func-
tion f : N → N that makes use of this fact is this: (a) specify
what the value of f is for argument 0, and (b) also specify how
to, given the value of f (x), compute the value of f (x + 1). For (a)
tells us directly what f (0) is, so f is defined for 0. Now, using the
instruction given by (b) for x = 0, we can compute f (1) = f (0+1)
from f (0). Using the same instructions for x = 1, we compute
f (2) = f (1 + 1) from f (1), and so on. For every natural num-
22 CHAPTER 2. RECURSIVE FUNCTIONS

ber x, we’ll eventually reach the step where we define f (x) from
f (x + 1), and so f (x) is defined for all x ∈ N.
For instance, suppose we specify h : N → N by the following
two equations:

h (0) = 1
h (x + 1) = 2 · h (x)

If we already know how to multiply, then these equations give us

the information required for (a) and (b) above. Successively the
second equation, we get that

h (1) = 2 · h (0) = 2,
h (2) = 2 · h (1) = 2 · 2,
h (3) = 2 · h (2) = 2 · 2 · 2,
..
.

We see that the function h we have specified is h (x) = 2x .

The characteristic feature of the natural numbers guarantees
that there is only one function d that meets these two criteria.
A pair of equations like these is called a definition by primitive
recursion of the function d . It is so-called because we define f
“recursively,” i.e., the definition, specifically the second equation,
involves f itself on the right-hand-side. It is “primitive” because in
defining f (x +1) we only use the value f (x), i.e., the immediately
preceding value. This is the simplest way of defining a function
on N recursively.
We can define even more fundamental functions like addi-
tion and multiplication by primitive recursion. In these cases,
however, the functions in question are 2-place. We fix one of the
argument places, and use the other for the recursion. E.g, to de-
fine add(x, y) we can fix x and define the value first for y = 0 and
then for y + 1 in terms of y. Since x is fixed, it will appear on the
left and on the right side of the defining equations.

add(x, 0) = x
23 2.2. PRIMITIVE RECURSION

add(x, y + 1) = add(x, y) + 1

These equations specify the value of add for all x and y. To find
add(2, 3), for instance, we apply the defining equations for x = 2,
using the first to find add(2, 0) = 2, then using the second to
successively find add(2, 1) = 2 + 1 = 3, add(2, 2) = 3 + 1 = 4,
add(2, 3) = 4 + 1 = 5.
In the definition of add we used + on the right-hand-side of the
second equation, but only to add 1. In other words, we used the
successor function succ(z ) = z + 1 and applied it to the previous
value add(x, y) to define add(x, y + 1). So we can think of the
recursive definition as given in terms of a single function which
we apply to the previous value. However, it doesn’t hurt—and
sometimes is necessary—to allow the function to depend not just
on the previous value but also on x and y. Consider:

mult(x, 0) = 0
mult(x, y + 1) = add(mult(x, y),x)

This is a primitive recursive definition of a function mult

by applying the function add to both the preceding value
mult(x, y) and the first argument x. It also defines the func-
tion mult(x, y) for all arguments x and y. For instance, mult(2, 3)
is determined by successively computing mult(2, 0), mult(2, 1),
mult(2, 2), and mult(2, 3):

mult(2, 0) = 0
mult(2, 1) = mult(2, 0 + 1) = add(mult(2, 0), 2) = add(0, 2) = 2
mult(2, 2) = mult(2, 1 + 1) = add(mult(2, 1), 2) = add(2, 2) = 4
mult(2, 3) = mult(2, 2 + 1) = add(mult(2, 2), 2) = add(4, 2) = 6

The general pattern then is this: to give a primitive recursive

definition of a function h (x 0 , . . . ,x k −1 , y), we provide two equa-
tions. The first defines the value of h (x 0 , . . . ,x k −1 , 0) without ref-
erence to f . The second defines the value of h (x 0 , . . . ,x k −1 , y + 1)
in terms of h (x 0 , . . . ,x k −1 , y), the other arguments x 0 , . . . , x k −1 ,
24 CHAPTER 2. RECURSIVE FUNCTIONS

and y. Only the immediately preceding value of h may be used

in that second equation. If we think of the operations given by
the right-hand-sides of these two equations as themselves being
functions f and g , then the pattern to define a new function h by
primitive recursion is this:

h (x 0 , . . . ,x k −1 , 0) = f (x 0 , . . . ,x k −1 )
h (x 0 , . . . ,x k −1 , y + 1) = g (x 0 , . . . ,x k −1 , y,h (x 0 , . . . ,x k −1 , y))

In the case of add, we have k = 0 and f (x 0 ) = x 0 (the identity

function), and g (x 0 , y, z ) = z + 1 (the 3-place function that returns
the successor of its third argument):

add(x 0 , 0) = f (x 0 ) = x 0
add(x 0 , y + 1) = g (x 0 , y, add(x 0 , y)) = succ(add(x 0 , y))

In the case of mult, we have f (x 0 ) = 0 (the constant function

always returning 0) and g (x 0 , y, z ) = add(z ,x 0 ) (the 3-place func-
tion that returns the sum of its last and first argument):

mult(x 0 , 0) = f (x 0 ) = 0
mult(x 0 , y + 1) = g (x 0 , y, mult(x 0 , y)) = add(mult(x 0 , y),x 0 )

2.3 Composition
If f and g are two one-place functions of natural numbers, we
can compose them: h (x) = g ( f (x)). The new function h (x) is
then defined by composition from the functions f and g . We’d like
to generalize this to functions of more than one argument.
Here’s one way of doing this: suppose f is a k -place function,
and g 0 , . . . , g k −1 are k functions which are all n-place. Then we
can define a new n-place function h as follows:

h (x 0 , . . . ,x n−1 ) = f (g 0 (x 0 , . . . ,x n−1 ), . . . , g k −1 (x 0 , . . . ,x n−1 ))

If f and all g i are computable, so is h: To compute h (x 0 , . . . ,x n−1 ),

first compute the values y i = g i (x 0 , . . . ,x n−1 ) for each i =
25 2.3. COMPOSITION

0, . . . , k − 1. Then feed these values into f to compute

h (x 0 , . . . ,x k −1 ) = f (y 0 , . . . , y k −1 ).
This may seem like an overly restrictive characterization of
what happens when we compute a new function using some ex-
isting ones. For one thing, sometimes we do not use all the argu-
ments of a function, as when we defined g (x, y, z ) = succ(z ) for
use in the primitive recursive definition of add. Suppose we are
allowed use of the following functions:

Pin (x 0 , . . . ,x n−1 ) = x i

The functions Pik are called projection functions: Pin is an n-place

function. Then g can be defined by

g (x, y, z ) = succ(P 23 ).

Here the role of f is played by the 1-place function succ, so k = 1.

And we have one 3-place function P 23 which plays the role of g 0 .
The result is a 3-place function that returns the successor of the
third argument.
The projection functions also allow us to define new func-
tions by reordering or identifying arguments. For instance, the
function h (x) = add(x,x) can be defined by

h (x 0 ) = add(P 01 (x 0 ),P 01 (x 0 )).

Here k = 2, n = 1, the role of f (y 0 , y 1 ) is played by add, and the

roles of g 0 (x 0 ) and g 1 (x 0 ) are both played by P 01 (x 0 ), the one-place
projection function (aka the identity function).
If f (y 0 , y 1 ) is a function we already have, we can define the
function h (x 0 ,x 1 ) = f (x 1 ,x 0 ) by

h (x 0 ,x 1 ) = f (P 12 (x 0 ,x 1 ),P 02 (x 0 ,x 1 )).

Here k = 2, n = 2, and the roles of g 0 and g 1 are played by P 12

and P 02 , respectively.
You may also worry that g 0 , . . . , g k −1 are all required to have
the same arity n. (Remember that the arity of a function is the
26 CHAPTER 2. RECURSIVE FUNCTIONS

number of arguments; an n-place function has arity n.) But

adding the projection functions provides the desired flexibility.
For example, suppose f and g are 3-place functions and h is the
2-place function defined by

h (x, y) = f (x, g (x,x, y), y).

The definition of h can be rewritten with the projection functions,

as

h (x, y) = f (P 02 (x, y), g (P 02 (x, y),P 02 (x, y),P 12 (x, y)),P 12 (x, y)).

Then h is the composition of f with P 02 , l , and P 12 , where

l (x, y) = g (P 02 (x, y),P 02 (x, y),P 12 (x, y)),

i.e., l is the composition of g with P 02 , P 02 , and P 12 .

2.4 Primitive Recursion Functions

Let us record again how we can define new functions from exist-
ing ones using primitive recursion and composition.

Definition 2.1. Suppose f is a k -place function (k ≥ 1) and

g is a (k + 2)-place function. The function defined by primitive
recursion from f and g is the (k + 1)-place function h defined by
the equations

h (x 0 , . . . ,x k −1 , 0) = f (x 0 , . . . ,x k −1 )
h (x 0 , . . . ,x k −1 , y + 1) = g (x 0 , . . . ,x k −1 , y,h (x 0 , . . . ,x k −1 , y))

Definition 2.2. Suppose f is a k -place function, and g 0 , . . . , g k −1

are k functions which are all n-place. The function defined by
composition from f and g 0 , . . . , g k −1 is the n-place function h defined
27 2.4. PRIMITIVE RECURSION FUNCTIONS

by

h (x 0 , . . . ,x n−1 ) = f (g 0 (x 0 , . . . ,x n−1 ), . . . , g k −1 (x 0 , . . . ,x n−1 )).

In addition to succ and the projection functions

Pin (x 0 , . . . ,x n−1 ) = x i ,

for each natural number n and i < n, we will include among the
primitive recursive functions the function zero(x) = 0.

Definition 2.3. The set of primitive recursive functions is the set

of functions from Nn to N, defined inductively by the following
clauses:

1. zero is primitive recursive.

2. succ is primitive recursive.

3. Each projection function Pin is primitive recursive.

4. If f is a k -place primitive recursive function and g 0 ,

. . . , g k −1 are n-place primitive recursive functions, then the
composition of f with g 0 , . . . , g k −1 is primitive recursive.

5. If f is a k -place primitive recursive function and g is a

k + 2-place primitive recursive function, then the function
defined by primitive recursion from f and g is primitive
recursive.

Put more concisely, the set of primitive recursive functions is

the smallest set containing zero, succ, and the projection func-
tions P jn , and which is closed under composition and primitive
recursion.
Another way of describing the set of primitive recursive func-
tions is by defining it in terms of “stages.” Let S 0 denote the set of
starting functions: zero, succ, and the projections. These are the
primitive recursive functions of stage 0. Once a stage S i has been
28 CHAPTER 2. RECURSIVE FUNCTIONS

defined, let S i +1 be the set of all functions you get by applying a

single instance of composition or primitive recursion to functions
already in S i . Then ⋃︂
S = Si
i ∈N

is the set of all primitive recursive functions

Let us verify that add is a primitive recursive function.

Proposition 2.4. The addition function add(x, y) = x + y is primi-

tive recursive.

Proof. We already have a primitive recursive definition of add

in terms of two functions f and g which matches the format of
Definition 2.1:

add(x 0 , 0) = f (x 0 ) = x 0
add(x 0 , y + 1) = g (x 0 , y, add(x 0 , y)) = succ(add(x 0 , y))

So add is primitive recursive provided f and g are as well.

f (x 0 ) = x 0 = P 01 (x 0 ), and the projection functions count as prim-
itive recursive, so f is primitive recursive. The function g is the
three-place function g (x 0 , y, z ) defined by

g (x 0 , y, z ) = succ(z ).

This does not yet tell us that g is primitive recursive, since g and
succ are not quite the same function: succ is one-place, and g has
to be three-place. But we can define g “officially” by composition
as
g (x 0 , y, z ) = succ(P 23 (x 0 , y, z ))
Since succ and P 23 count as primitive recursive functions, g does
as well, since it can be defined by composition from primitive
recursive functions. □
29 2.4. PRIMITIVE RECURSION FUNCTIONS

Proposition 2.5. The multiplication function mult(x, y) = x · y is

primitive recursive.

Proof. Exercise. □

Example 2.6. Here’s our very first example of a primitive recur-

sive definition:

h (0) = 1
h (y + 1) = 2 · h (y).

This function cannot fit into the form required by Definition 2.1,
since k = 0. The definition also involves the constants 1 and 2. To
get around the first problem, let’s introduce a dummy argument
and define the function h ′:

h ′ (x 0 , 0) = f (x 0 ) = 1
h ′ (x 0 , y + 1) = g (x 0 , y,h ′ (x 0 , y)) = 2 · h ′ (x 0 , y).

The function f (x 0 ) = 1 can be defined from succ and zero by

composition: f (x 0 ) = succ(zero(x 0 )). The function g can be
defined by composition from g ′ (z ) = 2 · z and projections:

g (x 0 , y, z ) = g ′ (P 23 (x 0 , y, z ))

and g ′ in turn can be defined by composition as

g ′ (z ) = mult(g ′′ (z ),P 01 (z ))

and

g ′′ (z ) = succ( f (z )),

where f is as above: f (z ) = succ(zero(z )). Now that we have

h ′ we can use composition again to let h (y) = h ′ (P 01 (y),P 01 (y)).
This shows that h can be defined from the basic functions us-
ing a sequence of compositions and primitive recursions, so h is
primitive recursive.
30 CHAPTER 2. RECURSIVE FUNCTIONS

2.5 Primitive Recursion Notations

One advantage to having the precise inductive description of the
primitive recursive functions is that we can be systematic in de-
scribing them. For example, we can assign a “notation” to each
such function, as follows. Use symbols zero, succ, and Pin for
zero, successor, and the projections. Now suppose f is defined
by composition from a k -place function h and n-place functions
g 0 , . . . , g k −1 , and we have assigned notations H , G 0 , . . . , G k −1
to the latter functions. Then, using a new symbol Compk ,n , we
can denote the function f by Compk ,n [H ,G 0 , . . . ,G k −1 ]. For the
functions defined by primitive recursion, we can use analogous
notations of the form Reck [G ,H ], where k + 1 is the arity of the
function being defined. With this setup, we can denote the addi-
tion function by
Rec2 [P 01 , Comp1,3 [succ,P 23 ]].
Having these notations sometimes proves useful.

2.6 Primitive Recursive Functions are

Computable
Suppose a function h is defined by primitive recursion
h (x⃗ , 0) = f (x⃗ )
h (x⃗ , y + 1) = g (x⃗ , y,h (x⃗ , y))
and suppose the functions f and g are computable. (We use x⃗
to abbreviate x 0 , . . . , x k −1 .) Then h (x⃗ , 0) can obviously be com-
puted, since it is just f (x⃗ ) which we assume is computable. h (x⃗ , 1)
can then also be computed, since 1 = 0 + 1 and so h (x⃗ , 1) is just
h (x⃗ , 1) = g (x⃗ , 0,h (x⃗ , 0)) = g (x⃗ , 0, f (x⃗ )).

We can go on in this way and compute

h (x⃗ , 2) = g (x⃗ , 1,h (x⃗ , 1)) = g (x⃗ , 1, g (x⃗ , 0, f (x⃗ )))

31 2.7. EXAMPLES OF PRIMITIVE RECURSIVE FUNCTIONS

h (x⃗ , 3) = g (x⃗ , 2,h (x⃗ , 2)) = g (x⃗ , 2, g (x⃗ , 1, g (x⃗ , 0, f (x⃗ ))))
h (x⃗ , 4) = g (x⃗ , 3,h (x⃗ , 3)) = g (x⃗ , 3, g (x⃗ , 2, g (x⃗ , 1, g (x⃗ , 0, f (x⃗ )))))
..
.

Thus, to compute h (x⃗ , y) in general, successively compute h (x⃗ , 0),

h (x⃗ , 1), . . . , until we reach h (x⃗ , y).
Thus, a primitive recursive definition yields a new com-
putable function if the functions f and g are computable. Com-
position of functions also results in a computable function if the
functions f and g i are computable.
Since the basic functions zero, succ, and Pin are computable,
and composition and primitive recursion yield computable func-
tions from computable functions, this means that every primitive
recursive function is computable.

2.7 Examples of Primitive Recursive

Functions
We already have some examples of primitive recursive functions:
the addition and multiplication functions add and mult. The
identity function id(x) = x is primitive recursive, since it is
just P 01 . The constant functions constn (x) = n are primitive recur-
sive since they can be defined from zero and succ by successive
composition. This is useful when we want to use constants in
primitive recursive definitions, e.g., if we want to define the func-
tion f (x) = 2 · x can obtain it by composition from constn (x) and
multiplication as f (x) = mult(const2 (x),P 01 (x)). We’ll make use
of this trick from now on.

Proposition 2.7. The exponentiation function exp(x, y) = x y is

primitive recursive.

Proof. We can define exp primitive recursively as

exp(x, 0) = 1
32 CHAPTER 2. RECURSIVE FUNCTIONS

exp(x, y + 1) = mult(x, exp(x, y)).

Strictly speaking, this is not a recursive definition from primitive
recursive functions. Officially, though, we have:
exp(x, 0) = f (x)
exp(x, y + 1) = g (x, y, exp(x, y)).
where
f (x) = succ(zero(x)) = 1
g (x, y, z ) = mult(P 03 (x, y, z ),P 23 (x, y, z )) = x · z
and so f and g are defined from primitive recursive functions by
composition. □

Proposition 2.8. The predecessor function pred(y) defined by

{︄
0 if y = 0
pred(y) =
y − 1 otherwise

is primitive recursive.

Proof. Note that

pred(0) = 0 and
pred(y + 1) = y .
This is almost a primitive recursive definition. It does not, strictly
speaking, fit into the pattern of definition by primitive recursion,
since that pattern requires at least one extra argument x. It is
also odd in that it does not actually use pred(y) in the definition
of pred(y + 1). But we can first define pred′ (x, y) by
pred′ (x, 0) = zero(x) = 0,
pred′ (x, y + 1) = P 13 (x, y, pred′ (x, y)) = y .
and then define pred from it by composition, e.g., as pred(x) =
pred′ (zero(x),P 01 (x)). □
33 2.7. EXAMPLES OF PRIMITIVE RECURSIVE FUNCTIONS

Proposition 2.9. The factorial function fac(x) = x ! = 1·2·3· · · · ·x

is primitive recursive.

Proof. The obvious primitive recursive definition is

fac(0) = 1
fac(y + 1) = fac(y) · (y + 1).

Officially, we have to first define a two-place function h

h (x, 0) = const1 (x)

h (x, y) = g (x, y,h (x, y))

where g (x, y, z ) = mult(P 23 (x, y, z ), succ(P 13 (x, y, z ))) and then let

fac(y) = h (P 01 (y),P 01 (y))

From now on we’ll be a bit more laissez-faire and not give the
official definitions by composition and primitive recursion. □

Proposition 2.10. Truncated subtraction, x −̇ y, defined by

{︄
0 if x > y
x −̇ y =
x − y otherwise

is primitive recursive.

Proof. We have:

x −̇ 0 = x
x −̇ (y + 1) = pred(x −̇ y) □
34 CHAPTER 2. RECURSIVE FUNCTIONS

|︁ |︁
Proposition 2.11. The distance between x and y, |︁x − y |︁, is primitive
recursive.
|︁ |︁
Proof. We have |︁x − y |︁ = (x −̇ y) + (y −̇ x), so the distance can
be defined by composition from + and −̇, which are primitive
recursive. □

Proposition 2.12. The maximum of x and y, max(x, y), is primitive

recursive.

Proof. We can define max(x, y) by composition from + and −̇ by

max(x, y) = x + (y −̇ x).

If x is the maximum, i.e., x ≥ y, then y −̇ x = 0, so x + (y −̇ x) =

x + 0 = x. If y is the maximum, then y −̇ x = y − x, and so
x + (y −̇ x) = x + (y − x) = y. □

Proposition 2.13. The minimum of x and y, min(x, y), is primitive

recursive.

Proof. Exercise. □

Proposition 2.14. The set of primitive recursive functions is closed

under the following two operations:

1. Finite sums: if f (x⃗ , z ) is primitive recursive, then so is the func-

tion
y
∑︂
g (x⃗ , y) = f (x⃗ , z ).
z =0

2. Finite products: if f (x⃗ , z ) is primitive recursive, then so is the

35 2.8. PRIMITIVE RECURSIVE RELATIONS

function
y
∏︂
h (x⃗ , y) = f (x⃗ , z ).
z =0

Proof. For example, finite sums are defined recursively by the

equations
g (x⃗ , 0) = f (x⃗ , 0)
g (x⃗ , y + 1) = g (x⃗ , y) + f (x⃗ , y + 1). □

2.8 Primitive Recursive Relations

Definition 2.15. A relation R (x⃗ ) is said to be primitive recursive
if its characteristic function,
{︃
1 if R (x⃗ )
𝜒 R (x⃗ ) =
0 otherwise

is primitive recursive.

In other words, when one speaks of a primitive recursive re-

lation R (x⃗ ), one is referring to a relation of the form 𝜒 R (x⃗ ) = 1,
where 𝜒 R is a primitive recursive function which, on any input,
returns either 1 or 0. For example, the relation IsZero(x), which
holds if and only if x = 0, corresponds to the function 𝜒 IsZero ,
defined using primitive recursion by
𝜒 IsZero (0) = 1, 𝜒 IsZero (x + 1) = 0.
It should be clear that one can compose relations with other
primitive recursive functions. So the following are also primitive
recursive:
|︁ |︁
1. The equality relation, x = y, defined by IsZero( |︁x − y |︁)
2. The less-than relation, x ≤ y, defined by IsZero(x −̇ y)
36 CHAPTER 2. RECURSIVE FUNCTIONS

Proposition 2.16. The set of primitive recursive relations is closed

under boolean operations, that is, if P (x⃗ ) and Q (x⃗ ) are primitive re-
cursive, so are

1. ¬P (x⃗ )

2. P (x⃗ ) ∧ Q (x⃗ )

3. P (x⃗ ) ∨ Q (x⃗ )

4. P (x⃗ ) → Q (x⃗ )

Proof. Suppose P (x⃗ ) and Q (x⃗ ) are primitive recursive, i.e., their
characteristic functions 𝜒 P and 𝜒 Q are. We have to show that
the characteristic functions of ¬P (x⃗ ), etc., are also primitive re-
cursive. {︄
0 if 𝜒 P (x⃗ ) = 1
𝜒 ¬P (x⃗ ) =
1 otherwise
We can define 𝜒 ¬P (x⃗ ) as 1 −̇ 𝜒 P (x⃗ ).
{︄
1 if 𝜒 P (x⃗ ) = 𝜒 Q (x⃗ ) = 1
𝜒 P ∧Q (x⃗ ) =
0 otherwise

We can define 𝜒 P ∧Q (x⃗ ) as 𝜒 P (x⃗ )· 𝜒 Q (x⃗ ) or as min( 𝜒 P (x⃗ ), 𝜒 Q (x⃗ )).

Similarly, 𝜒 P ∨Q (x⃗ ) = max( 𝜒 P (x⃗ ), 𝜒 Q (x⃗ )) and 𝜒 P →Q (x⃗ ) =
max(1 −̇ 𝜒 P (x⃗ ), 𝜒 Q (x⃗ )). □

Proposition 2.17. The set of primitive recursive relations is closed un-

der bounded quantification, i.e., if R (x⃗ , z ) is a primitive recursive rela-
tion, then so are the relations (∀z < y) R (x⃗ , z ) and (∃z < y) R (x⃗ , z ).
((∀z < y) R (x⃗ , z ) holds of x⃗ and y if and only if R (x⃗ , z ) holds for
every z less than y, and similarly for (∃z < y) R (x⃗ , z ).)

Proof. By convention, we take (∀z < 0) R (x⃗ , z ) to be true (for

the trivial reason that there are no z less than 0) and (∃z <
0) R (x⃗ , z ) to be false. A universal quantifier functions just like
37 2.8. PRIMITIVE RECURSIVE RELATIONS

a finite product or iterated minimum, i.e., if P (x⃗ , y) ⇔ (∀z <

y) R (x⃗ , z ) then 𝜒 P (x⃗ , y) can be defined by

𝜒 P (x⃗ , 0) = 1
𝜒 P (x⃗ , y + 1) = min( 𝜒 P (x⃗ , y), 𝜒 R (x⃗ , y))).

Bounded existential quantification can similarly be defined using

max. Alternatively, it can be defined from bounded universal
quantification, using the equivalence (∃z < y) R (x⃗ , z ) ↔ ¬(∀z <
y) ¬R (x⃗ , z ). Note that, for example, a bounded quantifier of the
form (∃x ≤ y) . . . x . . . is equivalent to (∃x < y + 1) . . . x . . . . □

Another useful primitive recursive function is the conditional

function, cond(x, y, z ), defined by
{︄
y if x = 0
cond(x, y, z ) =
z otherwise.

This is defined recursively by

cond(0, y, z ) = y, cond(x + 1, y, z ) = z .

One can use this to justify definitions of primitive recursive func-

tions by cases from primitive recursive relations:

Proposition 2.18. If g 0 (x⃗ ), . . . , g m (x⃗ ) are primitive recursive func-

tions, and R 0 (x⃗ ), . . . , R m−1 (x⃗ ) are primitive recursive relations, then
the function f defined by
⎧
⎪
⎪
⎪ g 0 (x⃗ ) if R 0 (x⃗ )
⎪
⎪
⎪
⎪
⎪ g 1 (x⃗ ) if R 1 (x⃗ ) and not R 0 (x⃗ )
⎨.
⎪
f (x⃗ ) = ..
⎪
⎪
g m−1 (x⃗ ) if R m−1 (x⃗ ) and none of the previous hold
⎪
⎪
⎪
⎪
⎪
⎪ g m (x⃗ )
⎪
otherwise
⎩
38 CHAPTER 2. RECURSIVE FUNCTIONS

is also primitive recursive.

Proof. When m = 1, this is just the function defined by

f (x⃗ ) = cond( 𝜒 ¬R0 (x⃗ ), g 0 (x⃗ ), g 1 (x⃗ )).

For m greater than 1, one can just compose definitions of this

form. □

2.9 Bounded Minimization

It is often useful to define a function as the least number sat-
isfying some property or relation P . If P is decidable, we can
compute this function simply by trying out all the possible num-
bers, 0, 1, 2, . . . , until we find the least one satisfying P . This
kind of unbounded search takes us out of the realm of primitive
recursive functions. However, if we’re only interested in the least
number less than some independently given bound, we stay primitive
recursive. In other words, and a bit more generally, suppose we
have a primitive recursive relation R (x, z ). Consider the function
that maps x and y to the least z < y such that R (x, z ). It, too, can
be computed, by testing whether R (x, 0), R (x, 1), . . . , R (x, y − 1).
But why is it primitive recursive?

Proposition 2.19. If R (x⃗ , z ) is primitive recursive, so is the function

mR (x⃗ , y) which returns the least z less than y such that R (x⃗ , z ) holds,
if there is one, and y otherwise. We will write the function mR as

(min z < y) R (x⃗ , z ),

Proof. Note than there can be no z < 0 such that R (x⃗ , z ) since
there is no z < 0 at all. So mR (x⃗ , 0) = 0.
In case the bound is of the form y + 1 we have three cases: (a)
There is a z < y such that R (x⃗ , z ), in which case mR (x⃗ , y + 1) =
mR (x⃗ , y). (b) There is no such z < y but R (x⃗ , y) holds, then
39 2.10. PRIMES

mR (x⃗ , y + 1) = y. (c) There is no z < y + 1 such that R (x⃗ , z ), then

mR (z⃗, y + 1) = y + 1. So,

mR (x⃗ , 0) = 0
⎧ m (x⃗ , y) if mR (x⃗ , y) ≠ y
⎨ R
⎪
⎪
⎪
mR (x⃗ , y + 1) = y if mR (x⃗ , y) = y and R (x⃗ , y)
⎪
⎪y + 1
⎪
otherwise.
⎩
Note that there is a z < y such that R (x⃗ , z ) iff mR (x⃗ , y) ≠ y. □

2.10 Primes
Bounded quantification and bounded minimization provide us
with a good deal of machinery to show that natural functions
and relations are primitive recursive. For example, consider the
relation “x divides y”, written x | y. The relation x | y holds if
division of y by x is possible without remainder, i.e., if y is an
integer multiple of x. (If it doesn’t hold, i.e., the remainder when
dividing x by y is > 0, we write x ∤ y.) In other words, x | y iff for
some z , x · z = y. Obviously, any such z , if it exists, must be ≤ y.
So, we have that x | y iff for some z ≤ y, x · z = y. We can define
the relation x | y by bounded existential quantification from =
and multiplication by

x | y ⇔ (∃z ≤ y) (x · z ) = y .

We’ve thus shown that x | y is primitive recursive.

A natural number x is prime if it is neither 0 nor 1 and is only
divisible by 1 and itself. In other words, prime numbers are such
that, whenever y | x, either y = 1 or y = x. To test if x is prime,
we only have to check if y | x for all y ≤ x, since if y > x, then
automatically y ∤ x. So, the relation Prime(x), which holds iff x
is prime, can be defined by

Prime(x) ⇔ x ≥ 2 ∧ (∀y ≤ x) (y | x → y = 1 ∨ y = x)

and is thus primitive recursive.

40 CHAPTER 2. RECURSIVE FUNCTIONS

The primes are 2, 3, 5, 7, 11, etc. Consider the function

p (x) which returns the xth prime in that sequence, i.e., p (0) = 2,
p (1) = 3, p (2) = 5, etc. (For convenience we will often write p (x)
as px (p 0 = 2, p 1 = 3, etc.)
If we had a function nextPrime(x), which returns the first
prime number larger than x, p can be easily defined using prim-
itive recursion:

p (0) = 2
p (x + 1) = nextPrime(p (x))

Since nextPrime(x) is the least y such that y > x and y is prime,

it can be easily computed by unbounded search. But it can also
be defined by bounded minimization, thanks to a result due to
Euclid: there is always a prime number between x and x ! + 1.

nextPrime(x) = (min y ≤ x ! + 1) (y > x ∧ Prime(y)).

This shows, that nextPrime(x) and hence p (x) are (not just com-
putable but) primitive recursive.
(If you’re curious, here’s a quick proof of Euclid’s theorem.
Suppose p n is the largest prime ≤ x and consider the product
p = p 0 · p 1 · · · · · p n of all primes ≤ x. Either p + 1 is prime or there
is a prime between x and p + 1. Why? Suppose p + 1 is not prime.
Then some prime number q | p + 1 where q < p + 1. None of the
primes ≤ x divide p + 1. (By definition of p, each of the primes
pi ≤ x divides p, i.e., with remainder 0. So, each of the primes
pi ≤ x divides p + 1 with remainder 1, and so pi ∤ p + 1.) Hence,
q is a prime > x and < p + 1. And p ≤ x !, so there is a prime > x
and ≤ x ! + 1.)

2.11 Sequences
The set of primitive recursive functions is remarkably robust.
But we will be able to do even more once we have developed
a adequate means of handling sequences. We will identify finite
41 2.11. SEQUENCES

sequences of natural numbers with natural numbers in the fol-

lowing way: the sequence ⟨a0 ,a1 ,a2 , . . . ,ak ⟩ corresponds to the
number
a +1
p 0a0 +1 · p 1a1 +1 · p 2a2 +1 · · · · · pk k .
We add one to the exponents to guarantee that, for example, the
sequences ⟨2, 7, 3⟩ and ⟨2, 7, 3, 0, 0⟩ have distinct numeric codes.
We can take both 0 and 1 to code the empty sequence; for con-
creteness, let 𝛬 denote 0.
The reason that this coding of sequences works is the so-called
Fundamental Theorem of Arithmetic: every natural number n ≥
2 can be written in one and only one way in the form
a
n = p 0a0 · p 1a1 · · · · · pk k

with ak ≥ 1. This guarantees that the mapping ⟨⟩(a0 , . . . ,ak ) =

⟨a 0 , . . . ,ak ⟩ is injective: different sequences are mapped to differ-
ent numbers; to each number only at most one sequence corre-
sponds.
We’ll now show that the operations of determining the length
of a sequence, determining its i th element, appending an element
to a sequence, and concatenating two sequences, are all primitive
recursive.

Proposition 2.20. The function len(s ), which returns the length of

the sequence s , is primitive recursive.

Proof. Let R (i ,s ) be the relation defined by

R (i ,s ) iff pi | s ∧ pi +1 ∤ s .

R is clearly primitive recursive. Whenever s is the code of a non-

empty sequence, i.e.,
a +1
s = p 0a0 +1 · · · · · pk k ,

R (i ,s ) holds if pi is the largest prime such that pi | s , i.e., i = k .

The length of s thus is i +1 iff pi is the largest prime that divides s ,
42 CHAPTER 2. RECURSIVE FUNCTIONS

so we can let
{︄
0 if s = 0 or s = 1
len(s ) =
1 + (min i < s ) R (i ,s ) otherwise

We can use bounded minimization, since there is only one i that

satisfies R (s ,i ) when s is a code of a sequence, and if i exists it
is less than s itself. □

Proposition 2.21. The function append(s ,a), which returns the re-
sult of appending a to the sequence s , is primitive recursive.

Proof. append can be defined by:

{︄
2a+1 if s = 0 or s = 1
append(s ,a) = a+1
s · p len(s )
otherwise. □

Proposition 2.22. The function element(s ,i ), which returns the i th

element of s (where the initial element is called the 0th), or 0 if i is
greater than or equal to the length of s , is primitive recursive.

Proof. Note that a is the i th element of s iff pia+1 is the largest

power of pi that divides s , i.e., pia+1 | s but pia+2 ∤ s . So:
{︄
0 if i ≥ len(s )
element(s ,i ) =
(min a < s ) (pia+2 ∤ s ) otherwise. □

Instead of using the official names for the functions defined

above, we introduce a more compact notation. We will use (s )i
instead of element(s ,i ), and ⟨s 0 , . . . ,sk ⟩ to abbreviate

append(append(. . . append(𝛬,s 0 ) . . . ),sk ).

Note that if s has length k , the elements of s are (s )0 , . . . , (s )k −1 .

43 2.11. SEQUENCES

Proposition 2.23. The function concat(s ,t ), which concatenates

two sequences, is primitive recursive.

Proof. We want a function concat with the property that

concat(⟨a 0 , . . . ,ak ⟩, ⟨b 0 , . . . ,b l ⟩) = ⟨a0 , . . . ,ak ,b 0 , . . . ,b l ⟩.

We’ll use a “helper” function hconcat(s ,t ,n) which concatenates

the first n symbols of t to s . This function can be defined by
primitive recursion as follows:

hconcat(s ,t , 0) = s
hconcat(s ,t ,n + 1) = append(hconcat(s ,t ,n), (t )n )

Then we can define concat by

concat(s ,t ) = hconcat(s ,t , len(t )). □

We will write s ⌢ t instead of concat(s ,t ).

It will be useful for us to be able to bound the numeric code of
a sequence in terms of its length and its largest element. Suppose
s is a sequence of length k , each element of which is less than or
equal to some number x. Then s has at most k prime factors,
each at most pk −1 , and each raised to at most x + 1 in the prime
factorization of s . In other words, if we define

sequenceBound(x,k ) = pkk −1
·(x+1)
,

then the numeric code of the sequence s described above is at

most sequenceBound(x,k ).
Having such a bound on sequences gives us a way of defining
new functions using bounded search. For example, we can define
concat using bounded search. All we need to do is write down
a primitive recursive specification of the object (number of the
concatenated sequence) we are looking for, and a bound on how
far to look. The following works:

concat(s ,t ) = (min v < sequenceBound(s + t , len(s ) + len(t )))

44 CHAPTER 2. RECURSIVE FUNCTIONS

(len(v ) = len(s ) + len(t ) ∧

(∀i < len(s )) ((v )i = (s )i ) ∧
(∀j < len(t )) ((v )len(s )+ j = (t ) j ))

Proposition 2.24. The function subseq(s ,i ,n) which returns the

subsequence of s of length n beginning at the i th element, is primitive
recursive.

Proof. Exercise. □

2.12 Trees
Sometimes it is useful to represent trees as natural numbers, just
like we can represent sequences by numbers and properties of and
operations on them by primitive recursive relations and functions
on their codes. We’ll use sequences and their codes to do this. A
tree can be either a single node (possibly with a label) or else a
node (possibly with a label) connected to a number of subtrees.
The node is called the root of the tree, and the subtrees it is
connected to its immediate subtrees.
We code trees recursively as a sequence ⟨k ,d1 , . . . ,dk ⟩, where
k is the number of immediate subtrees and d1 , . . . , dk the codes
of the immediate subtrees. If the nodes have labels, they can be
included after the immediate subtrees. So a tree consisting just
of a single node with label l would be coded by ⟨0,l ⟩, and a tree
consisting of a root (labelled l1 ) connected to two single nodes
(labelled l2 , l3 ) would be coded by ⟨2, ⟨0,l2 ⟩, ⟨0,l3 ⟩,l1 ⟩.

Proposition 2.25. The function SubtreeSeq(t ), which returns the

code of a sequence the elements of which are the codes of all subtrees of
the tree with code t , is primitive recursive.

Proof. First note that ISubtrees(t ) = subseq(t , 1, (t )0 ) is primitive

recursive and returns the codes of the immediate subtrees of a
tree t . Now we can define a helper function hSubtreeSeq(t ,n)
45 2.13. OTHER RECURSIONS

which computes the sequence of all subtrees which are n nodes

removed from the root. The sequence of subtrees of t which is 0
nodes removed from the root—in other words, begins at the root
of t —is the sequence consisting just of t . To obtain a sequence
of all level n + 1 subtrees of t , we concatenate the level n subtrees
with a sequence consisting of all immediate subtrees of the level n
subtrees. To get a list of all these, note that if f (x) is a primitive
recursive function returning codes of sequences, then g f (s ,k ) =
f ((s )0 ) ⌢ . . . ⌢ f ((s )k ) is also primitive recursive:

g (s , 0) = f ((s )0 )
g (s ,k + 1) = g (s ,k ) ⌢ f ((s )k +1 )

For instance, if s is a sequence of trees, then h (s ) =

g ISubtrees (s , len(s )) gives the sequence of the immediate subtrees
of the elements of s . We can use it to define hSubtreeSeq by

hSubtreeSeq(t , 0) = ⟨t ⟩
hSubtreeSeq(t ,n + 1) = hSubtreeSeq(t ,n) ⌢ h (hSubtree(t ,n)).

The maximum level of subtrees in a tree coded by t , i.e., the

maximum distance between the root and a leaf node, is bounded
by the code t . So a sequence of codes of all subtrees of the tree
coded by t is given by hSubtreeSeq(t ,t ). □

2.13 Other Recursions

Using pairing and sequencing, we can justify more exotic (and
useful) forms of primitive recursion. For example, it is often use-
ful to define two functions simultaneously, such as in the following
definition:

h0 (x⃗ , 0) = f 0 (x⃗ )
h1 (x⃗ , 0) = f 1 (x⃗ )
h0 (x⃗ , y + 1) = g 0 (x⃗ , y,h0 (x⃗ , y),h1 (x⃗ , y))
h1 (x⃗ , y + 1) = g 1 (x⃗ , y,h0 (x⃗ , y),h1 (x⃗ , y))
46 CHAPTER 2. RECURSIVE FUNCTIONS

This is an instance of simultaneous recursion. Another useful way

of defining functions is to give the value of h (x⃗ , y + 1) in terms of
all the values h (x⃗ , 0), . . . , h (x⃗ , y), as in the following definition:

h (x⃗ , 0) = f (x⃗ )
h (x⃗ , y + 1) = g (x⃗ , y, ⟨h (x⃗ , 0), . . . ,h (x⃗ , y)⟩).

The following schema captures this idea more succinctly:

h (x⃗ , y) = g (x⃗ , y, ⟨h (x⃗ , 0), . . . ,h (x⃗ , y − 1)⟩)

with the understanding that the last argument to g is just the

empty sequence when y is 0. In either formulation, the idea is
that in computing the “successor step,” the function h can make
use of the entire sequence of values computed so far. This is
known as a course-of-values recursion. For a particular example, it
can be used to justify the following type of definition:
{︄
g (x⃗ , y,h (x⃗ ,k (x⃗ , y))) if k (x⃗ , y) < y
h (x⃗ , y) =
f (x⃗ ) otherwise

In other words, the value of h at y can be computed in terms of

the value of h at any previous value, given by k .
You should think about how to obtain these functions using
ordinary primitive recursion. One final version of primitive recur-
sion is more flexible in that one is allowed to change the parameters
(side values) along the way:

h (x⃗ , 0) = f (x⃗ )
h (x⃗ , y + 1) = g (x⃗ , y,h (k (x⃗ ), y))

This, too, can be simulated with ordinary primitive recursion.

(Doing so is tricky. For a hint, try unwinding the computation by
hand.)
47 2.14. NON-PRIMITIVE RECURSIVE FUNCTIONS

2.14 Non-Primitive Recursive Functions

The primitive recursive functions do not exhaust the intuitively
computable functions. It should be intuitively clear that we can
make a list of all the unary primitive recursive functions, f0 , f1 ,
f2 , . . . such that we can effectively compute the value of fx on
input y; in other words, the function g (x, y), defined by

g (x, y) = fx (y)

is computable. But then so is the function

h (x) = g (x,x) + 1
= fx (x) + 1.

For each primitive recursive function fi , the value of h and fi

differ at i . So h is computable, but not primitive recursive; and
one can say the same about g . This is an “effective” version of
Cantor’s diagonalization argument.
One can provide more explicit examples of computable func-
tions that are not primitive recursive. For example, let the nota-
tion g n (x) denote g (g (. . . g (x))), with n g ’s in all; and define a
sequence g 0 , g 1 , . . . of functions by

g 0 (x) = x + 1
g n+1 (x) = g nx (x)

You can confirm that each function g n is primitive recursive. Each

successive function grows much faster than the one before; g 1 (x)
is equal to 2x, g 2 (x) is equal to 2x ·x, and g 3 (x) grows roughly like
an exponential stack of x 2’s. Ackermann’s function is essentially
the function G (x) = g x (x), and one can show that this grows
faster than any primitive recursive function.
Let us return to the issue of enumerating the primitive re-
cursive functions. Remember that we have assigned symbolic
notations to each primitive recursive function; so it suffices to
48 CHAPTER 2. RECURSIVE FUNCTIONS

enumerate notations. We can assign a natural number #(F ) to

each notation F , recursively, as follows:
#(0) = ⟨0⟩
#(S ) = ⟨1⟩
#(Pin ) = ⟨2,n,i ⟩
#(Compk ,l [H ,G 0 , . . . ,G k −1 ]) = ⟨3,k ,l , #(H ), #(G 0 ), . . . , #(G k −1 )⟩
#(Recl [G ,H ]) = ⟨4,l , #(G ), #(H )⟩
Here we are using the fact that every sequence of numbers can
be viewed as a natural number, using the codes from the last sec-
tion. The upshot is that every code is assigned a natural number.
Of course, some sequences (and hence some numbers) do not
correspond to notations; but we can let fi be the unary primitive
recursive function with notation coded as i , if i codes such a no-
tation; and the constant 0 function otherwise. The net result is
that we have an explicit way of enumerating the unary primitive
recursive functions.
(In fact, some functions, like the constant zero function, will
appear more than once on the list. This is not just an artifact
of our coding, but also a result of the fact that the constant zero
function has more than one notation. We will later see that one
can not computably avoid these repetitions; for example, there
is no computable function that decides whether or not a given
notation represents the constant zero function.)
We can now take the function g (x, y) to be given by fx (y),
where fx refers to the enumeration we have just described. How
do we know that g (x, y) is computable? Intuitively, this is clear:
to compute g (x, y), first “unpack” x, and see if it is a notation for
a unary function. If it is, compute the value of that function on
input y.
You may already be convinced that (with some work!) one
can write a program (say, in Java or C++) that does this; and
now we can appeal to the Church-Turing thesis, which says that
anything that, intuitively, is computable can be computed by a
Turing machine.
49 2.15. PARTIAL RECURSIVE FUNCTIONS

Of course, a more direct way to show that g (x, y) is com-

putable is to describe a Turing machine that computes it, explic-
itly. This would, in particular, avoid the Church-Turing thesis and
appeals to intuition. Soon we will have built up enough machin-
ery to show that g (x, y) is computable, appealing to a model of
computation that can be simulated on a Turing machine: namely,
the recursive functions.

2.15 Partial Recursive Functions

To motivate the definition of the recursive functions, note that
our proof that there are computable functions that are not primi-
tive recursive actually establishes much more. The argument was
simple: all we used was the fact was that it is possible to enumer-
ate functions f 0 , f 1 , . . . such that, as a function of x and y, fx (y)
is computable. So the argument applies to any class of functions
that can be enumerated in such a way. This puts us in a bind: we
would like to describe the computable functions explicitly; but
any explicit description of a collection of computable functions
cannot be exhaustive!
The way out is to allow partial functions to come into play.
We will see that it is possible to enumerate the partial computable
functions. In fact, we already pretty much know that this is the
case, since it is possible to enumerate Turing machines in a sys-
tematic way. We will come back to our diagonal argument later,
and explore why it does not go through when partial functions
are included.
The question is now this: what do we need to add to the
primitive recursive functions to obtain all the partial recursive
functions? We need to do two things:

1. Modify our definition of the primitive recursive functions

to allow for partial functions as well.

2. Add something to the definition, so that some new partial

functions are included.
50 CHAPTER 2. RECURSIVE FUNCTIONS

The first is easy. As before, we will start with zero, succes-

sor, and projections, and close under composition and primitive
recursion. The only difference is that we have to modify the def-
initions of composition and primitive recursion to allow for the
possibility that some of the terms in the definition are not de-
fined. If f and g are partial functions, we will write f (x) ↓ to
mean that f is defined at x, i.e., x is in the domain of f ; and
f (x) ↑ to mean the opposite, i.e., that f is not defined at x. We
will use f (x) ≃ g (x) to mean that either f (x) and g (x) are both
undefined, or they are both defined and equal. We will use these
notations for more complicated terms as well. We will adopt the
convention that if h and g 0 , . . . , g k all are partial functions, then

h (g 0 (x⃗ ), . . . , g k (x⃗ ))

is defined if and only if each g i is defined at x⃗ , and h is defined

at g 0 (x⃗ ), . . . , g k (x⃗ ). With this understanding, the definitions of
composition and primitive recursion for partial functions is just
as above, except that we have to replace “=” by “≃”.
What we will add to the definition of the primitive recursive
functions to obtain partial functions is the unbounded search oper-
ator. If f (x, z⃗) is any partial function on the natural numbers,
define 𝜇x f (x, z⃗) to be

the least x such that f (0, z⃗), f (1, z⃗), . . . , f (x, z⃗) are all
defined, and f (x, z⃗) = 0, if such an x exists

with the understanding that 𝜇x f (x, z⃗) is undefined otherwise.

This defines 𝜇x f (x, z⃗) uniquely.
Note that our definition makes no reference to Turing ma-
chines, or algorithms, or any specific computational model. But
like composition and primitive recursion, there is an opera-
tional, computational intuition behind unbounded search. When
it comes to the computability of a partial function, arguments
where the function is undefined correspond to inputs for which
the computation does not halt. The procedure for computing
𝜇x f (x, z⃗) will amount to this: compute f (0, z⃗), f (1, z⃗), f (2, z⃗)
51 2.16. THE NORMAL FORM THEOREM

until a value of 0 is returned. If any of the intermediate com-

putations do not halt, however, neither does the computation of
𝜇x f (x, z⃗).
If R (x, z⃗) is any relation, 𝜇x R (x, z⃗) is defined to be 𝜇x (1 −̇
𝜒 R (x, z⃗)). In other words, 𝜇x R (x, z⃗) returns the least value of x
such that R (x, z⃗) holds. So, if f (x, z⃗) is a total function, 𝜇x f (x, z⃗)
is the same as 𝜇x ( f (x, z⃗) = 0). But note that our original defini-
tion is more general, since it allows for the possibility that f (x, z⃗)
is not everywhere defined (whereas, in contrast, the characteristic
function of a relation is always total).

Definition 2.26. The set of partial recursive functions is the small-

est set of partial functions from the natural numbers to the nat-
ural numbers (of various arities) containing zero, successor, and
projections, and closed under composition, primitive recursion,
and unbounded search.

Of course, some of the partial recursive functions will happen

to be total, i.e., defined for every argument.

Definition 2.27. The set of recursive functions is the set of partial

recursive functions that are total.

A recursive function is sometimes called “total recursive” to

emphasize that it is defined everywhere.

2.16 The Normal Form Theorem

Theorem 2.28 (Kleene’s Normal Form Theorem). There is a
primitive recursive relation T (e ,x,s ) and a primitive recursive function
U (s ), with the following property: if f is any partial recursive function,
then for some e ,
f (x) ≃ U (𝜇s T (e ,x,s ))
for every x.
52 CHAPTER 2. RECURSIVE FUNCTIONS

The proof of the normal form theorem is involved, but the ba-
sic idea is simple. Every partial recursive function has an index e ,
intuitively, a number coding its program or definition. If f (x) ↓,
the computation can be recorded systematically and coded by
some number s , and that s codes the computation of f on in-
put x can be checked primitive recursively using only x and the
definition e . This means that T is primitive recursive. Given the
full record of the computation s , the “upshot” of s is the value
of f (x), and it can be obtained from s primitive recursively as
well.
The normal form theorem shows that only a single un-
bounded search is required for the definition of any partial recur-
sive function. We can use the numbers e as “names” of partial
recursive functions, and write 𝜑e for the function f defined by the
equation in the theorem. Note that any partial recursive function
can have more than one index—in fact, every partial recursive
function has infinitely many indices.

2.17 The Halting Problem

The halting problem in general is the problem of deciding, given
the specification e (e.g., program) of a computable function and
a number n, whether the computation of the function on input n
halts, i.e., produces a result. Famously, Alan Turing proved that
this problem itself cannot be solved by a computable function,
i.e., the function
{︄
1 if computation e halts on input n
h (e ,n) =
0 otherwise,

is not computable.
In the context of partial recursive functions, the role of the
specification of a program may be played by the index e given in
Kleene’s normal form theorem. If f is a partial recursive func-
tion, any e for which the equation in the normal form theorem
53 2.17. THE HALTING PROBLEM

holds, is an index of f . Given a number e , the normal form

theorem states that

𝜑e (x) ≃ U (𝜇s T (e ,x,s ))

is partial recursive, and for every partial recursive f : N → N,

there is an e ∈ N such that 𝜑e (x) ≃ f (x) for all x ∈ N. In fact,
for each such f there is not just one, but infinitely many such e .
The halting function h is defined by
{︄
1 if 𝜑e (x) ↓
h (e ,x) =
0 otherwise.

Note that h (e ,x) = 0 if 𝜑e (x) ↑, but also when e is not the index
of a partial recursive function at all.

Theorem 2.29. The halting function h is not partial recursive.

Proof. If h were partial recursive, we could define

{︄
1 if h (y, y) = 0
d (y) =
𝜇x x ≠ x otherwise.

From this definition it follows that

1. d (y) ↓ iff 𝜑y (y) ↑ or y is not the index of a partial recursive

function.

2. d (y) ↑ iff 𝜑y (y) ↓.

If h were partial recursive, then d would be partial recursive as

well. Thus, by the Kleene normal form theorem, it has an index
e d . Consider the value of h (e d ,e d ). There are two possible cases,
0 and 1.

1. If h (e d ,e d ) = 1 then 𝜑ed (e d ) ↓. But 𝜑ed ≃ d , and d (e d ) is

defined iff h (e d ,e d ) = 0. So h (e d ,e d ) ≠ 1.
54 CHAPTER 2. RECURSIVE FUNCTIONS

2. If h (e d ,e d ) = 0 then either e d is not the index of a partial

recursive function, or it is and 𝜑ed (e d ) ↑. But again, 𝜑ed ≃ d ,
and d (e d ) is undefined iff 𝜑ed (e d ) ↓.

The upshot is that e d cannot, after all, be the index of a partial

recursive function. But if h were partial recursive, d would be too,
and so our definition of e d as an index of it would be admissible.
We must conclude that h cannot be partial recursive. □

2.18 General Recursive Functions

There is another way to obtain a set of total functions. Say a total
function f (x, z⃗) is regular if for every sequence of natural numbers
z⃗, there is an x such that f (x, z⃗) = 0. In other words, the regu-
lar functions are exactly those functions to which one can apply
unbounded search, and end up with a total function. One can,
conservatively, restrict unbounded search to regular functions:

Definition 2.30. The set of general recursive functions is the small-

est set of functions from the natural numbers to the natural num-
bers (of various arities) containing zero, successor, and projec-
tions, and closed under composition, primitive recursion, and
unbounded search applied to regular functions.

Clearly every general recursive function is total. The differ-

ence between Definition 2.30 and Definition 2.27 is that in the
latter one is allowed to use partial recursive functions along the
way; the only requirement is that the function you end up with
at the end is total. So the word “general,” a historic relic, is
a misnomer; on the surface, Definition 2.30 is less general than
Definition 2.27. But, fortunately, the difference is illusory; though
the definitions are different, the set of general recursive functions
and the set of recursive functions are one and the same.
55 2.18. GENERAL RECURSIVE FUNCTIONS

Summary
In order to show that Q represents all computable functions, we
need a precise model of computability that we can take as the
basis for a proof. There are, of course, many models of com-
putability, such as Turing machines. One model that plays a sig-
nificant role historically—it’s one of the first models proposed,
and is also the one used by Gödel himself—is that of the recur-
sive functions. The recursive functions are a class of arithmeti-
cal functions—that is, their domain and range are the natural
numbers—that can be defined from a few basic functions using a
few operations. The basic functions are zero, succ, and the pro-
jection functions. The operations are composition, primitive
recursion, and regular minimization. Composition is simply a
general version of “chaining together” functions: first apply one,
then apply the other to the result. Primitive recursion defines a
new function f from two functions g , h already defined, by stipu-
lating that the value of f for 0 is given by g , and the value for any
number n + 1 is given by h applied to f (n). Functions that can
be defined using just these two principles are called primitive
recursive. A relation is primitive recursive iff its characteristic
function is. It turns out that a whole list of interesting functions
and relations are primitive recursive (such as addition, multi-
plication, exponentiation, divisibility), and that we can define
new primitive recursive functions and relations from old ones us-
ing principles such as bounded quantification and bounded min-
imization. In particular, this allowed us to show that we can deal
with sequences of numbers in primitive recursive ways. That is,
there is a way to “code” sequences of numbers as single num-
bers in such a way that we can compute the i -the element, the
length, the concatenation of two sequences, etc., all using prim-
itive recursive functions operating on these codes. To obtain all
the computable functions, we finally added definition by regular
minimization to composition and primitive recursion. A func-
tion g (x, y) is regular iff, for every y it takes the value 0 for at
last one x. If f is regular, the least x such that g (x, y) = 0 al-
56 CHAPTER 2. RECURSIVE FUNCTIONS

ways exists, and can be found simply by computing all the values
of g (0, y), g (1, y), etc., until one of them is = 0. The resulting
function f (y) = 𝜇x g (x, y) = 0 is the function defined by regular
minimization from g . It is always total and computable. The re-
sulting set of functions are called general recursive. One version
of the Church-Turing Thesis says that the computable arithmeti-
cal functions are exactly the general recursive ones.

Problems
Problem 2.1. Prove Proposition 2.5 by showing that the prim-
itive recursive definition of mult is can be put into the form re-
quired by Definition 2.1 and showing that the corresponding func-
tions f and g are primitive recursive.

Problem 2.2. Give the complete primitive recursive notation for

mult.

Problem 2.3. Prove Proposition 2.13.

Problem 2.4. Show that

x
}︃
.2
..
f (x, y) = 2 (2 ) y 2’s

is primitive recursive.

Problem 2.5. Show that integer division d (x, y) = ⌊x/y⌋ (i.e., di-
vision, where you disregard everything after the decimal point)
is primitive recursive. When y = 0, we stipulate d (x, y) = 0. Give
an explicit definition of d using primitive recursion and compo-
sition.

Problem 2.6. Suppose R (x⃗ , z ) is primitive recursive. Define the

function mR′ (x⃗ , y) which returns the least z less than y such that
R (x⃗ , z ) holds, if there is one, and 0 otherwise, by primitive recur-
sion from 𝜒 R .
57 2.18. GENERAL RECURSIVE FUNCTIONS

Problem 2.7. Define integer division d (x, y) using bounded min-

imization.

Problem 2.8. Show that there is a primitive recursive func-

tion sconcat(s ) with the property that

sconcat(⟨s 0 , . . . ,sk ⟩) = s 0 ⌢ . . . ⌢ sk .

Problem 2.9. Show that there is a primitive recursive func-

tion tail(s ) with the property that

tail(𝛬) = 0 and
tail(⟨s 0 , . . . ,sk ⟩) = ⟨s 1 , . . . ,sk ⟩.

Problem 2.10. Prove Proposition 2.24.

Problem 2.11. The definition of hSubtreeSeq in the proof of

Proposition 2.25 in general includes repetitions. Give an alterna-
tive definition which guarantees that the code of a subtree occurs
only once in the resulting list.
CHAPTER 3

Arithmetization
of Syntax
3.1 Introduction
In order to connect computability and logic, we need a way to talk
about the objects of logic (symbols, terms, formulas, derivations),
operations on them, and their properties and relations, in a way
amenable to computational treatment. We can do this directly,
by considering computable functions and relations on symbols,
sequences of symbols, and other objects built from them. Since
the objects of logical syntax are all finite and built from a count-
able sets of symbols, this is possible for some models of compu-
tation. But other models of computation—such as the recursive
functions—-are restricted to numbers, their relations and func-
tions. Moreover, ultimately we also want to be able to deal with
syntax within certain theories, specifically, in theories formulated
in the language of arithmetic. In these cases it is necessary to
arithmetize syntax, i.e., to represent syntactic objects, operations
on them, and their relations, as numbers, arithmetical functions,
and arithmetical relations, respectively. The idea, which goes
back to Leibniz, is to assign numbers to syntactic objects.
It is relatively straightforward to assign numbers to symbols
as their “codes.” Some symbols pose a bit of a challenge, since,

58
59 3.1. INTRODUCTION

e.g., there are infinitely many variables, and even infinitely many
function symbols of each arity n. But of course it’s possible to
assign numbers to symbols systematically in such a way that, say,
v2 and v3 are assigned different codes. Sequences of symbols
(such as terms and formulas) are a bigger challenge. But if we can
deal with sequences of numbers purely arithmetically (e.g., by the
powers-of-primes coding of sequences), we can extend the coding
of individual symbols to coding of sequences of symbols, and then
further to sequences or other arrangements of formulas, such as
derivations. This extended coding is called “Gödel numbering.”
Every term, formula, and derivation is assigned a Gödel number.
By coding sequences of symbols as sequences of their codes,
and by chosing a system of coding sequences that can be dealt
with using computable functions, we can then also deal with
Gödel numbers using computable functions. In practice, all the
relevant functions will be primitive recursive. For instance, com-
puting the length of a sequence and computing the i -th element
of a sequence from the code of the sequence are both primitive
recursive. If the number coding the sequence is, e.g., the Gödel
number of a formula A, we immediately see that the length of a
formula and the (code of the) i -th symbol in a formula can also be
computed from the Gödel number of A. It is a bit harder to prove
that, e.g., the property of being the Gödel number of a correctly
formed term or of a correct derivation is primitive recursive. It is
nevertheless possible, because the sequences of interest (terms,
formulas, derivations) are inductively defined.
As an example, consider the operation of substitution. If A
is a formula, x a variable, and t a term, then A[t /x] is the result
of replacing every free occurrence of x in A by t . Now suppose
we have assigned Gödel numbers to A, x, t —say, k , l , and m, re-
spectively. The same scheme assigns a Gödel number to A[t /x],
say, n. This mapping—of k , l , and m to n—is the arithmetical
analog of the substitution operation. When the substitution oper-
ation maps A, x, t to A[t /x], the arithmetized substitution func-
tions maps the Gödel numbers k , l , m to the Gödel number n.
We will see that this function is primitive recursive.
60 CHAPTER 3. ARITHMETIZATION OF SYNTAX

Arithmetization of syntax is not just of abstract interest, al-

though it was originally a non-trivial insight that languages like
the language of arithmetic, which do not come with mechanisms
for “talking about” languages can, after all, formalize complex
properties of expressions. It is then just a small step to ask what
a theory in this language, such as Peano arithmetic, can prove
about its own language (including, e.g., whether sentences are
provable or true). This leads us to the famous limitative theorems
of Gödel (about unprovability) and Tarski (the undefinability of
truth). But the trick of arithmetizing syntax is also important in
order to prove some important results in computability theory,
e.g., about the computational prower of theories or the relation-
ship between different models of computability. The arithmetiza-
tion of syntax serves as a model for arithmetizing other objects
and properties. For instance, it is similarly possible to arithme-
tize configurations and computations (say, of Turing machines).
This makes it possible to simulate computations in one model
(e.g., Turing machines) in another (e.g., recursive functions).

3.2 Coding Symbols

The basic language L of first order logic makes use of the symbols

⊥ ¬ ∨ ∧ → ∀ ∃ = ( ) ,

together with countable sets of variables and constant symbols,

and countable sets of function symbols and predicate symbols
of arbitrary arity. We can assign codes to each of these symbols
in such a way that every symbol is assigned a unique number
as its code, and no two different symbols are assigned the same
number. We know that this is possible since the set of all symbols
is countable and so there is a bijection between it and the set of
natural numbers. But we want to make sure that we can recover
the symbol (as well as some information about it, e.g., the arity of
a function symbol) from its code in a computable way. There are
many possible ways of doing this, of course. Here is one such way,
61 3.2. CODING SYMBOLS

which uses primitive recursive functions. (Recall that ⟨n 0 , . . . ,nk ⟩

is the number coding the sequence of numbers n 0 , . . . , nk .)

Definition 3.1. If s is a symbol of L, let the symbol code cs be

defined as follows:

1. If s is among the logical symbols, cs is given by the follow-

ing table:

⊥ ¬ ∨ ∧ → ∀
⟨0, 0⟩ ⟨0, 1⟩ ⟨0, 2⟩ ⟨0, 3⟩ ⟨0, 4⟩ ⟨0, 5⟩
∃ = ( ) ,
⟨0, 6⟩ ⟨0, 7⟩ ⟨0, 8⟩ ⟨0, 9⟩ ⟨0, 10⟩

2. If s is the i -th variable vi , then cs = ⟨1,i ⟩.

3. If s is the i -th constant symbol ci , then cs = ⟨2,i ⟩.

4. If s is the i -th n-ary function symbol fin , then cs = ⟨3,n,i ⟩.

5. If s is the i -th n-ary predicate symbol Pin , then cs = ⟨4,n,i ⟩.

Proposition 3.2. The following relations are primitive recursive:

1. Fn(x,n) iff x is the code of fin for some i , i.e., x is the code of an
n-ary function symbol.

2. Pred(x,n) iff x is the code of Pin for some i or x is the code of =

and n = 2, i.e., x is the code of an n-ary predicate symbol.
62 CHAPTER 3. ARITHMETIZATION OF SYNTAX

Definition 3.3. If s 0 , . . . ,sn−1 is a sequence of symbols, its Gödel

number is ⟨cs0 , . . . , csn−1 ⟩.

Note that codes and Gödel numbers are different things. For
instance, the variable v5 has a code cv5 = ⟨1, 5⟩ = 22 · 36 . But the
variable v5 considered as a term is also a sequence of symbols (of
length 1). The Gödel number # v5 # of the term v5 is ⟨cv5 ⟩ = 2cv5 +1 =
2 6
22 ·3 +1 .

Example 3.4. Recall that if k0 , . . . , kn−1 is a sequence of num-

bers, then the code of the sequence ⟨k 0 , . . . ,kn−1 ⟩ in the power-
of-primes coding is
kn−1
2k0 +1 · 3k1 +1 · · · · · p n−1 ,

where pi is the i -th prime (starting with p 0 = 2). So for instance,

the formula v0 = 0, or, more explicitly, =(v0 , c0 ), has the Gödel
number
⟨c= , c ( , cv0 , c,, cc0 , c) ⟩.
Here, c= is ⟨0, 7⟩ = 20+1 · 37=1 , cv0 is ⟨1, 0⟩ = 21+1 · 30+1 , etc. So
#
= (v0 , c0 ) # is

2c= +1 · 3c ( +1 · 5cv0 +1 · 7c,+1 · 11cc0 +1 · 13c) +1 =

1 ·38 +1 1 ·39 +1 2 ·31 +1 1 ·311 +1 3 ·31 +1 1 ·310 +1
22 · 32 · 52 · 72 · 112 · 132 =
13 123 39 367 13 354 295 25 118 099
2 ·3 ·5 ·7 · 11 · 13 .

3.3 Coding Terms

A term is simply a certain kind of sequence of symbols: it is
built up inductively from constants and variables according to
the formation rules for terms. Since sequences of symbols can be
coded as numbers—using a coding scheme for the symbols plus
a way to code sequences of numbers—assigning Gödel numbers
to terms is not difficult. The challenge is rather to show that the
63 3.3. CODING TERMS

property a number has if it is the Gödel number of a correctly

formed term is computable, or in fact primitive recursive.
Variables and constant symbols are the simplest terms, and
testing whether x is the Gödel number of such a term is easy:
Var(x) holds if x is # vi # for some i . In other words, x is a se-
quence of length 1 and its single element (x)0 is the code of some
variable vi , i.e., x is ⟨⟨1,i ⟩⟩ for some i . Similarly, Const(x) holds
if x is # ci # for some i . Both of these relations are primitive recur-
sive, since if such an i exists, it must be < x:
Var(x) ⇔ (∃i < x) x = ⟨⟨1,i ⟩⟩
Const(x) ⇔ (∃i < x) x = ⟨⟨2,i ⟩⟩

Proposition 3.5. The relations Term(x) and ClTerm(x) which

hold iff x is the Gödel number of a term or a closed term, respectively,
are primitive recursive.

Proof. A sequence of symbols s is a term iff there is a sequence s0 ,

. . . , sk −1 = s of terms which records how the term s was formed
from constant symbols and variables according to the formation
rules for terms. To express that such a putative formation se-
quence follows the formation rules it has to be the case that, for
each i < k , either
1. si is a variable v j , or
2. si is a constant symbol c j , or
3. si is built from n terms t1 , . . . , tn occurring prior to place i
using an n-place function symbol f jn .
To show that the corresponding relation on Gödel numbers is
primitive recursive, we have to express this condition primitive
recursively, i.e., using primitive recursive functions, relations, and
bounded quantification.
Suppose y is the number that codes the sequence s 0 , . . . , sk −1 ,
i.e., y = ⟨# s 0 # , . . . , # sk −1 # ⟩. It codes a formation sequence for the
term with Gödel number x iff for all i < k :
64 CHAPTER 3. ARITHMETIZATION OF SYNTAX

1. Var((y)i ), or

2. Const((y)i ), or

3. there is an n and a number z = ⟨z 1 , . . . , z n ⟩ such that each

zl is equal to some (y)i ′ for i ′ < i and

(y)i = # f jn ( # ⌢ flatten(z ) ⌢ # ) # ,

and moreover (y)k −1 = x. (The function flatten(z ) turns the se-

quence ⟨#t1 # , . . . , #tn # ⟩ into #t1 , . . . ,tn # and is primitive recursive.)
The indices j , n, the Gödel numbers zl of the terms tl , and the
code z of the sequence ⟨z 1 , . . . , z n ⟩, in (3) are all less than y. We
can replace k above with len(y). Hence we can express “y is the
code of a formation sequence of the term with Gödel number x”
in a way that shows that this relation is primitive recursive.
We now just have to convince ourselves that there is a prim-
itive recursive bound on y. But if x is the Gödel number of
a term, it must have a formation sequence with at most len(x)
terms (since every term in the formation sequence of s must start
at some place in s , and no two subterms can start at the same
place). The Gödel number of each subterm of s is of course ≤ x.
Hence, there always is a formation sequence with code ≤ x len(x) .
For ClTerm, simply leave out the clause for variables. □

Proposition 3.6. The function num(n) = # n # is primitive recursive.

Proof. We define num(n) by primitive recursion:

num(0) = # 0#
num(n + 1) = #′( # ⌢ num(n) ⌢ # ) # . □
65 3.4. CODING FORMULAS

3.4 Coding Formulas

Proposition 3.7. The relation Atom(x) which holds iff x is the
Gödel number of an atomic formula, is primitive recursive.

Proof. The number x is the Gödel number of an atomic formula

iff one of the following holds:

1. There are n, j < x, and z < x such that for each i < n,
Term((z )i ) and x =
#
P jn ( # ⌢ flatten(z ) ⌢ # ) # .

2. There are z 1 , z 2 < x such that Term(z 1 ), Term(z 2 ), and x =

#
=( # ⌢ z 1 ⌢ # ,# ⌢ z 2 ⌢ # ) # .

3. x = # ⊥# . □

Proposition 3.8. The relation Frm(x) which holds iff x is the Gödel
number of a formula is primitive recursive.

Proof. A sequence of symbols s is a formula iff there is formation

sequence s 0 , . . . , sk −1 = s of formula which records how s was
formed from atomic formulas according to the formation rules.
The code for each si (and indeed of the code of the sequence
⟨s 0 , . . . ,sk −1 ⟩) is less than the code x of s . □

Proposition 3.9. The relation FreeOcc(x, z ,i ), which holds iff the

i -th symbol of the formula with Gödel number x is a free occurrence of
the variable with Gödel number z , is primitive recursive.

Proof. Exercise. □
66 CHAPTER 3. ARITHMETIZATION OF SYNTAX

Proposition 3.10. The property Sent(x) which holds iff x is the

Gödel number of a sentence is primitive recursive.

Proof. A sentence is a formula without free occurrences of vari-

ables. So Sent(x) holds iff

(∀i < len(x)) (∀z < x)

((∃ j < z ) z = # v j # → ¬FreeOcc(x, z ,i )). □

3.5 Substitution
Recall that substitution is the operation of replacing all free oc-
currences of a variable u in a formula A by a term t , written
A[t /u]. This operation, when carried out on Gödel numbers of
variables, formulas, and terms, is primitive recursive.

Proposition 3.11. There is a primitive recursive function

Subst(x, y, z ) with the property that

Subst( #A# , #t # , #u # ) = #A[t /u] #

Proof. We can then define a function hSubst by primitive recur-

sion as follows:

hSubst(x, y, z , 0) = 𝛬
hSubst(x, y, z ,i + 1) =
{︄
hSubst(x, y, z ,i ) ⌢ y if FreeOcc(x, z ,i )
append(hSubst(x, y, z ,i ), (x)i ) otherwise.

Subst(x, y, z ) can now be defined as hSubst(x, y, z , len(x)). □

Proposition 3.12. The relation FreeFor(x, y, z ), which holds iff the

term with Gödel number y is free for the variable with Gödel number z
67 3.6. DERIVATIONS IN NATURAL DEDUCTION

in the formula with Gödel number x, is primitive recursive.

Proof. Exercise. □

3.6 Derivations in Natural Deduction

In order to arithmetize derivations, we must represent derivations
as numbers. Since derivations are trees of formulas where each
inference carries one or two labels, a recursive representation
is the most obvious approach: we represent a derivation as a
tuple, the components of which are the number of immediate
sub-derivations leading to the premises of the last inference, the
representations of these sub-derivations, and the end-formula, the
discharge label of the last inference, and a number indicating the
type of the last inference.

Definition 3.13. If 𝛿 is a derivation in natural deduction, then

# #
𝛿 is defined inductively as follows:

1. If 𝛿 consists only of the assumption A, then # 𝛿 # is ⟨0, #A# ,n⟩.

The number n is 0 if it is an undischarged assumption, and
the numerical label otherwise.

2. If 𝛿 ends in an inference with one, two, or three premises,

then # 𝛿 # is

⟨1, # 𝛿1 # , #A# ,n,k ⟩,

⟨2, # 𝛿1 # , # 𝛿2 # , #A# ,n,k ⟩, or
⟨3, # 𝛿1 # , # 𝛿2 # , # 𝛿3 # , #A# ,n,k ⟩,

respectively. Here 𝛿1 , 𝛿2 , 𝛿3 are the sub-derivations ending

in the premise(s) of the last inference in 𝛿, A is the conclu-
sion of the last inference in 𝛿, n is the discharge label of the
last inference (0 if the inference does not discharge any as-
68 CHAPTER 3. ARITHMETIZATION OF SYNTAX

sumptions), and k is given by the following table according

to which rule was used in the last inference.
Rule: ∧Intro ∧Elim ∨Intro ∨Elim
k: 1 2 3 4
Rule: →Intro →Elim ¬Intro ¬Elim
k: 5 6 7 8
Rule: ⊥I ⊥C ∀Intro ∀Elim
k: 9 10 11 12
Rule: ∃Intro ∃Elim =Intro =Elim
k: 13 14 15 16

Example 3.14. Consider the very simple derivation

[A ∧ B] 1
∧Elim
A
1 →Intro
(A ∧ B) → A

The Gödel number of the assumption would be d0 =

⟨0, #A ∧ B # , 1⟩. The Gödel number of the derivation ending in
the conclusion of ∧Elim would be d1 = ⟨1,d0 , #A# , 0, 2⟩ (1 since
∧Elim has one premise, the Gödel number of conclusion A, 0
because no assumption is discharged, and 2 is the number cod-
ing ∧Elim). The Gödel number of the entire derivation then is
⟨1,d1 , # ((A ∧ B) → A) # , 1, 5⟩, i.e.,

⟨1, ⟨1, ⟨0, # (A ∧ B) # , 1⟩, #A# , 0, 2⟩, # ((A ∧ B) → A) # , 1, 5⟩.

Having settled on a representation of derivations, we must

also show that we can manipulate Gödel numbers of such deriva-
tions primitive recursively, and express their essential properties
and relations. Some operations are simple: e.g., given a Gödel
number d of a derivation, EndFmla(d ) = (d ) (d )0 +1 gives us the
Gödel number of its end-formula, DischargeLabel(d ) = (d ) (d )0 +2
gives us the discharge label and LastRule(d ) = (d ) (d )0 +3 the num-
ber indicating the type of the last inference. Some are much
69 3.6. DERIVATIONS IN NATURAL DEDUCTION

harder. We’ll at least sketch how to do this. The goal is to show

that the relation “𝛿 is a derivation of A from 𝛤” is a primitive
recursive relation of the Gödel numbers of 𝛿 and A.

Proposition 3.15. The following relations are primitive recursive:

1. A occurs as an assumption in 𝛿 with label n.

2. All assumptions in 𝛿 with label n are of the form A (i.e., we can

discharge the assumption A using label n in 𝛿).

Proof. We have to show that the corresponding relations between

Gödel numbers of formulas and Gödel numbers of derivations
are primitive recursive.

1. We want to show that Assum(x,d ,n), which holds if x is

the Gödel number of an assumption of the derivation with
Gödel number d labelled n, is primitive recursive. This is
the case if the derivation with Gödel number ⟨0,x,n⟩ is a
sub-derivation of d . Note that the way we code derivations
is a special case of the coding of trees introduced in sec-
tion 2.12, so the primitive recursive function SubtreeSeq(d )
gives a sequence of Gödel numbers of all sub-derivations
of d (of length a most d ). So we can define

Assum(x,d ,n) ⇔ (∃i < d ) (SubtreeSeq(d ))i = ⟨0,x,n⟩.

2. We want to show that Discharge(x,d ,n), which holds if all

assumptions with label n in the derivation with Gödel num-
ber d all are the formula with Gödel number x. But this
relation holds iff (∀y < d ) (Assum(y,d ,n) → y = x). □
70 CHAPTER 3. ARITHMETIZATION OF SYNTAX

Proposition 3.16. The property Correct(d ) which holds iff the last
inference in the derivation 𝛿 with Gödel number d is correct, is primitive
recursive.

Proof. Here we have to show that for each rule of infer-

ence R the relation FollowsByR (d ) is primitive recursive, where
FollowsByR (d ) holds iff d is the Gödel number of derivation 𝛿,
and the end-formula of 𝛿 follows by a correct application of R
from the immediate sub-derivations of 𝛿.
A simple case is that of the ∧Intro rule. If 𝛿 ends in a correct
∧Intro inference, it looks like this:

𝛿1 𝛿2
A B
∧Intro
A∧B

Then the Gödel number d of 𝛿 is ⟨2,d1 ,d2 , # (A ∧ B) # , 0,k ⟩ where

EndFmla(d1 ) = #A# , EndFmla(d2 ) = # B # , n = 0, and k = 1. So we
can define FollowsBy∧Intro (d ) as

(d )0 = 2 ∧ DischargeLabel(d ) = 0 ∧ LastRule(d ) = 1 ∧
EndFmla(d ) = # ( # ⌢ EndFmla((d )1 ) ⌢ # ∧# ⌢ EndFmla((d )2 ) ⌢ # ) # .

Another simple example if the =Intro rule. Here the premise

is an empty derivation, i.e., (d )1 = 0, and no discharge label, i.e.,
n = 0. However, A must be of the form t = t , for a closed term t .
Here, a primitive recursive definition is

(d )0 = 1 ∧ (d )1 = 0 ∧ DischargeLabel(d ) = 0 ∧
(∃t < d ) (ClTerm(t )∧EndFmla(d ) = # =( # ⌢ t ⌢ # ,# ⌢ t ⌢ # ) # )

For a more complicated example, FollowsBy→Intro (d ) holds

iff the end-formula of 𝛿 is of the form (A → B), where the end-
formula of 𝛿1 is B, and any assumption in 𝛿 labelled n is of the
form A. We can express this primitive recursively by
71 3.6. DERIVATIONS IN NATURAL DEDUCTION

(d )0 = 1 ∧
(∃a < d ) (Discharge(a, (d )1 , DischargeLabel(d )) ∧
EndFmla(d ) = ( # ( # ⌢ a ⌢ # →# ⌢ EndFmla((d )1 ) ⌢ # ) # ))

(Think of a as the Gödel number of A).

For another example, consider ∃Intro. Here, the last infer-
ence in 𝛿 is correct iff there is a formula A, a closed term t and
a variable x such that A[t /x] is the end-formula of the deriva-
tion 𝛿1 and ∃x A is the conclusion of the last inference. So,
FollowsBy ∃Intro (d ) holds iff

(d )0 = 1 ∧ DischargeLabel(d ) = 0 ∧
(∃a < d ) (∃x < d ) (∃t < d ) (ClTerm(t ) ∧ Var(x) ∧
Subst(a,t ,x) = EndFmla((d )1 )∧EndFmla(d ) = ( # ∃# ⌢ x ⌢ a)).

We then define Correct(d ) as

Sent(EndFmla(d )) ∧
(LastRule(d ) = 1 ∧ FollowsBy∧Intro (d )) ∨ · · · ∨
(LastRule(d ) = 16 ∧ FollowsBy=Elim (d )) ∨
(∃n < d ) (∃x < d ) (d = ⟨0,x,n⟩).

The first line ensures that the end-formula of d is a sentence. The

last line covers the case where d is just an assumption. □

Proposition 3.17. The relation Deriv(d ) which holds if d is the

Gödel number of a correct derivation 𝛿, is primitive recursive.

Proof. A derivation 𝛿 is correct if every one of its inferences

is a correct application of a rule, i.e., if every one of its sub-
derivations ends in a correct inference. So, Deriv(d ) iff

(∀i < len(SubtreeSeq(d ))) Correct((SubtreeSeq(d ))i ) □

72 CHAPTER 3. ARITHMETIZATION OF SYNTAX

Proposition 3.18. The relation OpenAssum(z ,d ) that holds if z is

the Gödel number of an undischarged assumption A of the derivation 𝛿
with Gödel number d , is primitive recursive.

Proof. An occurrence of an assumption is discharged if it occurs

with label n in a sub-derivation of 𝛿 that ends in a rule with dis-
charge label n. So A is an undischarged assumption of 𝛿 if at
least one of its occurrences is not discharged in 𝛿. We must be
careful: 𝛿 may contain both discharged and undischarged occur-
rences of A.
Consider a sequence 𝛿0 , . . . , 𝛿k where 𝛿0 = d , 𝛿k is the
assumption [A] n (for some n), and 𝛿i is an immediate sub-
derivation of 𝛿i +1 . If such a sequence exists in which no 𝛿i ends
in an inference with discharge label n, then A is an undischarged
assumption of 𝛿.
The primitive recursive function SubtreeSeq(d ) provides us
with a sequence of Gödel numbers of all sub-derivations of 𝛿. Any
sequence of Gödel numbers of sub-derivations of 𝛿 is a subse-
quence of it. Being a subsequence of is a primitive recursive rela-
tion: Subseq(s ,s ′) holds iff (∀i < len(s )) ∃ j < len(s ′) (s )i = (s ) j .
Being an immediate sub-derivation is as well: Subderiv(d ,d ′) iff
(∃ j < (d ′)0 ) d = (d ′) j . So we can define OpenAssum(z ,d ) by

(∃s < SubtreeSeq(d )) (Subseq(s , SubtreeSeq(d )) ∧ (s )0 = d ∧

(∃n < d ) ((s )len(s )−̇1 = ⟨0, z ,n⟩ ∧
(∀i < (len(s ) −̇ 1)) (Subderiv((s )i , (s )i +1 )] ∧
DischargeLabel((s )i +1 ) ≠ n))). □

Proposition 3.19. Suppose 𝛤 is a primitive recursive set of sentences.

Then the relation Prf 𝛤 (x, y) expressing “x is the code of a derivation 𝛿
of A from undischarged assumptions in 𝛤 and y is the Gödel number
of A” is primitive recursive.

Proof. Suppose “y ∈ 𝛤” is given by the primitive recursive pred-

icate R 𝛤 (y). We have to show that Prf 𝛤 (x, y) which holds iff y is
73 3.6. DERIVATIONS IN NATURAL DEDUCTION

the Gödel number of a sentence A and x is the code of a natural

deduction derivation with end formula A and all undischarged
assumptions in 𝛤 is primitive recursive.
By Proposition 3.17, the property Deriv(x) which holds iff x is
the Gödel number of a correct derivation 𝛿 in natural deduction
is primitive recursive. Thus we can define Prf 𝛤 (x, y) by

Prf 𝛤 (x, y) ⇔ Deriv(x) ∧ EndFmla(x) = y ∧

(∀z < x) (OpenAssum(z ,x) → R 𝛤 (z )). □

Summary
The proof of the incompleteness theorems requires that we have
a way to talk about provability in a theory (such as PA) in the
language of the theory itself, i.e., in the language of arithmetic.
But the language of arithmetic only deals with numbers, not with
formulas or derivations. The solution to this problem is to define
a systematic mapping from formulas and derivations to numbers.
The number associated with a formula or a derivation is called its
Gödel number. If A is a formula, #A# is its Gödel number. We
showed that important operations on formulas turn into primi-
tive recursive functions on the respective Gödel numbers. For
instance, A[t /x], the operation of substituting a term t for ev-
ery free occurrence of x in A, corresponds to an arithmetical
function subst(n,m,k ) which, if applied to the Gödel numbers
of A, t , and x, yields the Gödel number of A[t /x]. In other
words, subst( #A# , #t # , # x # ) = #A[t /x] # . Likewise, properties of
derivations turn into primitive recursive relations on the respec-
tive Gödel numbers. In particular, the property Deriv(n) that
holds of n if it is the Gödel number of a correct derivation in
natural deduction, is primitive recursive. Showing that these are
primitive recursive required a fair amount of work, and at times
some ingenuity, and depended essentially on the fact that op-
erating with sequences is primitive recursive. If a theory T is
decidable, then we can use Deriv to define a decidable relation
74 CHAPTER 3. ARITHMETIZATION OF SYNTAX

Prf T (n,m) which holds if n is the Gödel number of a derivation

of the sentence with Gödel number m from T. This relation is
primitive recursive if the set of axioms of T is, and merely gen-
eral recursive if the axioms of T are decidable but not primitive
recursive.

Problems
Problem 3.1. Show that the function flatten(z ), which turns the
sequence ⟨#t1 # , . . . , #tn # ⟩ into #t1 , . . . ,tn # , is primitive recursive.

Problem 3.2. Give a detailed proof of Proposition 3.8 along the

lines of the first proof of Proposition 3.5

Problem 3.3. Give a detailed proof of Proposition 3.8 along the

lines of the alternate proof of Proposition 3.5

Problem 3.4. Prove Proposition 3.9. You may make use of the
fact that any substring of a formula which is a formula is a sub-
formula of it.

Problem 3.5. Prove Proposition 3.12

Problem 3.6. Define the following properties as in Proposi-

tion 3.16:

1. FollowsBy→Elim (d ),

2. FollowsBy=Elim (d ),

3. FollowsBy∨Elim (d ),

4. FollowsBy∀Intro (d ).

For the last one, you will have to also show that you can test
primitive recursively if the last inference of the derivation with
Gödel number d satisfies the eigenvariable condition, i.e., the
75 3.6. DERIVATIONS IN NATURAL DEDUCTION

eigenvariable a of the ∀Intro inference occurs neither in the end-

formula of d nor in an open assumption of d . You may use the
primitive recursive predicate OpenAssum from Proposition 3.18
for this.
CHAPTER 4

Representability
in Q
4.1 Introduction
The incompleteness theorems apply to theories in which basic
facts about computable functions can be expressed and proved.
We will describe a very minimal such theory called “Q ” (or,
sometimes, “Robinson’s Q ,” after Raphael Robinson). We will
say what it means for a function to be representable in Q , and
then we will prove the following:
A function is representable in Q if and only if it is
computable.
For one thing, this provides us with another model of computabil-
ity. But we will also use it to show that the set {A : Q ⊢ A} is not
decidable, by reducing the halting problem to it. By the time we
are done, we will have proved much stronger things than this.
The language of Q is the language of arithmetic; Q consists
of the following axioms (to be used in conjunction with the other
axioms and rules of first-order logic with identity predicate):

∀x ∀y (x ′ = y ′ → x = y) (Q 1 )
′
∀x 0 ≠ x (Q 2 )

76
77 4.1. INTRODUCTION

∀x (x = 0 ∨ ∃y x = y ′) (Q 3 )
∀x (x + 0) = x (Q 4 )
′ ′
∀x ∀y (x + y ) = (x + y) (Q 5 )
∀x (x × 0) = 0 (Q 6 )
′
∀x ∀y (x × y ) = ((x × y) + x) (Q 7 )
′
∀x ∀y (x < y ↔ ∃z (z + x) = y) (Q 8 )

For each natural number n, define the numeral n to be the term

0′′...′ where there are n tick marks in all. So, 0 is the constant
symbol 0 by itself, 1 is 0′, 2 is 0′′, etc.
As a theory of arithmetic, Q is extremely weak; for example,
you can’t even prove very simple facts like ∀x x ≠ x ′ or ∀x ∀y (x +
y) = (y + x). But we will see that much of the reason that Q is so
interesting is because it is so weak. In fact, it is just barely strong
enough for the incompleteness theorem to hold. Another reason
Q is interesting is because it has a finite set of axioms.
A stronger theory than Q (called Peano arithmetic PA) is ob-
tained by adding a schema of induction to Q :

(A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x)

where A(x) is any formula. If A(x) contains free variables other

than x, we add universal quantifiers to the front to bind all of them
(so that the corresponding instance of the induction schema is a
sentence). For instance, if A(x, y) also contains the variable y
free, the corresponding instance is

∀y ((A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x))

Using instances of the induction schema, one can prove much

more from the axioms of PA than from those of Q . In fact, it
takes a good deal of work to find “natural” statements about the
natural numbers that can’t be proved in Peano arithmetic!
78 CHAPTER 4. REPRESENTABILITY IN Q

Definition 4.1. A function f (x 0 , . . . ,x k ) from the natural num-

bers to the natural numbers is said to be representable in Q if there
is a formula A f (x 0 , . . . ,x k , y) such that whenever f (n 0 , . . . ,nk ) =
m, Q proves

1. A f (n 0 , . . . ,nk ,m)

2. ∀y (A f (n0 , . . . ,nk , y) → m = y).

There are other ways of stating the definition; for example, we

could equivalently require that Q proves ∀y (A f (n 0 , . . . ,nk , y) ↔
y = m).

Theorem 4.2. A function is representable in Q if and only if it is

computable.

There are two directions to proving the theorem. The left-

to-right direction is fairly straightforward once arithmetization
of syntax is in place. The other direction requires more work.
Here is the basic idea: we pick “general recursive” as a way of
making “computable” precise, and show that every general re-
cursive function is representable in Q . Recall that a function
is general recursive if it can be defined from zero, the successor
function succ, and the projection functions Pin , using composi-
tion, primitive recursion, and regular minimization. So one way
of showing that every general recursive function is representable
in Q is to show that the basic functions are representable, and
whenever some functions are representable, then so are the func-
tions defined from them using composition, primitive recursion,
and regular minimization. In other words, we might show that
the basic functions are representable, and that the representable
functions are “closed under” composition, primitive recursion,
and regular minimization. This guarantees that every general
recursive function is representable.
It turns out that the step where we would show that repre-
sentable functions are closed under primitive recursion is hard.
79 4.2. FUNCTIONS REPRESENTABLE IN Q ARE COMPUTABLE

In order to avoid this step, we show first that in fact we can do

without primitive recursion. That is, we show that every gen-
eral recursive function can be defined from basic functions using
composition and regular minimization alone. To do this, we show
that primitive recursion can actually be done by a specific regular
minimization. However, for this to work, we have to add some
additional basic functions: addition, multiplication, and the char-
acteristic function of the identity relation 𝜒 = . Then, we can prove
the theorem by showing that all of these basic functions are repre-
sentable in Q , and the representable functions are closed under
composition and regular minimization.

4.2 Functions Representable in Q are

Computable
Lemma 4.3. Every function that is representable in Q is computable.

Proof. Let’s first give the intuitive idea for why this is
true. If f (x 0 , . . . ,x k ) is representable in Q , there is a for-
mula A(x 0 , . . . ,x k , y) such that

Q ⊢ A f (n 0 , . . . ,nk ,m) iff m = f (n 0 , . . . ,nk ).

To compute f , we do the following. List all the possible deriva-

tions 𝛿 in the language of arithmetic. This is possible to do me-
chanically. For each one, check if it is a derivation of a formula
of the form A f (n 0 , . . . ,nk ,m). If it is, m must be = f (n 0 , . . . ,nk )
and we’ve found the value of f . The search terminates because
Q ⊢ A f (n 0 , . . . ,nk , f (n 0 , . . . ,nk )), so eventually we find a 𝛿 of the
right sort.
This is not quite precise because our procedure operates on
derivations and formulas instead of just on numbers, and we
haven’t explained exactly why “listing all possible derivations”
is mechanically possible. But as we’ve seen, it is possible to code
terms, formulas, and derivations by Gödel numbers. We’ve also
80 CHAPTER 4. REPRESENTABILITY IN Q

introduced a precise model of computation, the general recursive

functions. And we’ve seen that the relation Prf Q (d , y), which
holds iff d is the Gödel number of a derivation of the formula
with Gödel number x from the axioms of Q , is (primitive) re-
cursive. Other primitive recursive functions we’ll need are num
(Proposition 3.6) and Subst (Proposition 3.11). From these, it is
possible to define f by minimization; thus, f is recursive.
First, define

A(n 0 , . . . ,nk ,m) =

Subst(Subst(. . . Subst( #A f # , num(n0 ), # x 0 # ),
. . . ), num(nk ), # x k # ), num(m), # y # )
This looks complicated, but it’s just the function
A(n 0 , . . . ,nk ,m) = #A f (n 0 , . . . ,nk ,m) # .
Now, consider the relation R (n 0 , . . . ,nk ,s ) which holds
if (s )0 is the Gödel number of a derivation from Q of
A f (n 0 , . . . ,nk , (s )1 ):
R (n 0 , . . . ,nk ,s ) iff Prf Q ((s )0 ,A(n 0 , . . . ,nk , (s )1 ))
If we can find an s such that R (n 0 , . . . ,nk ,s ) hold, we have found
a pair of numbers—(s )0 and (s 1 )—such that (s )0 is the Gödel
number of a derivation of A f (n 0 , . . . ,nk , (s )1 ). So looking for s
is like looking for the pair d and m in the informal proof. And a
computable function that “looks for” such an s can be defined by
regular minimization. Note that R is regular: for every n 0 , . . . ,
nk , there is a derivation 𝛿 of Q ⊢ A f (n0 , . . . ,nk , f (n 0 , . . . ,nk )),
so R (n 0 , . . . ,nk ,s ) holds for s = ⟨# 𝛿 # , f (n 0 , . . . ,nk )⟩. So, we can
write f as
f (n 0 , . . . ,nk ) = (𝜇s R (n 0 , . . . ,nk ,s ))1 . □

4.3 The Beta Function Lemma

In order to show that we can carry out primitive recursion if ad-
dition, multiplication, and 𝜒 = are available, we need to develop
81 4.3. THE BETA FUNCTION LEMMA

functions that handle sequences. (If we had exponentiation as

well, our task would be easier.) When we had primitive recur-
sion, we could define things like the “n-th prime,” and pick a
fairly straightforward coding. But here we do not have primi-
tive recursion—in fact we want to show that we can do primitive
recursion using minimization—so we need to be more clever.

Lemma 4.4. There is a function 𝛽 (d ,i ) such that for every sequence

a0 , . . . , an there is a number d , such that for every i ≤ n, 𝛽 (d ,i ) = ai .
Moreover, 𝛽 can be defined from the basic functions using just composi-
tion and regular minimization.

Think of d as coding the sequence ⟨a0 , . . . ,an ⟩, and 𝛽 (d ,i )

returning the i -th element. (Note that this “coding” does not use
the prower-of-primes coding we’re already familiar with!). The
lemma is fairly minimal; it doesn’t say we can concatenate se-
quences or append elements, or even that we can compute d from
a0 , . . . , an using functions definable by composition and regular
minimization. All it says is that there is a “decoding” function
such that every sequence is “coded.”
The use of the notation 𝛽 is Gödel’s. To repeat, the hard
part of proving the lemma is defining a suitable 𝛽 using the
seemingly restricted resources, i.e., using just composition and
minimization—however, we’re allowed to use addition, multipli-
cation, and 𝜒 = . There are various ways to prove this lemma, but
one of the cleanest is still Gödel’s original method, which used a
number-theoretic fact called the Chinese Remainder theorem.

Definition 4.5. Two natural numbers a and b are relatively prime

if their greatest common divisor is 1; in other words, they have
no other divisors in common.
82 CHAPTER 4. REPRESENTABILITY IN Q

Definition 4.6. a ≡ b mod c means c | (a −b), i.e., a and b have

the same remainder when divided by c .

Here is the Chinese Remainder theorem:

Theorem 4.7. Suppose x 0 , . . . , x n are (pairwise) relatively prime. Let

y 0 , . . . , y n be any numbers. Then there is a number z such that

z ≡ y0 mod x 0
z ≡ y1 mod x 1
..
.
z ≡ yn mod x n .

Here is how we will use the Chinese Remainder theorem: if

x 0 , . . . , x n are bigger than y 0 , . . . , y n respectively, then we can
take z to code the sequence ⟨y 0 , . . . , y n ⟩. To recover y i , we need
only divide z by x i and take the remainder. To use this coding,
we will need to find suitable values for x 0 , . . . , x n .
A couple of observations will help us in this regard. Given y 0 ,
. . . , y n , let
j = max(n, y 0 , . . . , y n ) + 1,
and let

x0 = 1 + j !
x1 = 1 + 2 · j !
x2 = 1 + 3 · j !
..
.
x n = 1 + (n + 1) · j !

Then two things are true:

1. x 0 , . . . , x n are relatively prime.

2. For each i , y i < x i .

83 4.3. THE BETA FUNCTION LEMMA

To see that (1) is true, note that if p is a prime number and p | x i

and p | x k , then p | 1 + (i + 1) j ! and p | 1 + (k + 1) j !. But then p
divides their difference,

(1 + (i + 1) j !) − (1 + (k + 1) j !) = (i − k ) j !.

Since p divides 1 + (i + 1) j !, it can’t divide j ! as well (otherwise,

the first division would leave a remainder of 1). So p divides i −k ,
since p divides (i − k ) j !. But |i − k | is at most n, and we have
chosen j > n, so this implies that p | j !, again a contradiction.
So there is no prime number dividing both x i and x k . Clause (2)
is easy: we have y i < j < j ! < x i .
Now let us prove the 𝛽 function lemma. Remember that we
can use 0, successor, plus, times, 𝜒 = , projections, and any func-
tion defined from them using composition and minimization ap-
plied to regular functions. We can also use a relation if its charac-
teristic function is so definable. As before we can show that these
relations are closed under boolean combinations and bounded
quantification; for example:

1. not(x) = 𝜒 = (x, 0)

2. (min x ≤ z ) R (x, y) = 𝜇x (R (x, y) ∨ x = z )

3. (∃x ≤ z ) R (x, y) ⇔ R ((min x ≤ z ) R (x, y), y)

We can then show that all of the following are also definable
without primitive recursion:

1. The pairing function, J (x, y) = 21 [(x + y) (x + y + 1)] + x

2. Projections

K (z ) = (min x ≤ q ) (∃y ≤ z [z = J (x, y)])

and

L(z ) = (min y ≤ q ) (∃x ≤ z [z = J (x, y)]).

84 CHAPTER 4. REPRESENTABILITY IN Q

3. x < y

4. x | y

5. The function rem(x, y) which returns the remainder when

y is divided by x

Now define

𝛽 ∗ (d0 ,d1 ,i ) = rem(1 + (i + 1)d1 ,d0 )

and
𝛽 (d ,i ) = 𝛽 ∗ (K (d ),L(d ),i ).
This is the function we need. Given a0 , . . . ,an , as above, let

j = max(n,a0 , . . . ,an ) + 1,

and let d1 = j !. By the observations above, we know that 1 +

d1 , 1 + 2d1 , . . . , 1 + (n + 1)d1 are relatively prime and all are bigger
than a0 , . . . ,an . By the Chinese Remainder theorem there is a
value d0 such that for each i ,

d0 ≡ ai mod (1 + (i + 1)d1 )

and so (because d1 is greater than ai ),

ai = rem(1 + (i + 1)d1 ,d0 ).

Let d = J (d0 ,d1 ). Then for each i ≤ n, we have

𝛽 (d ,i ) = 𝛽 ∗ (d0 ,d1 ,i )
= rem(1 + (i + 1)d1 ,d0 )
= ai

which is what we need. This completes the proof of the 𝛽 -function

lemma.
85 4.4. SIMULATING PRIMITIVE RECURSION

4.4 Simulating Primitive Recursion

Now we can show that definition by primitive recursion can be
“simulated” by regular minimization using the beta function.
Suppose we have f (x⃗ ) and g (x⃗ , y, z ). Then the function h (x, z⃗)
defined from f and g by primitive recursion is
h (x⃗ , y) = f (z⃗)
h (x⃗ , y + 1) = g (x⃗ , y,h (x⃗ , y)).
We need to show that h can be defined from f and g using just
composition and regular minimization, using the basic functions
and functions defined from them using composition and regular
minimization (such as 𝛽 ).

Lemma 4.8. If h can be defined from f and g using primitive recur-

sion, it can be defined from f , g , the functions zero, succ, Pin , add,
mult, 𝜒 = , using composition and regular minimization.

Proof. First, define an auxiliary function ĥ (x⃗ , y) which returns the

least number d such that d codes a sequence which satisfies
1. (d )0 = f (x⃗ ), and
2. for each i < y, (d )i +1 = g (x⃗ ,i , (d )i ),
where now (d )i is short for 𝛽 (d ,i ). In other words, ĥ returns the
sequence ⟨h (x⃗ , 0),h (x⃗ , 1), . . . ,h (x⃗ , y)⟩. We can write ĥ as
ĥ (x⃗ , y) = 𝜇d (𝛽 (d , 0) = f (x⃗ )∧(∀i < y) 𝛽 (d ,i +1) = g (x⃗ ,i , 𝛽 (d ,i )).
Note: no primitive recursion is needed here, just minimization.
The function we minimize is regular because of the beta function
lemma Lemma 4.4.
But now we have
h (x⃗ , y) = 𝛽 (ĥ (x⃗ , y), y),
so h can be defined from the basic functions using just composi-
tion and regular minimization. □
86 CHAPTER 4. REPRESENTABILITY IN Q

4.5 Basic Functions are Representable in Q

First we have to show that all the basic functions are representable
in Q . In the end, we need to show how to assign to each k -ary
basic function f (x 0 , . . . ,x k −1 ) a formula A f (x 0 , . . . ,x k −1 , y) that
represents it.
We will be able to represent zero, successor, plus, times, the
characteristic function for equality, and projections. In each case,
the appropriate representing function is entirely straightforward;
for example, zero is represented by the formula y = 0, successor is
represented by the formula x 0′ = y, and addition is represented by
the formula (x 0 + x 1 ) = y. The work involves showing that Q can
prove the relevant sentences; for example, saying that addition is
represented by the formula above involves showing that for every
pair of natural numbers m and n, Q proves

n + m = n + m and
∀y ((n + m) = y → y = n + m).

Proposition 4.9. The zero function zero(x) = 0 is represented in Q

by y = 0.

Proposition 4.10. The successor function succ(x) = x + 1 is repre-

sented in Q by y = x ′.

Proposition 4.11. The projection function Pin (x 0 , . . . ,x n−1 ) = x i is

represented in Q by y = x i .

Proposition 4.12. The characteristic function of =,

{︄
1 if x 0 = x 1
𝜒 = (x 0 ,x 1 ) =
0 other wise
87 4.5. BASIC FUNCTIONS ARE REPRESENTABLE IN Q

is represented in Q by

(x 0 = x 1 ∧ y = 1) ∨ (x 0 ≠ x 1 ∧ y = 0).

The proof requires the following lemma.

Lemma 4.13. Given natural numbers n and m, if n ≠ m, then Q ⊢

n ≠ m.

Proof. Use induction on n to show that for every m, if n ≠ m, then

Q ⊢ n ≠ m.
In the base case, n = 0. If m is not equal to 0, then m = k + 1
for some natural number k . We have an axiom that says ∀x 0 ≠ x ′.
′
By a quantifier axiom, replacing x by k , we can conclude 0 ≠ k .
′
But k is just m.
In the induction step, we can assume the claim is true for n,
and consider n + 1. Let m be any natural number. There are
two possibilities: either m = 0 or for some k we have m = k + 1.
The first case is handled as above. In the second case, suppose
n + 1 ≠ k + 1. Then n ≠ k . By the induction hypothesis for n we
have Q ⊢ n ≠ k . We have an axiom that says ∀x ∀y x ′ = y ′ → x =
′
y. Using a quantifier axiom, we have n ′ = k → n = k . Using
′
propositional logic, we can conclude, in Q , n ≠ k → n ′ ≠ k .
′
Using modus ponens, we can conclude n ′ ≠ k , which is what we
′
want, since k is m. □

Note that the lemma does not say much: in essence it says
that Q can prove that different numerals denote different objects.
For example, Q proves 0′′ ≠ 0′′′. But showing that this holds in
general requires some care. Note also that although we are using
induction, it is induction outside of Q .

Proof of Proposition 4.12. If n = m, then n and m are the same

term, and 𝜒 = (n,m) = 1. But Q ⊢ (n = m ∧ 1 = 1), so it proves
A= (n,m, 1). If n ≠ m, then 𝜒 = (n,m) = 0. By Lemma 4.13, Q ⊢
n ≠ m and so also (n ≠ m ∧ 0 = 0). Thus Q ⊢ A= (n,m, 0).
88 CHAPTER 4. REPRESENTABILITY IN Q

For the second part, we also have two cases. If n = m, we have

to show that Q ⊢ ∀y (A= (n,m, y) → y = 1). Arguing informally,
suppose A= (n,m, y), i.e.,

(n = n ∧ y = 1) ∨ (n ≠ n ∧ y = 0)

The left disjunct implies y = 1 by logic; the right contradicts n = n

which is provable by logic.
Suppose, on the other hand, that n ≠ m. Then A= (n,m, y) is

(n = m ∧ y = 1) ∨ (n ≠ m ∧ y = 0)

Here, the left disjunct contradicts n ≠ m, which is provable in Q

by Lemma 4.13; the right disjunct entails y = 0. □

Proposition 4.14. The addition function add(x 0 ,x 1 ) = x 0 + x 1 is

represented in Q by
y = (x 0 + x 1 ).

Lemma 4.15. Q ⊢ (n + m) = n + m

Proof. We prove this by induction on m. If m = 0, the claim is

that Q ⊢ (n + 0) = n. This follows by axiom Q 4 . Now suppose the
claim for m; let’s prove the claim for m + 1, i.e., prove that Q ⊢
(n + m + 1) = n + m + 1. Note that m + 1 is just m ′, and n + m + 1
is just n + m ′. By axiom Q 5 , Q ⊢ (n +m ′) = (n +m) ′. By induction
hypothesis, Q ⊢ (n + m) = n + m. So Q ⊢ (n + m ′) = n + m ′. □

Proof of Proposition 4.14. The formula Aadd (x 0 ,x 1 , y) representing

add is y = (x 0 + x 1 ). First we show that if add(n,m) = k , then
Q ⊢ Aadd (n,m,k ), i.e., Q ⊢ k = (n + m). But since k = n + m, k
just is n + m, and we’ve shown in Lemma 4.15 that Q ⊢ (n + m) =
n + m.
We also have to show that if add(n,m) = k , then

Q ⊢ ∀y (Aadd (n,m, y) → y = k ).
89 4.6. COMPOSITION IS REPRESENTABLE IN Q

Suppose we have (n + m) = y. Since

Q ⊢ (n + m) = n + m,

we can replace the left side with n + m and get n + m = y, for

arbitrary y. □

Proposition 4.16. The multiplication function mult(x 0 ,x 1 ) = x 0 ·x 1

is represented in Q by
y = (x 0 × x 1 ).

Proof. Exercise. □

Lemma 4.17. Q ⊢ (n × m) = n · m

Proof. Exercise. □

Recall that we use × for the function symbol of the language

of arithmetic, and · for the ordinary multiplication operation on
numbers. So · can appear between expressions for numbers (such
as in m · n) while × appears only between terms of the language
of arithmetic (such as in (m × n)). Even more confusingly, + is
used for both the function symbol and the addition operation.
When it appears between terms—e.g., in (n + m)—it is the 2-
place function symbol of the language of arithmetic, and when it
appears between numbers—e.g., in n +m—it is the addition oper-
ation. This includes the case n + m: this is the standard numeral
corresponding to the number n + m.

4.6 Composition is Representable in Q

Suppose h is defined by

h (x 0 , . . . ,xl −1 ) = f (g 0 (x 0 , . . . ,xl −1 ), . . . , g k −1 (x 0 , . . . ,xl −1 )).

90 CHAPTER 4. REPRESENTABILITY IN Q

where we have already found formulas A f ,A g 0 , . . . ,A gk −1 repre-

senting the functions f , and g 0 , . . . , g k −1 , respectively. We have
to find a formula Ah representing h.
Let’s start with a simple case, where all functions are 1-
place, i.e., consider h (x) = f (g (x)). If A f (y, z ) represents f ,
and A g (x, y) represents g , we need a formula Ah (x, z ) that repre-
sents h. Note that h (x) = z iff there is a y such that both z = f (y)
and y = g (x). (If h (x) = z , then g (x) is such a y; if such a y exists,
then since y = g (x) and z = f (y), z = f (g (x)).) This suggests
that ∃y (A g (x, y) ∧ A f (y, z )) is a good candidate for Ah (x, z ). We
just have to verify that Q proves the relevant formulas.

Proposition 4.18. If h (n) = m, then Q ⊢ Ah (n,m).

Proof. Suppose h (n) = m, i.e., f (g (n)) = m. Let k = g (n). Then

Q ⊢ A g (n,k )

since A g represents g , and

Q ⊢ A f (k ,m)

since A f represents f . Thus,

Q ⊢ A g (n,k ) ∧ A f (k ,m)

and consequently also

Q ⊢ ∃y (A g (n, y) ∧ A f (y,m)),

i.e., Q ⊢ Ah (n,m). □
91 4.7. REGULAR MINIMIZATION IS REPRESENTABLE IN Q

Proposition 4.19. If h (n) = m, then Q ⊢ ∀z (Ah (n, z ) → z = m).

Proof. Suppose h (n) = m, i.e., f (g (n)) = m. Let k = g (n). Then

Q ⊢ ∀y (A g (n, y) → y = k )

since A g represents g , and

Q ⊢ ∀z (A f (k , z ) → z = m)
since A f represents f . Using just a little bit of logic, we can show
that also

Q ⊢ ∀z (∃y (A g (n, y) ∧ A f (y, z )) → z = m).

i.e., Q ⊢ ∀y (Ah (n, y) → y = m). □

The same idea works in the more complex case where f

and g i have arity greater than 1.

Proposition 4.20. If A f (y 0 , . . . , y k −1 , z ) represents f (y 0 , . . . , y k −1 )

in Q , and A gi (x 0 , . . . ,xl −1 , y) represents g i (x 0 , . . . ,xl −1 ) in Q , then

∃y 0 , . . . ∃y k −1 (A g 0 (x 0 , . . . ,xl −1 , y 0 ) ∧ · · · ∧
A gk −1 (x 0 , . . . ,xl −1 , y k −1 ) ∧ A f (y 0 , . . . , y k −1 , z ))

represents

h (x 0 , . . . ,xl −1 ) = f (g 0 (x 0 , . . . ,xl −1 ), . . . , g k −1 (x 0 , . . . ,xl −1 )).

Proof. Exercise. □

4.7 Regular Minimization is Representable

in Q
Let’s consider unbounded search. Suppose g (x, z ) is regular and
representable in Q , say by the formula A g (x, z , y). Let f be de-
fined by f (z ) = 𝜇x [g (x, z ) = 0]. We would like to find a for-
mula A f (z , y) representing f . The value of f (z ) is that number x
92 CHAPTER 4. REPRESENTABILITY IN Q

which (a) satisfies g (x, z ) = 0 and (b) is the least such, i.e., for
any w < x, g (w, z ) ≠ 0. So the following is a natural choice:

A f (z , y) ≡ A g (y, z , 0) ∧ ∀w (w < y → ¬A g (w, z , 0)).

In the general case, of course, we would have to replace z with

z 0 , . . . , zk .
The proof, again, will involve some lemmas about things Q
is strong enough to prove.

Lemma 4.21. For every constant symbol a and every natural num-
ber n,
Q ⊢ (a ′ + n) = (a + n) ′ .

Proof. The proof is, as usual, by induction on n. In the base case,

n = 0, we need to show that Q proves (a ′ + 0) = (a + 0) ′. But we
have:

Q ⊢ (a ′ + 0) = a ′ by axiom Q 4 (4.1)
Q ⊢ (a + 0) = a by axiom Q 4 (4.2)
′ ′
Q ⊢ (a + 0) = a by eq. (4.2) (4.3)
′ ′
Q ⊢ (a + 0) = (a + 0) by eq. (4.1) and eq. (4.3)

In the induction step, we can assume that we have shown that

Q ⊢ (a ′ + n) = (a + n) ′. Since n + 1 is n ′, we need to show that
Q proves (a ′ + n ′) = (a + n ′) ′. We have:

Q ⊢ (a ′ + n ′) = (a ′ + n) ′ by axiom Q 5 (4.4)
′ ′ ′ ′
Q ⊢ (a + n ) = (a + n ) inductive hypothesis (4.5)
′ ′ ′ ′
Q ⊢ (a + n) = (a + n ) by eq. (4.4) and eq. (4.5). □

It is again worth mentioning that this is weaker than saying

that Q proves ∀x ∀y (x ′ + y) = (x + y) ′. Although this sentence is
true in N, Q does not prove it.
93 4.7. REGULAR MINIMIZATION IS REPRESENTABLE IN Q

Lemma 4.22. Q ⊢ ∀x ¬x < 0.

Proof. We give the proof informally (i.e., only giving hints as to

how to construct the formal derivation).
We have to prove ¬a < 0 for an arbitrary a. By the definition
of <, we need to prove ¬∃y (y ′ + a) = 0 in Q . We’ll assume
∃y (y ′ + a) = 0 and prove a contradiction. Suppose (b ′ + a) = 0.
Using Q 3 , we have that a = 0 ∨ ∃y a = y ′. We distinguish cases.
Case 1: a = 0 holds. From (b ′ + a) = 0, we have (b ′ + 0) = 0.
By axiom Q 4 of Q , we have (b ′ + 0) = b ′, and hence b ′ = 0. But
by axiom Q 2 we also have b ′ ≠ 0, a contradiction.
Case 2: For some c , a = c ′. But then we have (b ′ + c ′) = 0. By
axiom Q 5 , we have (b ′ + c ) ′ = 0, again contradicting axiom Q 2 .□

Lemma 4.23. For every natural number n,

Q ⊢ ∀x (x < n + 1 → (x = 0 ∨ · · · ∨ x = n)).

Proof. We use induction on n. Let us consider the base case,

when n = 0. In that case, we need to show a < 1 → a = 0, for
arbitrary a. Suppose a < 1. Then by the defining axiom for <,
we have ∃y (y ′ + a) = 0′ (since 1 ≡ 0′).
Suppose b has that property, i.e., we have (b ′ + a) = 0′. We
need to show a = 0. By axiom Q 3 , we have either a = 0 or that
there is a c such that a = c ′. In the former case, there is nothing to
show. So suppose a = c ′. Then we have (b ′ +c ′) = 0′. By axiom Q 5
of Q , we have (b ′ +c ) ′ = 0′. By axiom Q 1 , we have (b ′ +c ) = 0. But
this means, by axiom Q 8 , that c < 0, contradicting Lemma 4.22.
Now for the inductive step. We prove the case for n + 1, as-
suming the case for n. So suppose a < n + 2. Again using Q 3 we
can distinguish two cases: a = 0 and for some b, a = c ′. In the
first case, a = 0 ∨ · · · ∨ a = n + 1 follows trivially. In the second
′
case, we have c ′ < n + 2, i.e., c ′ < n + 1 . By axiom Q 8 , for some
′ ′
d , (d ′ + c ′) = n + 1 . By axiom Q 5 , (d ′ + c ) ′ = n + 1 . By axiom Q 1 ,
94 CHAPTER 4. REPRESENTABILITY IN Q

(d ′ + c ) = n + 1, and so c < n + 1 by axiom Q 8 . By inductive hy-

pothesis, c = 0∨ · · · ∨c = n. From this, we get c ′ = 0′ ∨ · · · ∨c ′ = n ′
by logic, and so a = 1 ∨ · · · ∨ a = n + 1 since a = c ′. □

Lemma 4.24. For every natural number m,

Q ⊢ ∀y ((y < m ∨ m < y) ∨ y = m).

Proof. By induction on m. First, consider the case m = 0. Q ⊢

∀y (y = 0 ∨ ∃z y = z ′) by Q 3 . Let a be arbitrary. Then either
a = 0 or for some b, a = b ′. In the former case, we also have
(a < 0 ∨ 0 < a) ∨ a = 0. But if a = b ′, then (b ′ + 0) = (a + 0) by the
logic of =. By Q 4 , (a + 0) = a, so we have (b ′ + 0) = a, and hence
∃z (z ′ + 0) = a. By the definition of < in Q 8 , 0 < a. If 0 < a, then
also (0 < a ∨ a < 0) ∨ a = 0.
Now suppose we have

Q ⊢ ∀y ((y < m ∨ m < y) ∨ y = m)

and we want to show

Q ⊢ ∀y ((y < m + 1 ∨ m + 1 < y) ∨ y = m + 1)

Let a be arbitrary. By Q 3 , either a = 0 or for some b, a = b ′. In

the first case, we have m ′ + a = m + 1 by Q 4 , and so a < m + 1 by
Q 8.
Now consider the second case, a = b ′. By the induction hy-
pothesis, (b < m ∨ m < b) ∨ b = m.
The first disjunct b < m is equivalent (by Q 8 ) to ∃z (z ′ +b) = m.
Suppose c has this property. If (c ′ + b) = m, then also (c ′ + b) ′ =
m ′. By Q 5 , (c ′ + b) ′ = (c ′ + b ′). Hence, (c ′ + b ′) = m ′. We get
∃u (u ′ +b ′) = m + 1 by existentially generalizing on c ′ and keeping
in mind that m ′ ≡ m + 1. Hence, if b < m then b ′ < m + 1 and so
a < m + 1.
Now suppose m < b, i.e., ∃z (z ′ + m) = b. Suppose c is such
a z , i.e., (c ′ + m) = b. By logic, (c ′ + m) ′ = b ′. By Q 5 , (c ′ + m ′) = b ′.
Since a = b ′ and m ′ ≡ m + 1, (c ′ + m + 1) = a. By Q 8 , m + 1 < a.
95 4.7. REGULAR MINIMIZATION IS REPRESENTABLE IN Q

Finally, assume b = m. Then, by logic, b ′ = m ′, and so a =

m + 1.
Hence, from each disjunct of the case for m and b, we can
obtain the corresponding disjunct for for m + 1 and a. □

Proposition 4.25. If A g (x, z , y) represents g (x, z ) in Q , then

A f (z , y) ≡ A g (y, z , 0) ∧ ∀w (w < y → ¬A g (w, z , 0)).

represents f (z ) = 𝜇x [g (x, z ) = 0].

Proof. First we show that if f (n) = m, then Q ⊢ A f (n,m), i.e.,

Q ⊢ A g (m,n, 0) ∧ ∀w (w < m → ¬A g (w,n, 0)).

Since A g (x, z , y) represents g (x, z ) and g (m,n) = 0 if f (n) = m,

we have

Q ⊢ A g (m,n, 0).

If f (n) = m, then for every k < m, g (k ,n) ≠ 0. So

Q ⊢ ¬A g (k ,n, 0).

We get that

Q ⊢ ∀w (w < m → ¬A g (w,n, 0)). (4.6)

by Lemma 4.22 in case m = 0 and by Lemma 4.23 otherwise.

Now let’s show that if f (n) = m, then Q ⊢ ∀y (A f (n, y) →
y = m). We again sketch the argument informally, leaving the
formalization to the reader.
Suppose A f (n,b). From this we get (a) A g (b,n, 0) and (b)
∀w (w < b → ¬A g (w,n, 0)). By Lemma 4.24, (b < m ∨ m <
b) ∨ b = m. We’ll show that both b < m and m < b leads to a
contradiction.
96 CHAPTER 4. REPRESENTABILITY IN Q

If m < b, then ¬A g (m,n, 0) from (b). But m = f (n), so

g (m,n) = 0, and so Q ⊢ A g (m,n, 0) since A g represents g . So
we have a contradiction.
Now suppose b < m. Then since Q ⊢ ∀w (w < m →
¬A g (w,n, 0)) by eq. (4.6), we get ¬A g (b,n, 0). This again con-
tradicts (a). □

4.8 Computable Functions are

Representable in Q
Theorem 4.26. Every computable function is representable in Q .

Proof. For definiteness, and using the Church-Turing Thesis, let’s

say that a function is computable iff it is general recursive. The
general recursive functions are those which can be defined from
the zero function zero, the successor function succ, and the pro-
jection function Pin using composition, primitive recursion, and
regular minimization. By Lemma 4.8, any function h that can be
defined from f and g can also be defined using composition and
regular minimization from f , g , and zero, succ, Pin , add, mult,
𝜒 = . Consequently, a function is general recursive iff it can be de-
fined from zero, succ, Pin , add, mult, 𝜒 = using composition and
regular minimization.
We’ve furthermore shown that the basic functions in question
are representable in Q (Propositions 4.9 to 4.12, 4.14 and 4.16),
and that any function defined from representable functions by
composition or regular minimization (Proposition 4.20, Propo-
sition 4.25) is also representable. Thus every general recursive
function is representable in Q . □

We have shown that the set of computable functions can be

characterized as the set of functions representable in Q . In fact,
the proof is more general. From the definition of representability,
it is not hard to see that any theory extending Q (or in which one
can interpret Q ) can represent the computable functions. But,
97 4.9. REPRESENTING RELATIONS

conversely, in any proof system in which the notion of proof is

computable, every representable function is computable. So, for
example, the set of computable functions can be characterized
as the set of functions representable in Peano arithmetic, or even
Zermelo-Fraenkel set theory. As Gödel noted, this is somewhat
surprising. We will see that when it comes to provability, ques-
tions are very sensitive to which theory you consider; roughly,
the stronger the axioms, the more you can prove. But across a
wide range of axiomatic theories, the representable functions are
exactly the computable ones; stronger theories do not represent
more functions as long as they are axiomatizable.

4.9 Representing Relations

Let us say what it means for a relation to be representable.

Definition 4.27. A relation R (x 0 , . . . ,x k ) on the natural num-

bers is representable in Q if there is a formula AR (x 0 , . . . ,x k ) such
that whenever R (n 0 , . . . ,nk ) is true, Q proves AR (n 0 , . . . ,nk ), and
whenever R (n 0 , . . . ,nk ) is false, Q proves ¬AR (n 0 , . . . ,nk ).

Theorem 4.28. A relation is representable in Q if and only if it is

computable.

Proof. For the forwards direction, suppose R (x 0 , . . . ,x k ) is repre-

sented by the formula AR (x 0 , . . . ,x k ). Here is an algorithm for
computing R: on input n0 , . . . , nk , simultaneously search for a
proof of AR (n 0 , . . . ,nk ) and a proof of ¬AR (n 0 , . . . ,nk ). By our
hypothesis, the search is bound to find one or the other; if it is
the first, report “yes,” and otherwise, report “no.”
In the other direction, suppose R (x 0 , . . . ,x k ) is computable.
By definition, this means that the function 𝜒 R (x 0 , . . . ,x k ) is
computable. By Theorem 4.2, 𝜒 R is represented by a for-
mula, say A 𝜒 R (x 0 , . . . ,x k , y). Let AR (x 0 , . . . ,x k ) be the formula
98 CHAPTER 4. REPRESENTABILITY IN Q

A 𝜒 R (x 0 , . . . ,x k , 1). Then for any n0 , . . . , nk , if R (n 0 , . . . ,nk )

is true, then 𝜒 R (n0 , . . . ,nk ) = 1, in which case Q proves
A 𝜒 R (n 0 , . . . ,nk , 1), and so Q proves AR (n 0 , . . . ,nk ). On the other
hand, if R (n0 , . . . ,nk ) is false, then 𝜒 R (n 0 , . . . ,nk ) = 0. This
means that Q proves

∀y (A 𝜒 R (n 0 , . . . ,nk , y) → y = 0).

Since Q proves 0 ≠ 1, Q proves ¬A 𝜒 R (n 0 , . . . ,nk , 1), and so it

proves ¬AR (n 0 , . . . ,nk ). □

4.10 Undecidability
We call a theory T undecidable if there is no computational pro-
cedure which, after finitely many steps and unfailingly, provides
a correct answer to the question “does T prove A?” for any
sentence A in the language of T. So Q would be decidable iff
there were a computational procedure which decides, given a sen-
tence A in the language of arithmetic, whether Q ⊢ A or not. We
can make this more precise by asking: Is the relation ProvQ (y),
which holds of y iff y is the Gödel number of a sentence provable
in Q , recursive? The answer is: no.

Theorem 4.29. Q is undecidable, i.e., the relation

ProvQ (y) ⇔ Sent(y) ∧ ∃x Prf Q (x, y)

is not recursive.

Proof. Suppose it were. Then we could solve the halting prob-

lem as follows: Given e and n, we know that 𝜑e (n) ↓ iff there
is an s such that T (e ,n,s ), where T is Kleene’s predicate from
Theorem 2.28. Since T is primitive recursive it is representable
in Q by a formula BT , that is, Q ⊢ BT (e ,n,s ) iff T (e ,n,s ). If
Q ⊢ BT (e ,n,s ) then also Q ⊢ ∃y BT (e ,n, y). If no such s exists,
then Q ⊢ ¬BT (e ,n,s ) for every s . But Q is 𝜔-consistent, i.e.,
if Q ⊢ ¬A(n) for every n ∈ N, then Q ⊬ ∃y A(y). We know
99 4.10. UNDECIDABILITY

this because the axioms of Q are true in the standard model N.

So, Q ⊬ ∃y BT (e ,n, y). In other words, Q ⊢ ∃y BT (e ,n, y) iff
there is an s such that T (e ,n,s ), i.e., iff 𝜑e (n) ↓. From e and n we
can compute # ∃y BT (e ,n, y) # , let g (e ,n) be the primitive recursive
function which does that. So
{︄
1 if ProvQ (g (e ,n))
h (e ,n) =
0 otherwise.

This would show that h is recursive if ProvQ is. But h is not

recursive, by Theorem 2.29, so ProvQ cannot be either. □

Corollary 4.30. First-order logic is undecidable.

Proof. If first-order logic were decidable, provability in Q would

be as well, since Q ⊢ A iff ⊢ O → A, where O is the conjunction
of the axioms of Q . □

Summary
In order to show how theories like Q can “talk” about com-
putable functions—and especially about provability (via Gödel
numbers)—we established that Q represents all computable
functions. By “Q represents f (n)” we mean that there is a for-
mula A f (x, y) in LA which expresses that f (x) = y, and Q can
prove that it does. This, in turn, means that whenever f (n) = m,
then T ⊢ A f (n,m) and T ⊢ ∀y (A f (n, y) → y = m). (Here, n is the
standard numeral for n, i.e., the term 0′...′ with n ′s. The term n
picks out the number n in the standard model N, so it’s a conve-
nient way of representing the number n in LA .) To prove that Q
represents all computable functions we go back to the characteri-
zation of computable functions as those that can be defined from
zero, succ, and the projection functions, by composition, prim-
itive recursion, and regular minimization. While it is relatively
easy to prove that the basic functions are representable and that
100 CHAPTER 4. REPRESENTABILITY IN Q

functions defined by composition and regular minimization from

representable functions are also representable, primitive recur-
sion is harder. We showed that we can actually avoid definition
by primitive recursion, if we allow a few additional basic functions
(namely, addition, multiplication, and the characteristic function
of =). This required a beta function which allows us to deal with
sequences of numbers in a rudimentary way, and which can be
defined without using primitive recursion.

Problems
Problem 4.1. Prove that y = 0, y = x ′, and y = x i represent zero,
succ, and Pin , respectively.

Problem 4.2. Prove Lemma 4.17.

Problem 4.3. Use Lemma 4.17 to prove Proposition 4.16.

Problem 4.4. Using the proofs of Proposition 4.19 and Proposi-

tion 4.19 as a guide, carry out the proof of Proposition 4.20 in
detail.

Problem 4.5. Show that if R is representable in Q , so is 𝜒 R .

CHAPTER 5

Incompleteness
and Provability
5.1 Introduction
Hilbert thought that a system of axioms for a mathematical struc-
ture, such as the natural numbers, is inadequate unless it allows
one to derive all true statements about the structure. Combined
with his later interest in formal systems of deduction, this suggests
that he thought that we should guarantee that, say, the formal sys-
tems we are using to reason about the natural numbers is not only
consistent, but also complete, i.e., every statement in its language
is either derivable or its negation is. Gödel’s first incomplete-
ness theorem shows that no such system of axioms exists: there
is no complete, consistent, axiomatizable formal system for arith-
metic. In fact, no “sufficiently strong,” consistent, axiomatizable
mathematical theory is complete.
A more important goal of Hilbert’s, the centerpiece of his
program for the justification of modern (“classical”) mathemat-
ics, was to find finitary consistency proofs for formal systems rep-
resenting classical reasoning. With regard to Hilbert’s program,
then, Gödel’s second incompleteness theorem was a much bigger
blow. The second incompleteness theorem can be stated in vague
terms, like the first incompleteness theorem. Roughly speaking,

101
102 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

it says that no sufficiently strong theory of arithmetic can prove

its own consistency. We will have to take “sufficiently strong” to
include a little bit more than Q .
The idea behind Gödel’s original proof of the incompleteness
theorem can be found in the Epimenides paradox. Epimenides,
a Cretan, asserted that all Cretans are liars; a more direct form of
the paradox is the assertion “this sentence is false.” Essentially,
by replacing truth with derivability, Gödel was able to formalize
a sentence which, in a roundabout way, asserts that it itself is
not derivable. If that sentence were derivable, the theory would
then be inconsistent. Gödel showed that the negation of that
sentence is also not derivable from the system of axioms he was
considering. (For this second part, Gödel had to assume that the
theory T is what’s called “𝜔-consistent.” 𝜔-Consistency is related
to consistency, but is a stronger property. A few years after Gödel,
Rosser showed that assuming simple consistency of T is enough.)
The first challenge is to understand how one can construct a
sentence that refers to itself. For every formula A in the language
of Q , let ⌜A⌝ denote the numeral corresponding to #A# . Think
about what this means: A is a formula in the language of Q , #A#
is a natural number, and ⌜A⌝ is a term in the language of Q . So
every formula A in the language of Q has a name, ⌜A⌝, which is
a term in the language of Q ; this provides us with a conceptual
framework in which formulas in the language of Q can “say”
things about other formulas. The following lemma is known as
the fixed-point lemma.

Lemma 5.1. Let T be any theory extending Q , and let B (x) be any
formula with only the variable x free. Then there is a sentence A such
that T ⊢ A ↔ B (⌜A⌝).

The lemma asserts that given any property B (x), there is a

sentence A that asserts “B (x) is true of me,” and T “knows” this.
How can we construct such a sentence? Consider the follow-
ing version of the Epimenides paradox, due to Quine:
103 5.2. THE FIXED-POINT LEMMA

“Yields falsehood when preceded by its quotation”

yields falsehood when preceded by its quotation.

This sentence is not directly self-referential. It simply makes an

assertion about the syntactic objects between quotes, and, in do-
ing so, it is on par with sentences like

1. “Robert” is a nice name.

2. “I ran.” is a short sentence.

3. “Has three words” has three words.

But what happens when one takes the phrase “yields falsehood
when preceded by its quotation,” and precedes it with a quoted
version of itself? Then one has the original sentence! In short,
the sentence asserts that it is false.

5.2 The Fixed-Point Lemma

The fixed-point lemma says that for any formula B (x), there is
a sentence A such that T ⊢ A ↔ B (⌜A⌝), provided T extends Q .
In the case of the liar sentence, we’d want A to be equivalent
(provably in T) to “⌜A⌝ is false,” i.e., the statement that #A# is
the Gödel number of a false sentence. To understand the idea
of the proof, it will be useful to compare it with Quine’s infor-
mal gloss of A as, “‘yields a falsehood when preceded by its own
quotation’ yields a falsehood when preceded by its own quota-
tion.” The operation of taking an expression, and then forming a
sentence by preceding this expression by its own quotation may
be called diagonalizing the expression, and the result its diago-
nalization. So, the diagonalization of ‘yields a falsehood when
preceded by its own quotation’ is “‘yields a falsehood when pre-
ceded by its own quotation’ yields a falsehood when preceded by
its own quotation.” Now note that Quine’s liar sentence is not the
diagonalization of ‘yields a falsehood’ but of ‘yields a falsehood
104 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

when preceded by its own quotation.’ So the property being diag-

onalized to yield the liar sentence itself involves diagonalization!
In the language of arithmetic, we form quotations of a formula
with one free variable by computing its Gödel numbers and then
substituting the standard numeral for that Gödel number into
the free variable. The diagonalization of E (x) is E (n), where
n = # E (x) # . (From now on, let’s abbreviate # E (x) # as ⌜E (x)⌝.)
So if B (x) is “is a falsehood,” then “yields a falsehood if pre-
ceded by its own quotation,” would be “yields a falsehood when
applied to the Gödel number of its diagonalization.” If we had a
symbol di ag for the function diag(n) which computes the Gödel
number of the diagonalization of the formula with Gödel num-
ber n, we could write E (x) as B (di ag(x)). And Quine’s version
of the liar sentence would then be the diagonalization of it, i.e.,
E (⌜E (x)⌝) or B (di ag(⌜B (di ag(x))⌝)). Of course, B (x) could
now be any other property, and the same construction would
work. For the incompleteness theorem, we’ll take B (x) to be “x
is not derivable in T.” Then E (x) would be “yields a sentence
not derivable in T when applied to the Gödel number of its di-
agonalization.”
To formalize this in T, we have to find a way to formalize
diag. The function diag(n) is computable, in fact, it is primitive
recursive: if n is the Gödel number of a formula E (x), diag(n)
returns the Gödel number of E (⌜E (x)⌝). (Recall, ⌜E (x)⌝ is the
standard numeral of the Gödel number of E (x), i.e., # E (x) # ).
If di ag were a function symbol in T representing the function
diag, we could take A to be the formula B (di ag(⌜B (di ag(x))⌝)).
Notice that
diag( # B (di ag(x)) # ) = # B (di ag(⌜B (di ag(x))⌝)) #
= #A# .
Assuming T can derive
di ag(⌜B (di ag(x))⌝) = ⌜A⌝,
it can derive B (di ag(⌜B (di ag(x))⌝))↔B (⌜A⌝). But the left hand
side is, by definition, A.
105 5.2. THE FIXED-POINT LEMMA

Of course, di ag will in general not be a function symbol of

T, and certainly is not one of Q . But, since diag is computable,
it is representable in Q by some formula D diag (x, y). So instead of
writing B (di ag(x)) we can write ∃y (D diag (x, y) ∧ B (y)). Other-
wise, the proof sketched above goes through, and in fact, it goes
through already in Q .

Lemma 5.2. Let B (x) be any formula with one free variable x. Then
there is a sentence A such that Q ⊢ A ↔ B (⌜A⌝).

Proof. Given B (x), let E (x) be the formula ∃y (D diag (x, y) ∧ B (y))
and let A be its diagonalization, i.e., the formula E (⌜E (x)⌝).
Since D diag represents diag, and diag( # E (x) # ) = #A# , Q can
derive

D diag (⌜E (x)⌝, ⌜A⌝) (5.1)

∀y (D diag (⌜E (x)⌝, y) → y = ⌜A⌝). (5.2)

Now we show that Q ⊢ A ↔ B (⌜A⌝). We argue informally, using

just logic and facts derivable in Q .
First, suppose A, i.e., E (⌜E (x)⌝). Going back to the definition
of E (x), we see that E (⌜E (x)⌝) just is

∃y (D diag (⌜E (x)⌝, y) ∧ B (y)).

Consider such a y. Since D diag (⌜E (x)⌝, y), by eq. (5.2), y = ⌜A⌝.
So, from B (y) we have B (⌜A⌝).
Now suppose B (⌜A⌝). By eq. (5.1), we
have D diag (⌜E (x)⌝, ⌜A⌝) ∧ B (⌜A⌝). It follows that
∃y (D diag (⌜E (x)⌝, y) ∧ B (y)). But that’s just E (⌜E ⌝), i.e., A. □

You should compare this to the proof of the fixed-point lemma

in computability theory. The difference is that here we want to
define a statement in terms of itself, whereas there we wanted to
define a function in terms of itself; this difference aside, it is really
the same idea.
106 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

5.3 The First Incompleteness Theorem

We can now describe Gödel’s original proof of the first incom-
pleteness theorem. Let T be any computably axiomatized theory
in a language extending the language of arithmetic, such that T
includes the axioms of Q . This means that, in particular, T rep-
resents computable functions and relations.
We have argued that, given a reasonable coding of formu-
las and proofs as numbers, the relation PrfT (x, y) is computable,
where PrfT (x, y) holds if and only if x is the Gödel number of
a derivation of the formula with Gödel number y in T. In fact,
for the particular theory that Gödel had in mind, Gödel was able
to show that this relation is primitive recursive, using the list of
45 functions and relations in his paper. The 45th relation, xBy,
is just PrfT (x, y) for his particular choice of T. Remember that
where Gödel uses the word “recursive” in his paper, we would
now use the phrase “primitive recursive.”
Since PrfT (x, y) is computable, it is representable in T. We
will use PrfT (x, y) to refer to the formula that represents it. Let
ProvT (y) be the formula ∃x PrfT (x, y). This describes the 46th
relation, Bew(y), on Gödel’s list. As Gödel notes, this is the only
relation that “cannot be asserted to be recursive.” What he prob-
ably meant is this: from the definition, it is not clear that it is
computable; and later developments, in fact, show that it isn’t.
Let T be an axiomatizable theory containing Q . Then
PrfT (x, y) is decidable, hence representable in Q by a for-
mula PrfT (x, y). Let ProvT (y) be the formula we described above.
By the fixed-point lemma, there is a formula G T such that Q (and
hence T) derives

G T ↔ ¬ProvT (⌜G T ⌝). (5.3)

Note that G T says, in essence, “G T is not derivable in T.”

107 5.3. THE FIRST INCOMPLETENESS THEOREM

Lemma 5.3. If T is a consistent, axiomatizable theory extending Q ,

then T ⊬ G T .

Proof. Suppose T derives G T . Then there is a derivation, and

so, for some number m, the relation PrfT (m, #G T # ) holds. But
then Q derives the sentence PrfT (m, ⌜G T ⌝). So Q derives
∃x PrfT (x, ⌜G T ⌝), which is, by definition, ProvT (⌜G T ⌝). By
eq. (5.3), Q derives ¬G T , and since T extends Q , so does T.
We have shown that if T derives G T , then it also derives ¬G T ,
and hence it would be inconsistent. □

Definition 5.4. A theory T is 𝜔-consistent if the following holds:

if ∃x A(x) is any sentence and T derives ¬A(0), ¬A(1), ¬A(2),
. . . then T does not prove ∃x A(x).

Note that every 𝜔-consistent theory is also consistent. This

follows simply from the fact that if T is inconsistent, then T ⊢ A
for every A. In particular, if T is inconsistent, it derives both
¬A(n) for every n and also derives ∃x A(x). So, if T is inconsis-
tent, it is 𝜔-inconsistent. By contraposition, if T is 𝜔-consistent,
it must be consistent.

Lemma 5.5. If T is an 𝜔-consistent, axiomatizable theory extend-

ing Q , then T ⊬ ¬G T .

Proof. We show that if T derives ¬G T , then it is 𝜔-inconsistent.

Suppose T derives ¬G T . If T is inconsistent, it is 𝜔-inconsistent,
and we are done. Otherwise, T is consistent, so it does not derive
G T by Lemma 5.3. Since there is no derivation of G T in T, Q
derives

¬PrfT (0, ⌜G T ⌝), ¬PrfT (1, ⌜G T ⌝), ¬PrfT (2, ⌜G T ⌝), . . .

and so does T. On the other hand, by eq. (5.3), ¬G T is equivalent

to ∃x PrfT (x, ⌜G T ⌝). So T is 𝜔-inconsistent. □
108 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

Theorem 5.6. Let T be any 𝜔-consistent, axiomatizable theory ex-

tending Q . Then T is not complete.

Proof. If T is 𝜔-consistent, it is consistent, so T ⊬ G T by

Lemma 5.3. By Lemma 5.5, T ⊬ ¬G T . This means that T is
incomplete, since it derives neither G T nor ¬G T . □

5.4 Rosser’s Theorem

Can we modify Gödel’s proof to get a stronger result, replacing
“𝜔-consistent” with simply “consistent”? The answer is “yes,”
using a trick discovered by Rosser. Rosser’s trick is to use a
“modified” derivability predicate RProvT (y) instead of ProvT (y).

Theorem 5.7. Let T be any consistent, axiomatizable theory extend-

ing Q . Then T is not complete.

Proof. Recall that ProvT (y) is defined as ∃x PrfT (x, y), where
PrfT (x, y) represents the decidable relation which holds iff x is the
Gödel number of a derivation of the sentence with Gödel num-
ber y. The relation that holds between x and y if x is the Gödel
number of a refutation of the sentence with Gödel number y is
also decidable. Let not(x) be the primitive recursive function
which does the following: if x is the code of a formula A, not(x)
is a code of ¬A. Then RefT (x, y) holds iff PrfT (x, not(y)). Let
RefT (x, y) represent it. Then, if T ⊢ ¬A and 𝛿 is a corresponding
derivation, Q ⊢ RefT (⌜𝛿⌝, ⌜A⌝). We define RProvT (y) as

∃x ( PrfT (x, y) ∧ ∀z (z < x → ¬RefT (z , y))).

Roughly, RProvT (y) says “there is a proof of y in T, and there is

no shorter refutation of y.” Assuming T is consistent, RProvT (y)
is true of the same numbers as ProvT (y); but from the point of
view of provability in T (and we now know that there is a differ-
ence between truth and provability!) the two have different prop-
erties. If T is inconsistent, then the two do not hold of the same
109 5.4. ROSSER’S THEOREM

numbers! (RProvT (y) is often read as “y is Rosser provable.”

Since, as just discussed, Rosser provability is not some special
kind of provability—in inconsistent theories, there are sentences
that are provable but not Rosser provable—this may be confus-
ing. To avoid the confusion, you could instead read it as “y is
shmovable.”)
By the fixed-point lemma, there is a formula R T such that

Q ⊢ R T ↔ ¬RProvT (⌜R T ⌝). (5.4)

In contrast to the proof of Theorem 5.6, here we claim that if T is

consistent, T doesn’t derive R T , and T also doesn’t derive ¬R T .
(In other words, we don’t need the assumption of 𝜔-consistency.)
First, let’s show that T ⊬ RT . Suppose it did, so there is
a derivation of RT from T ; let n be its Gödel number. Then Q ⊢
PrfT (n, ⌜RT ⌝), since PrfT represents PrfT in Q . Also, for each
k < n, k is not the Gödel number of ¬RT , since T is consistent.
So for each k < n, Q ⊢ ¬RefT (k , ⌜RT ⌝). By Lemma 4.23, Q ⊢
∀z (z < n → ¬RefT (z , ⌜RT ⌝)). Thus,

Q ⊢ ∃x ( PrfT (x, ⌜RT ⌝) ∧ ∀z (z < x → ¬RefT (z , ⌜RT ⌝))),

but that’s just RProvT (⌜RT ⌝). By eq. (5.4), Q ⊢ ¬RT . Since T
extends Q , also T ⊢ ¬RT . We’ve assumed that T ⊢ RT , so T
would be inconsistent, contrary to the assumption of the theorem.
Now, let’s show that T ⊬ ¬RT . Again, suppose it did,
and suppose n is the Gödel number of a derivation of ¬RT .
Then RefT (n, #RT # ) holds, and since RefT represents RefT in
Q , Q ⊢ RefT (n, ⌜RT ⌝). We’ll again show that T would then be
inconsistent because it would also derive RT . Since

Q ⊢ RT ↔ ¬RProvT (⌜RT ⌝),

and since T extends Q , it suffices to show that

Q ⊢ ¬RProvT (⌜RT ⌝).

110 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

The sentence ¬RProvT (⌜RT ⌝), i.e.,

¬∃x ( PrfT (x, ⌜RT ⌝) ∧ ∀z (z < x → ¬RefT (z , ⌜RT ⌝)))

is logically equivalent to

∀x ( PrfT (x, ⌜RT ⌝) → ∃z (z < x ∧ RefT (z , ⌜RT ⌝)))

We argue informally using logic, making use of facts about what

Q derives. Suppose x is arbitrary and PrfT (x, ⌜RT ⌝). We already
know that T ⊬ RT , and so for every k , Q ⊢ ¬PrfT (k , ⌜RT ⌝).
Thus, for every k it follows that x ≠ k . In particular, we have (a)
that x ≠ n. We also have ¬(x = 0 ∨ x = 1 ∨ · · · ∨ x = n − 1) and
so by Lemma 4.23, (b) ¬(x < n). By Lemma 4.24, n < x. Since
Q ⊢ RefT (n, ⌜RT ⌝), we have n < x ∧ RefT (n, ⌜RT ⌝), and from
that ∃z (z < x ∧ RefT (z , ⌜RT ⌝)). Since x was arbitrary we get, as
required, that

∀x ( PrfT (x, ⌜RT ⌝) → ∃z (z < x ∧ RefT (z , ⌜RT ⌝))). □

5.5 Comparison with Gödel’s Original

Paper
It is worthwhile to spend some time with Gödel’s 1931 paper.
The introduction sketches the ideas we have just discussed. Even
if you just skim through the paper, it is easy to see what is go-
ing on at each stage: first Gödel describes the formal system P
(syntax, axioms, proof rules); then he defines the primitive recur-
sive functions and relations; then he shows that xBy is primitive
recursive, and argues that the primitive recursive functions and
relations are represented in P. He then goes on to prove the in-
completeness theorem, as above. In Section 3, he shows that one
can take the unprovable assertion to be a sentence in the lan-
guage of arithmetic. This is the origin of the 𝛽 -lemma, which is
111 5.6. THE DERIVABILITY CONDITIONS FOR PA

what we also used to handle sequences in showing that the recur-

sive functions are representable in Q . Gödel doesn’t go so far
to isolate a minimal set of axioms that suffice, but we now know
that Q will do the trick. Finally, in Section 4, he sketches a proof
of the second incompleteness theorem.

5.6 The Derivability Conditions for PA

Peano arithmetic, or PA, is the theory extending Q with induc-
tion axioms for all formulas. In other words, one adds to Q
axioms of the form
(A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x)
for every formula A. Notice that this is really a schema, which is to
say, infinitely many axioms (and it turns out that PA is not finitely
axiomatizable). But since one can effectively determine whether
or not a string of symbols is an instance of an induction axiom,
the set of axioms for PA is computable. PA is a much more robust
theory than Q . For example, one can easily prove that addition
and multiplication are commutative, using induction in the usual
way. In fact, most finitary number-theoretic and combinatorial
arguments can be carried out in PA.
Since PA is computably axiomatized, the derivability predi-
cate Prf PA (x, y) is computable and hence represented in Q (and
so, in PA). As before, we will take PrfPA (x, y) to denote the for-
mula representing the relation. Let ProvPA (y) be the formula
∃x Prf PA (x, y), which, intuitively says, “y is provable from the
axioms of PA.” The reason we need a little bit more than the
axioms of Q is we need to know that the theory we are using is
strong enough to derive a few basic facts about this derivability
predicate. In fact, what we need are the following facts:
P1. If PA ⊢ A, then PA ⊢ ProvPA (⌜A⌝).
P2. For all formulas A and B,
PA ⊢ ProvPA (⌜A → B ⌝) → ( ProvPA (⌜A⌝) → ProvPA (⌜B ⌝)).
112 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

P3. For every formula A,

PA ⊢ ProvPA (⌜A⌝) → ProvPA (⌜ProvPA (⌜A⌝)⌝).

The only way to verify that these three properties hold is to de-
scribe the formula ProvPA (y) carefully and use the axioms of PA
to describe the relevant formal proofs. Conditions (1) and (2) are
easy; it is really condition (3) that requires work. (Think about
what kind of work it entails . . . ) Carrying out the details would
be tedious and uninteresting, so here we will ask you to take it on
faith that PA has the three properties listed above. A reasonable
choice of ProvPA (y) will also satisfy

P4. If PA ⊢ ProvPA (⌜A⌝), then PA ⊢ A.

But we will not need this fact.

Incidentally, Gödel was lazy in the same way we are being
now. At the end of the 1931 paper, he sketches the proof of
the second incompleteness theorem, and promises the details in
a later paper. He never got around to it; since everyone who
understood the argument believed that it could be carried out
(he did not need to fill in the details.)

5.7 The Second Incompleteness Theorem

How can we express the assertion that PA doesn’t prove its own
consistency? Saying PA is inconsistent amounts to saying that
PA ⊢ 0 = 1. So we can take the consistency statement ConPA to be
the sentence ¬ProvPA (⌜0 = 1⌝), and then the following theorem
does the job:
113 5.7. THE SECOND INCOMPLETENESS THEOREM

Theorem 5.8. Assuming PA is consistent, then PA does not derive

ConPA .

It is important to note that the theorem depends on the par-

ticular representation of ConPA (i.e., the particular representa-
tion of ProvPA (y)). All we will use is that the representation of
ProvPA (y) satisfies the three derivability conditions, so the theo-
rem generalizes to any theory with a derivability predicate having
these properties.
It is informative to read Gödel’s sketch of an argument, since
the theorem follows like a good punch line. It goes like this. Let
G PA be the Gödel sentence that we constructed in the proof of
Theorem 5.6. We have shown “If PA is consistent, then PA does
not derive G PA .” If we formalize this in PA, we have a proof of

ConPA → ¬ProvPA (⌜G PA ⌝).

Now suppose PA derives ConPA . Then it derives

¬ProvPA (⌜G PA ⌝). But since G PA is a Gödel sentence, this is
equivalent to G PA . So PA derives G PA .
But: we know that if PA is consistent, it doesn’t derive G PA !
So if PA is consistent, it can’t derive ConPA .
To make the argument more precise, we will let G PA be the
Gödel sentence for PA and use the derivability conditions (P1)–
(P3) to show that PA derives ConPA → G PA . This will show that
PA doesn’t derive ConPA . Here is a sketch of the proof, in PA.
(For simplicity, we drop the PA subscripts.)

G ↔ ¬Prov (⌜G ⌝) (5.5)

G is a Gödel sentence
G → ¬Prov (⌜G ⌝) (5.6)
from eq. (5.5)
G → ( Prov (⌜G ⌝) → ⊥) (5.7)
from eq. (5.6) by logic
Prov (⌜G → ( Prov (⌜G ⌝) → ⊥)⌝) (5.8)
114 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

by from eq. (5.7) by condition P1

Prov (⌜G ⌝) → Prov (⌜( Prov (⌜G ⌝) → ⊥)⌝) (5.9)
from eq. (5.8) by condition P2
Prov (⌜G ⌝) → ( Prov (⌜ Prov (⌜G ⌝)⌝) → Prov (⌜⊥⌝)) (5.10)
from eq. (5.9) by condition P2 and logic
Prov (⌜G ⌝) → Prov (⌜ Prov (⌜G ⌝)⌝) (5.11)
by P3
Prov (⌜G ⌝) → Prov (⌜⊥⌝) (5.12)
from eq. (5.10) and eq. (5.11) by logic
Con → ¬Prov (⌜G ⌝) (5.13)
contraposition of eq. (5.12) and Con ≡ ¬Prov (⌜⊥⌝)
Con → G
from eq. (5.5) and eq. (5.13) by logic

The use of logic in the above just elementary facts from proposi-
tional logic, e.g., eq. (5.7) uses ⊢ ¬A ↔ (A → ⊥) and eq. (5.12)
uses A → (B → C ),A → B ⊢ A → C . The use of condi-
tion P2 in eq. (5.9) and eq. (5.10) relies on instances of P2,
Prov (⌜A → B ⌝) → ( Prov (⌜A⌝) → Prov (⌜B ⌝)). In the first one,
A ≡ G and B ≡ Prov (⌜G ⌝) → ⊥; in the second, A ≡ Prov (⌜G ⌝)
and B ≡ ⊥.
The more abstract version of the second incompleteness the-
orem is as follows:

Theorem 5.9. Let T be any consistent, axiomatized theory extending

Q and let ProvT (y) be any formula satisfying derivability conditions
P1–P3 for T. Then T does not derive ConT .

The moral of the story is that no “reasonable” consistent the-

ory for mathematics can derive its own consistency statement.
Suppose T is a theory of mathematics that includes Q and
Hilbert’s “finitary” reasoning (whatever that may be). Then, the
whole of T cannot derive the consistency statement of T, and
so, a fortiori, the finitary fragment can’t derive the consistency
115 5.8. LÖB’S THEOREM

statement of T either. In that sense, there cannot be a finitary

consistency proof for “all of mathematics.”
There is some leeway in interpreting the term “finitary,” and
Gödel, in the 1931 paper, grants the possibility that something we
may consider “finitary” may lie outside the kinds of mathematics
Hilbert wanted to formalize. But Gödel was being charitable;
today, it is hard to see how we might find something that can
reasonably be called finitary but is not formalizable in, say, ZFC.

5.8 Löb’s Theorem

The Gödel sentence for a theory T is a fixed point of ¬ProvT (x),
i.e., a sentence G such that

T ⊢ ¬ProvT (⌜G ⌝) ↔ G .

It is not derivable, because if T ⊢ G , (a) by derivability con-

dition (1), T ⊢ ProvT (⌜G ⌝), and (b) T ⊢ G together with
T ⊢ ¬ProvT (⌜G ⌝) ↔ G gives T ⊢ ¬ProvT (⌜G ⌝), and so T would
be inconsistent. Now it is natural to ask about the status of a
fixed point of ProvT (x), i.e., a sentence H such that

T ⊢ ProvT (⌜H ⌝) ↔ H .

If it were derivable, T ⊢ ProvT (⌜H ⌝) by condition (1), but the

same conclusion follows if we apply modus ponens to the equiv-
alence above. Hence, we don’t get that T is inconsistent, at least
not by the same argument as in the case of the Gödel sentence.
This of course does not show that T does derive H .
We can make headway on this question if we generalize it
a bit. The left-to-right direction of the fixed point equivalence,
ProvT (⌜H ⌝) → H , is an instance of a general schema called a
reflection principle: ProvT (⌜A⌝) → A. It is called that because it
expresses, in a sense, that T can “reflect” about what it can derive;
basically it says, “If T can derive A, then A is true,” for any A.
This is true for sound theories only, of course, and this suggests
116 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

that theories will in general not derive every instance of it. So

which instances can a theory (strong enough, and satisfying the
derivability conditions) derive? Certainly all those where A itself
is derivable. And that’s it, as the next result shows.

Theorem 5.10. Let T be an axiomatizable theory extending Q , and

suppose ProvT (y) is a formula satisfying conditions P1–P3 from sec-
tion 5.7. If T derives ProvT (⌜A⌝) → A, then in fact T derives A.

Put differently, if T ⊬ A, then T ⊬ ProvT (⌜A⌝) → A. This

result is known as Löb’s theorem.
The heuristic for the proof of Löb’s theorem is a clever proof
that Santa Claus exists. (If you don’t like that conclusion, you
are free to substitute any other conclusion you would like.) Here
it is:

1. Let X be the sentence, “If X is true, then Santa Claus ex-

ists.”

2. Suppose X is true.

3. Then what it says holds; i.e., we have: if X is true, then

Santa Claus exists.

4. Since we are assuming X is true, we can conclude that

Santa Claus exists, by modus ponens from (2) and (3).

5. We have succeeded in deriving (4), “Santa Claus exists,”

from the assumption (2), “X is true.” By conditional proof,
we have shown: “If X is true, then Santa Claus exists.”

6. But this is just the sentence X . So we have shown that X is

true.

7. But then, by the argument (2)–(4) above, Santa Claus ex-

ists.

A formalization of this idea, replacing “is true” with “is deriv-

able,” and “Santa Claus exists” with A, yields the proof of Löb’s
117 5.8. LÖB’S THEOREM

theorem. The trick is to apply the fixed-point lemma to the for-

mula ProvT (y) → A. The fixed point of that corresponds to the
sentence X in the preceding sketch.
Proof of Theorem 5.10. Suppose A is a sentence such that T derives
ProvT (⌜A⌝) → A. Let B (y) be the formula ProvT (y) → A, and use
the fixed-point lemma to find a sentence D such that T derives
D ↔ B (⌜D ⌝). Then each of the following is derivable in T:
D ↔ ( ProvT (⌜D ⌝) → A) (5.14)
D is a fixed point of B (y)
D → ( ProvT (⌜D ⌝) → A) (5.15)
from eq. (5.14)
ProvT (⌜D → ( ProvT (⌜D ⌝) → A)⌝) (5.16)
from eq. (5.15) by condition P1
ProvT (⌜D ⌝) → ProvT (⌜ ProvT (⌜D ⌝) → A⌝) (5.17)
from eq. (5.16) using condition P2
ProvT (⌜D ⌝) → ( ProvT (⌜ ProvT (⌜D ⌝)⌝) → ProvT (⌜A⌝)) (5.18)
from eq. (5.17) using P2 again
ProvT (⌜D ⌝) → ProvT (⌜ ProvT (⌜D ⌝)⌝) (5.19)
by derivability condition P3
ProvT (⌜D ⌝) → ProvT (⌜A⌝) (5.20)
from eq. (5.18) and eq. (5.19)
ProvT (⌜A⌝) → A (5.21)
by assumption of the theorem
ProvT (⌜D ⌝) → A (5.22)
from eq. (5.20) and eq. (5.21)
( ProvT (⌜D ⌝) → A) → D (5.23)
from eq. (5.14)
D (5.24)
from eq. (5.22) and eq. (5.23)
ProvT (⌜D ⌝) (5.25)
118 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

from eq. (5.24) by condition P1

A from eq. (5.21) and eq. (5.25) □

With Löb’s theorem in hand, there is a short proof of the sec-

ond incompleteness theorem (for theories having a derivability
predicate satisfying conditions P1–P3): if T ⊢ ProvT (⌜⊥⌝) → ⊥,
then T ⊢ ⊥. If T is consistent, T ⊬ ⊥. So, T ⊬ ProvT (⌜⊥⌝) → ⊥,
i.e., T ⊬ ConT . We can also apply it to show that H , the fixed
point of ProvT (x), is derivable. For since

T ⊢ ProvT (⌜H ⌝) ↔ H

in particular

T ⊢ ProvT (⌜H ⌝) → H

and so by Löb’s theorem, T ⊢ H .

5.9 The Undefinability of Truth

The notion of definability depends on having a formal semantics
for the language of arithmetic. We have described a set of formu-
las and sentences in the language of arithmetic. The “intended
interpretation” is to read such sentences as making assertions
about the natural numbers, and such an assertion can be true or
false. Let N be the structure with domain N and the standard in-
terpretation for the symbols in the language of arithmetic. Then
N ⊨ A means “A is true in the standard interpretation.”
119 5.9. THE UNDEFINABILITY OF TRUTH

Definition 5.11. A relation R (x 1 , . . . ,x k ) of natural numbers is

definable in N if and only if there is a formula A(x 1 , . . . ,x k )
in the language of arithmetic such that for every n1 , . . . ,nk ,
R (n 1 , . . . ,nk ) if and only if N ⊨ A(n 1 , . . . ,n k ).

Put differently, a relation is definable in N if and only if it

is representable in the theory TA, where TA = {A : N ⊨ A} is
the set of true sentences of arithmetic. (If this is not immediately
clear to you, you should go back and check the definitions and
convince yourself that this is the case.)

Lemma 5.12. Every computable relation is definable in N.

Proof. It is easy to check that the formula representing a relation

in Q defines the same relation in N. □

Now one can ask, is the converse also true? That is, is ev-
ery relation definable in N computable? The answer is no. For
example:

Lemma 5.13. The halting relation is definable in N.

Proof. Let H be the halting relation, i.e.,

H = {⟨e ,x⟩ : ∃s T (e ,x,s )}.

Let DT define T in N. Then

H = {⟨e ,x⟩ : N ⊨ ∃s DT (e ,x,s )},

so ∃s DT (z ,x,s ) defines H in N. □

What about TA itself? Is it definable in arithmetic? That is:

is the set {#A# : N ⊨ A} definable in arithmetic? Tarski’s theorem
answers this in the negative.
120 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

Theorem 5.14. The set of true sentences of arithmetic is not definable

in arithmetic.

Proof. Suppose D (x) defined it, i.e., N ⊨ A iff N ⊨ D (⌜A⌝). By

the fixed-point lemma, there is a formula A such that Q ⊢ A ↔
¬D (⌜A⌝), and hence N ⊨ A ↔ ¬D (⌜A⌝). But then N ⊨ A if
and only if N ⊨ ¬D (⌜A⌝), which contradicts the fact that D (y) is
supposed to define the set of true statements of arithmetic. □

Tarski applied this analysis to a more general philosophical

notion of truth. Given any language L, Tarski argued that an
adequate notion of truth for L would have to satisfy, for each
sentence X ,

‘X ’ is true if and only if X .

Tarski’s oft-quoted example, for English, is the sentence

‘Snow is white’ is true if and only if snow is white.

However, for any language strong enough to represent the diago-

nal function, and any linguistic predicate T (x), we can construct
a sentence X satisfying “X if and only if not T (‘X ’).” Given that
we do not want a truth predicate to declare some sentences to be
both true and false, Tarski concluded that one cannot specify a
truth predicate for all sentences in a language without, somehow,
stepping outside the bounds of the language. In other words, a
the truth predicate for a language cannot be defined in the lan-
guage itself.

Summary
The first incompleteness theorem states that for any consistent,
axiomatizable theory T that extends Q , there is a sentence G T
such that T ⊬ G T . G T is constructed in such a way that G T , in
a roundabout way, says “T does not prove G T .” Since T does
not, in fact, prove it, what it says is true. If N ⊨ T, then T
121 5.9. THE UNDEFINABILITY OF TRUTH

does not prove any false claims, so T ⊬ ¬G T . Such a sentence

is independent or undecidable in T. Gödel’s original proof
established that G T is independent on the assumption that T is
𝜔-consistent. Rosser improved the result by finding a different
sentence R T with is neither provable nor refutable in T as long
as T is simply consistent.
The construction of G T is effective: given an axiomatization
of T we could, in principle, write down G T . The “roundabout
way” in which G T states its own unprovability, is a special case of
a general result, the fixed-point lemma. It states that for any for-
mula B (y) in LA , there is a sentence A such that Q ⊢ A↔B (⌜A⌝).
(Here, ⌜A⌝ is the standard numeral for the Gödel number of A,
i.e., #A# .) To obtain G T , we use the formula ¬ProvT (y) as B (y).
We get ProvT as the culmination of our previous efforts: We know
that Prf T (n,m), which holds if n is the Gödel number of a deriva-
tion of the sentence with Gödel number m from T, is primitive
recursive. We also know that Q represents all primitive recur-
sive relations, and so there is some formula PrfT (x, y) that repre-
sents Prf T in Q . The provability predicate for T is ProvT (y)
is ∃x Prf T (x, y) then expresses provability in T. (It doesn’t rep-
resent it though: if T ⊢ A, then Q ⊢ ProvT (⌜A⌝); but if T ⊬ A,
then Q does not in general prove ¬ProvT (⌜A⌝).)
The second incompleteness theorem establishes that the
sentence ConT that expresses that T is consistent, i.e., T also
does not prove ¬ProvT (⌜⊥⌝). The proof of the second incom-
pleteness theorem requires some additional conditions on T, the
provability conditions. PA satisfies them, although Q does not.
Theories that satisfy the provability conditions also satisfy Löb’s
theorem: T ⊢ ProvT (⌜A⌝) → A iff T ⊢ A.
The fixed-point theorem also has another important conse-
quence. We say a relation R (n) is definable in LA if there is
a formula AR (x) such that N ⊨ AR (n) iff R (n) holds. For in-
stance, ProvT is definable, since ProvT defines it. The property n
has iff it is the Gödel number of a sentence true in N, however, is
not definable. This is Tarski’s theorem about the undefinability
of truth.
122 CHAPTER 5. INCOMPLETENESS AND PROVABILITY

Problems
Problem 5.1. Every 𝜔-consistent theory is consistent. Show that
the converse does not hold, i.e., that there are consistent but 𝜔-
inconsistent theories. Do this by showing that Q ∪ {¬G Q } is
consistent but 𝜔-inconsistent.

Problem 5.2. Show that PA derives G PA → ConPA .

Problem 5.3. Let T be a computably axiomatized theory, and

let ProvT be a derivability predicate for T. Consider the following
four statements:

1. If T ⊢ A, then T ⊢ ProvT (⌜A⌝).

2. T ⊢ A → ProvT (⌜A⌝).

3. If T ⊢ ProvT (⌜A⌝), then T ⊢ A.

4. T ⊢ ProvT (⌜A⌝) → A

Under what conditions are each of these statements true?

Problem 5.4. Show that Q (n) ⇔ n ∈ {#A# : Q ⊢ A} is definable

in arithmetic.
CHAPTER 6

Models of
Arithmetic
6.1 Introduction
The standard model of arithmetic is the structure N with |N| = N
in which 0, ′, +, ×, and < are interpreted as you would expect.
That is, 0 is 0, ′ is the successor function, + is interpeted as
addition and × as multiplication of the numbers in N. Specifically,

0N = 0
′N (n) = n + 1
+N (n,m) = n + m
×N (n,m) = nm

Of course, there are structures for LA that have domains other

than N. For instance, we can take M with domain |M| = {a}∗
(the finite sequences of the single symbol a, i.e., ∅, a, aa, aaa,
. . . ), and interpretations

0M = ∅
′M (s ) = s ⌢ a
+M (n,m) = a n+m

123
124 CHAPTER 6. MODELS OF ARITHMETIC

×M (n,m) = a nm

These two structures are “essentially the same” in the sense that
the only difference is the elements of the domains but not how
the elements of the domains are related among each other by
the interpretation functions. We say that the two structures are
isomorphic.
It is an easy consequence of the compactness theorem that
any theory true in N also has models that are not isomorphic
to N. Such structures are called non-standard. The interesting
thing about them is that while the elements of a standard model
(i.e., N, but also all structures isomorphic to it) are exhausted by
the values of the standard numerals n, i.e.,

|N| = {ValN (n) : n ∈ N}

that isn’t the case in non-standard models: if M is non-standard,

then there is at least one x ∈ |M| such that x ≠ ValM (n) for all n.
These non-standard elements are pretty neat: they are “in-
finite natural numbers.” But their existence also explains, in a
sense, the incompleteness phenomena. Consider an example,
e.g., the consistency statement for Peano arithmetic, ConPA , i.e.,
¬∃x PrfPA (x, ⌜⊥⌝). Since PA neither proves ConPA nor ¬ConPA ,
either can be consistently added to PA. Since PA is consistent,
N ⊨ ConPA , and consequently N ⊭ ¬ConPA . So N is not a model
of PA∪{¬ConPA }, and all its models must be nonstandard. Mod-
els of PA ∪ {¬ConPA } must contain some element that serves as
the witness that makes ∃x PrfPA (⌜⊥⌝) true, i.e., a Gödel number
of a derivation of a contradiction from PA. Such an element can’t
be standard—since PA ⊢ ¬PrfPA (n, ⌜⊥⌝) for every n.

6.2 Reducts and Expansions

Often it is useful or necessary to compare languages which have
symbols in common, as well as structures for these languages.
The most comon case is when all the symbols in a language L
125 6.3. ISOMORPHIC STRUCTURES

are also part of a language L′, i.e., L ⊆ L′. An L-structure M

can then always be expanded to an L′-structure by adding inter-
pretations of the additional symbols while leaving the interpre-
tations of the common symbols the same. On the other hand,
from an L′-structure M ′ we can obtain an L-structure simply by
“forgetting” the interpretations of the symbols that do not occur
in L.

Definition 6.1. Suppose L ⊆ L′, M is an L-structure and M ′

is an L′-structure. M is the reduct of M ′ to L, and M ′ is an
expansion of M to L′ iff

1. |M| = |M ′ |
′
2. For every constant symbol c ∈ L, c M = c M .

3. For every function symbol f ∈ L, f M =f M′ .

′
4. For every predicate symbol P ∈ L, P M = P M .

Proposition 6.2. If an L-structure M is a reduct of an L′-structure

M ′, then for all L-sentences A,

M ⊨ A iff M ′ ⊨ A.

Proof. Exercise. □

Definition 6.3. When we have an L-structure M, and L′ = L∪

{P } is the expansion of L obtained by adding a single n-place
predicate symbol P , and R ⊆ |M| n is an n-place relation, then we
′
write (M,R) for the expansion M ′ of M with P M = R.

6.3 Isomorphic Structures

First-order structures can be alike in one of two ways. One way
in which the can be alike is that they make the same sentences
126 CHAPTER 6. MODELS OF ARITHMETIC

true. We call such structures elementarily equivalent. But structures

can be very different and still make the same sentences true—for
instance, one can be countable and the other not. This is because
there are lots of features of a structure that cannot be expressed
in first-order languages, either because the language is not rich
enough, or because of fundamental limitations of first-order logic
such as the Löwenheim-Skolem theorem. So another, stricter,
aspect in which structures can be alike is if they are fundamentally
the same, in the sense that they only differ in the objects that make
them up, but not in their structural features. A way of making
this precise is by the notion of an isomorphism.

Definition 6.4. Given two structures M and M ′ for the same

language L, we say that M is elementarily equivalent to M ′, written
M ≡ M ′, if and only if for every sentence A of L, M ⊨ A iff
M ′ ⊨ A.

Definition 6.5. Given two structures M and M ′ for the same

language L, we say that M is isomorphic to M ′, written M ≃ M ′,
if and only if there is a function h : |M| → |M ′ | such that:

1. h is injective: if h (x) = h (y) then x = y;

2. h is surjective: for every y ∈ |M ′ | there is x ∈ |M| such that

h (x) = y;
′
3. for every constant symbol c : h (c M ) = c M ;

4. for every n-place predicate symbol P :

′
⟨a 1 , . . . ,an ⟩ ∈ P M iff ⟨h (a1 ), . . . ,h (an )⟩ ∈ P M ;

5. for every n-place function symbol f :

M M′
h(f (a1 , . . . ,an )) = f (h (a 1 ), . . . ,h (an )).
127 6.3. ISOMORPHIC STRUCTURES

Theorem 6.6. If M ≃ M ′ then M ≡ M ′.

Proof. Let h be an isomorphism of M onto M ′. For any assign-

ment s , h ◦ s is the composition of h and s , i.e., the assignment in
M ′ such that (h ◦ s )(x) = h (s (x)). By induction on t and A one
can prove the stronger claims:
′
M
a. h (ValsM (t )) = Valh◦s (t ).
b. M,s ⊨ A iff M ′,h ◦ s ⊨ A.
The first is proved by induction on the complexity of t .
′ ′
M
1. If t ≡ c , then ValsM (c ) = c M and Valh◦s (c ) = c M . Thus,
′
h (ValsM (t )) = h (c M ) = c M (by (3) of Definition 6.5)
M′
= Valh◦s (t ).
′
2. If t ≡ x, then ValsM (x) = s (x) and Valh◦s M
(x) = h (s (x)).
M M′
Thus, h (Vals (x)) = h (s (x)) = Valh◦s (x).
3. If t ≡ f (t1 , . . . ,tn ), then
ValsM (t ) = f M
(ValsM (t1 ), . . . , ValsM (tn )) and
M′ M M′ M′
Valh◦s (t ) =f (Valh◦s (t1 ), . . . , Valh◦s (tn )).
The induction hypothesis is that for each i , h (ValsM (ti )) =
M′
Valh◦s (ti ). So,
h (ValsM (t )) = h ( f M
(ValsM (t1 ), . . . , ValsM (tn ))
M M ′ M ′
= h(f (Valh◦s (t1 ), . . . , Valh◦s (tn )) (6.1)
M′ M′ M′
= f (Valh◦s (t1 ), . . . , Valh◦s (tn )) (6.2)
M′
= Valh◦s (t )
Here, eq. (6.1) follows by induction hypothesis and eq. (6.2)
by (5) of Definition 6.5.
Part (b) is left as an exercise.
If A is a sentence, the assignments s and h ◦ s are irrelevant,
and we have M ⊨ A iff M ′ ⊨ A. □
128 CHAPTER 6. MODELS OF ARITHMETIC

Definition 6.7. An automorphism of a structure 𝔐 is an isomor-

phism of 𝔐 onto itself.

6.4 The Theory of a Structure

Every structure M makes some sentences true, and some false.
The set of all the sentences it makes true is called its theory. That
set is in fact a theory, since anything it entails must be true in all
its models, including M.

Definition 6.8. Given a structure M, the theory of M is the set

Th(M) of sentences that are true in M, i.e., Th(M) = {A : M ⊨
A}.

We also use the term “theory” informally to refer to sets of

sentences having an intended interpretation, whether deductively
closed or not.
Proposition 6.9. For any M, Th(M) is complete.

Proof. For any sentence A either M ⊨ A or M ⊨ ¬A, so either

A ∈ Th(M) or ¬A ∈ Th(M). □

Proposition 6.10. If N |= A for every A ∈ Th(M), then M ≡ N.

Proof. Since N ⊨ A for all A ∈ Th(M), Th(M) ⊆ Th(N). If

N ⊨ A, then N ⊭ ¬A, so ¬A ∉ Th(M). Since Th(M) is complete,
A ∈ Th(M). So, Th(N) ⊆ Th(M), and we have M ≡ N. □
Remark 1. Consider R = ⟨R, <⟩, the structure whose domain is
the set R of the real numbers, in the language comprising only a
2-place predicate symbol interpreted as the < relation over the re-
als. Clearly R is uncountable; however, since Th(R) is obviously
consistent, by the Löwenheim-Skolem theorem it has a countable
model, say S, and by Proposition 6.10, R ≡ S. Moreover, since R
and S are not isomorphic, this shows that the converse of Theo-
rem 6.6 fails in general.
129 6.5. STANDARD MODELS OF ARITHMETIC

6.5 Standard Models of Arithmetic

The language of arithmetic LA is obviously intended to be about
numbers, specifically, about natural numbers. So, “the” standard
model N is special: it is the model we want to talk about. But
in logic, we are often just interested in structural properties, and
any two structures that are isomorphic share those. So we can be
a bit more liberal, and consider any structure that is isomorphic
to N “standard.”

Definition 6.11. A structure for LA is standard if it is isomorphic

to N.

Proposition 6.12. If a structure M standard, its domain is the set of

values of the standard numerals, i.e.,

|M| = {ValM (n) : n ∈ N}

Proof. Clearly, every ValM (n) ∈ |M|. We just have to show that ev-
ery x ∈ |M| is equal to ValM (n) for some n. Since M is standard,
it is isomorphic to N. Suppose g : N → |M| is an isomorphism.
Then g (n) = g (ValN (n)) = ValM (n). But for every x ∈ |M|, there
is an n ∈ N such that g (n) = x, since g is surjective. □

If a structure M for LA is standard, the elements of its do-

main can all be named by the standard numerals 0, 1, 2, . . . , i.e.,
the terms 0, 0′, 0′′, etc. Of course, this does not mean that the
elements of |M| are the numbers, just that we can pick them out
the same way we can pick out the numbers in |N|.

Proposition 6.13. If M ⊨ Q , and |M| = {ValM (n) : n ∈ N}, then

M is standard.

Proof. We have to show that M is isomorphic to N. Consider the

function g : N → |M| defined by g (n) = ValM (n). By the hypoth-
esis, g is surjective. It is also injective: Q ⊢ n ≠ m whenever
130 CHAPTER 6. MODELS OF ARITHMETIC

n ≠ m. Thus, since M ⊨ Q , M ⊨ n ≠ m, whenever n ≠ m. Thus,

if n ≠ m, then ValM (n) ≠ ValM (m), i.e., g (n) ≠ g (m).
We also have to verify that g is an isomorphism.

1. We have g (0N ) = g (0) since, 0N = 0. By definition of g ,

g (0) = ValM (0). But 0 is just 0, and the value of a term
which happens to be a constant symbol is given by what the
structure assigns to that constant symbol, i.e., ValM (0) =
0M . So we have g (0N ) = 0M as required.

2. g (′N (n)) = g (n + 1), since ′ in N is the successor function

on N. Then, g (n + 1) = ValM (n + 1) by definition of g . But
n + 1 is the same term as n ′, so ValM (n + 1) = ValM (n ′). By
the definition of the value function, this is = ′M (ValM (n)).
Since ValM (n) = g (n) we get g (′N (n)) = ′M (g (n)).

3. g (+N (n,m)) = g (n + m), since + in N is the addition

function on N. Then, g (n + m) = ValM (n + m) by defini-
tion of g . But Q ⊢ n + m = (n + m), so ValM (n + m) =
ValM (n + m). By the definition of the value function, this
is = +M (ValM (n), ValM (m)). Since ValM (n) = g (n) and
ValM (m) = g (m), we get g (+N (n,m)) = +M (g (n), g (m)).

4. g (×N (n,m)) = ×M (g (n), g (m)): Exercise.

5. ⟨n,m⟩ ∈ <N iff n < m. If n < m, then Q ⊢ n < m, and

also M ⊨ n < m. Thus ⟨ValM (n), ValM (m)⟩ ∈ <M , i.e.,
⟨g (n), g (m)⟩ ∈ <M . If n ≮ m, then Q ⊢ ¬n < m, and
consequently M ⊭ n < m. Thus, as before, ⟨g (n), g (m)⟩ ∉
<M . Together, we get: ⟨n,m⟩ ∈ <N iff ⟨g (n), g (m)⟩ ∈ <M .
□

The function g is the most obvious way of defining a mapping

from N to the domain of any other structure M for LA , since
every such M contains elements named by 0, 1, 2, etc. So it isn’t
surprising that if M makes at least some basic statements about
the n’s true in the same way that N does, and g is also bijective,
131 6.5. STANDARD MODELS OF ARITHMETIC

then g will turn into an isomorphism. In fact, if |M| contains no

elements other than what the n’s name, it’s the only one.

Proposition 6.14. If M is standard, then g from the proof of Propo-

sition 6.13 is the only isomorphism from N to M.

Proof. Suppose h : N → |M| is an isomorphism between N

and M. We show that g = h by induction on n. If n = 0, then
g (0) = 0M by definition of g . But since h is an isomorphism,
h (0) = h (0N ) = 0M , so g (0) = h (0).
Now consider the case for n + 1. We have

g (n + 1) = ValM (n + 1) by definition of g
= ValM (n ′) since n + 1 ≡ n ′
= ′M (ValM (n)) by definition of ValM (t ′)
= ′M (g (n)) by definition of g
= ′M (h (n)) by induction hypothesis
= h (′N (n)) since h is an isomorphism
= h (n + 1) □

For any countably infinite set M , there’s a bijection between

N and M , so every such set M is potentially the domain of a
standard model M. In fact, once you pick an object z ∈ M and
a suitable function s as 0M and ′M , the interpretations of +, ×,
and < is already fixed. Only functions s : M → M \ {z } that
are both injective and surjective are suitable in a standard model
as ′M . The range of s cannot contain z , since otherwise ∀x 0 ≠ x ′
would be false. That sentence is true in N, and so M also has
to make it true. The function s has to be injective, since the
successor function ′N in N is, and that ′N is injective is expressed
by a sentence true in N. It has to be surjective because otherwise
there would be some x ∈ M \ {z } not in the domain of s , i.e., the
sentence ∀x (x = 0 ∨ ∃y y ′ = x) would be false in M—but it is true
in N.
132 CHAPTER 6. MODELS OF ARITHMETIC

6.6 Non-Standard Models

We call a structure for LA standard if it is isomorphic to N. If a
structure isn’t isomorphic to N, it is called non-standard.

Definition 6.15. A structure M for LA is non-standard if it is

not isomorphic to N. The elements x ∈ |M| which are equal to
ValM (n) for some n ∈ N are called standard numbers (of M), and
those not, non-standard numbers.

By Proposition 6.12, any standard structure for LA contains

only standard elements. Consequently, a non-standard structure
must contain at least one non-standard element. In fact, the ex-
istence of a non-standard element guarantees that the structure
is non-standard.

Proposition 6.16. If a structure M for LA contains a non-standard

number, M is non-standard.

Proof. Suppose not, i.e., suppose M standard but contains a non-

standard number x. Let g : N → |M| be an isomorphism. It is
easy to see (by induction on n) that g (ValN (n)) = ValM (n). In
other words, g maps standard numbers of N to standard num-
bers of M. If M contains a non-standard number, g cannot be
surjective, contrary to hypothesis. □

It is easy enough to specify non-standard structures for LA .

For instance, take the structure with domain Z and interpret all
non-logical symbols as usual. Since negative numbers are not
values of n for any n, this structure is non-standard. Of course,
it will not be a model of arithmetic in the sense that it makes
the same sentences true as N. For instance, ∀x x ′ ≠ 0 is false.
However, we can prove that non-standard models of arithmetic
exist easily enough, using the compactness theorem.
133 6.7. MODELS OF Q

Proposition 6.17. Let TA = {A : N ⊨ A} be the theory of N. TA

has a countable non-standard model.

Proof. Expand LA by a new constant symbol c and consider the

set of sentences

𝛤 = TA ∪ {c ≠ 0,c ≠ 1,c ≠ 2, . . . }

Any model Mc of 𝛤 would contain an element x = c M which is

non-standard, since x ≠ ValM (n) for all n ∈ N. Also, obviously,
Mc ⊨ TA, since TA ⊆ 𝛤. If we turn Mc into a structure M for
LA simply by forgetting about c , its domain still contains the non-
standard x, and also M ⊨ TA. The latter is guaranteed since c
does not occur in TA. So, it suffices to show that 𝛤 has a model.
We use the compactness theorem to show that 𝛤 has a model.
If every finite subset of 𝛤 is satisfiable, so is 𝛤. Consider any finite
subset 𝛤0 ⊆ 𝛤. 𝛤0 includes some sentences of TA and some of
the form c ≠ n, but only finitely many. Suppose k is the largest
number so that c ≠ k ∈ 𝛤0 . Define Nk by expanding N to include
the interpretation c Nk = k + 1. Nk ⊨ 𝛤0 : if A ∈ TA, Nk ⊨ A since
Nk is just like N in all respects except c , and c does not occur
in A. And Nk ⊨ c ≠ n, since n ≤ k , and ValNk (c ) = k + 1. Thus,
every finite subset of 𝛤 is satisfiable. □

6.7 Models of Q
We know that there are non-standard structures that make the
same sentences true as N does, i.e., is a model of TA. Since
N ⊨ Q , any model of TA is also a model of Q . Q is much
weaker than TA, e.g., Q ⊬ ∀x ∀y (x +y) = (y +x). Weaker theories
are easier to satisfy: they have more models. E.g., Q has models
which make ∀x ∀y (x + y) = (y + x) false, but those cannot also
be models of TA, or PA for that matter. Models of Q are also
relatively simple: we can specify them explicitly.
134 CHAPTER 6. MODELS OF ARITHMETIC

Example 6.18. Consider the structure K with domain |K| = N ∪

{a} and interpretations

0K = 0
{︄
x +1 if x ∈ N
′K (x) =
a if x = a
{︄
K x +y if x, y ∈ N
+ (x, y) =
a otherwise
⎧
⎪
⎪ xy if x, y ∈ N
K
⎨
⎪
× (x, y) = 0 if x = 0 or y = 0
⎪
⎪a otherwise
⎪
⎩
<K = {⟨x, y⟩ : x, y ∈ N and x < y } ∪ {⟨x,a⟩ : x ∈ |K|}

To show that K ⊨ Q we have to verify that all axioms of Q are

true in K. For convenience, let’s write x ∗ for ′K (x) (the “succes-
sor” of x in K), x ⊕ y for +K (x, y) (the “sum” of x and y in K,
x ⊗ y for ×K (x, y) (the “product” of x and y in K), and x 4 y for
⟨x, y⟩ ∈ <K . With these abbreviations, we can give the operations
in K more perspicuously as

x ⊕y 0 m a x ⊗y 0 m a
x x∗
0 0 m a 0 0 0 0
n n+1
n n n +m a n 0 nm a
a a
a a a a a 0 a a

We have n 4 m iff n < m for n, m ∈ N and x 4 a for all x ∈ |K|.

K ⊨ ∀x ∀y (x ′ = y ′ → x = y) since ∗ is injective. K ⊨ ∀x 0 ≠ x ′
since 0 is not a ∗-successor in K. K ⊨ ∀x (x = 0 ∨ ∃y x = y ′) since
for every n > 0, n = (n − 1) ∗ , and a = a ∗ .
K ⊨ ∀x (x + 0) = x since n ⊕ 0 = n + 0 = n, and a ⊕ 0 = a by
definition of ⊕. K ⊨ ∀x ∀y (x + y ′) = (x + y) ′ is a bit trickier. If n,
m are both standard, we have:

(n ⊕ m ∗ ) = (n + (m + 1)) = (n + m) + 1 = (n ⊕ m) ∗
135 6.7. MODELS OF Q

since ⊕ and ∗ agree with + and ′ on standard numbers. Now

suppose x ∈ |K|. Then

(x ⊕ a ∗ ) = (x ⊕ a) = a = a ∗ = (x ⊕ a) ∗

The remaining case is if y ∈ |K| but x = a. Here we also have to

distinguish cases according to whether y = n is standard or y = b:

(a ⊕ n ∗ ) = (a ⊕ (n + 1)) = a = a ∗ = (a ⊕ n) ∗
(a ⊕ a ∗ ) = (a ⊕ a) = a = a ∗ = (a ⊕ a) ∗

This is of course a bit more detailed than needed. For instance,

since a⊕z = a whatever z is, we can immediately conclude a⊕a ∗ =
a. The remaining axioms can be verified the same way.
K is thus a model of Q . Its “addition” ⊕ is also commutative.
But there are other sentences true in N but false in K, and vice
versa. For instance, a 4 a, so K ⊨ ∃x x < x and K ⊭ ∀x ¬x < x.
This shows that Q ⊬ ∀x ¬x < x.

Example 6.19. Consider the structure L with domain |L| = N ∪

{a,b } and interpretations ′L = ∗, +L = ⊕ given by

x x∗ x ⊕y m a b
n n+1 n n +m b a
a a a a b a
b b b b b a

Since ∗ is injective, 0 is not in its range, and every x ∈ |L| other

than 0 is, axioms Q 1 –Q 3 are true in L. For any x, x ⊕ 0 = x, so
Q 4 is true as well. For Q 5 , consider x ⊕ y ∗ and (x ⊕ y) ∗ . They
are equal if x and y are both standard, since then ∗ and ⊕ agree
with ′ and +. If x is non-standard, and y is standard, we have
x ⊕ y ∗ = x = x ∗ = (x ⊕ y) ∗ . If x and y are both non-standard, we
have four cases:

a ⊕ a ∗ = b = b ∗ = (a ⊕ a) ∗
b ⊕ b ∗ = a = a ∗ = (b ⊕ b) ∗
136 CHAPTER 6. MODELS OF ARITHMETIC

b ⊕ a ∗ = b = b ∗ = (b ⊕ y) ∗
a ⊕ b ∗ = a = a ∗ = (a ⊕ b) ∗

If x is standard, but y is non-standard, we have

n ⊕ a ∗ = n ⊕ a = b = b ∗ = (n ⊕ a) ∗
n ⊕ b ∗ = n ⊕ b = a = a ∗ = (n ⊕ b) ∗

So, L ⊨ Q 5 . However, a ⊕ 0 ≠ 0 ⊕ a, so L ⊭ ∀x ∀y (x + y) = (y + x).

We’ve explicitly constructed models of Q in which the non-

standard elements live “beyond” the standard elements. In fact,
that much is required by the axioms. A non-standard element x
cannot be 4 0, since Q ⊢ ∀x ¬x < 0 (see Lemma 4.22). Also,
for every n, Q ⊢ ∀x (x < n ′ → (x = 0 ∨ x = 1 ∨ · · · ∨ x = n))
(Lemma 4.23), so we can’t have a 4 n for any n > 0.

6.8 Models of PA
Any non-standard model of TA is also one of PA. We know that
non-standard models of TA and hence of PA exist. We also know
that such non-standard models contain non-standard “numbers,”
i.e., elements of the domain that are “beyond” all the standard
“numbers.” But how are they arranged? How many are there?
We’ve seen that models of the weaker theory Q can contain as
few as a single non-standard number. But these simple structures
are not models of PA or TA.
The key to understanding the structure of models of PA or
TA is to see what facts are derivable in these theories. For in-
stance, already PA proves that ∀x x ≠ x ′ and ∀x ∀y (x +y) = (y +x),
so this rules out simple structures (in which these sentences are
false) as models of PA.
Suppose M is a model of PA. Then if PA ⊢ A, M ⊨ A. Let’s
again use z for 0M , ∗ for ′M , ⊕ for +M , ⊗ for ×M , and 4 for <M .
Any sentence A then states some condition about z, ∗, ⊕, ⊗, and
137 6.8. MODELS OF PA

4, and if M ⊨ A that condition must be satisfied. For instance,

if M ⊨ Q 1 , i.e., M ⊨ ∀x ∀y (x ′ = y ′ → x = y), then ∗ must be
injective.

Proposition 6.20. In M, 4 is a linear strict order, i.e., it satisfies:

1. Not x 4 x for any x ∈ |M|.

2. If x 4 y and y 4 z then x 4 z .

3. For any x ≠ y, x 4 y or y 4 x

Proof. PA proves:
1. ∀x ¬x < x
2. ∀x ∀y ∀z ((x < y ∧ y < z ) → x < z )
3. ∀x ∀y ((x < y ∨ y < x) ∨ x = y)) □

Proposition 6.21. z is the least element of |M| in the 4-ordering.

For any x, x 4 x ∗ , and x ∗ is the 4-least element with that property. For
any x, there is a unique y such that y ∗ = x. (We call y the “predecessor”
of x in M, and denote it by ∗ x.)

Proof. Exercise. □

Proposition 6.22. All standard elements of M are less than (accord-

ing to 4) all non-standard elements.

Proof. We’ll use n as short for ValM (n), a standard element of M.

Already Q proves that, for any n ∈ N, ∀x (x < n ′ → (x = 0 ∨ x =
1 ∨ · · · ∨ x = n)). There are no elements that are 4z. So if n
is standard and x is non-standard, we cannot have x 4 n. By
definition, a non-standard element is one that isn’t ValM (n) for
any n ∈ N, so x ≠ n as well. Since 4 is a linear order, we must
have n 4 x. □
138 CHAPTER 6. MODELS OF ARITHMETIC

Proposition 6.23. Every nonstandard element x of |M| is an element

of the subset

. . .∗∗∗ x 4∗∗ x 4∗ x 4 x 4 x ∗ 4 x ∗∗ 4 x ∗∗∗ 4 . . .

We call this subset the block of x and write it as [x]. It has no least and
no greatest element. It can be characterized as the set of those y ∈ |M|
such that, for some standard n, x ⊕ n = y or y ⊕ n = x.

Proof. Clearly, such a set [x] always exists since every element y
of |M| has a unique successor y ∗ and unique predecessor ∗ y. For
successive elements y, y ∗ we have y 4 y ∗ and y ∗ is the 4-least
element of |M| such that y is 4-less than it. Since always ∗ y 4 y
and y 4 y ∗ , [x] has no least or greatest element. If y ∈ [x] then
x ∈ [y], for then either y ∗...∗ = x or x ∗...∗ = y. If y ∗...∗ = x (with n
∗’s), then y ⊕ n = x and conversely, since PA ⊢ ∀x x ′...′ = (x + n)
(if n is the number of ′’s). □

Proposition 6.24. If [x] ≠ [y] and x 4 y, then for any u ∈ [x] and
any v ∈ [y], u 4 v .

Proof. Note that PA ⊢ ∀x ∀y (x < y → (x ′ < y ∨ x ′ = y)). Thus, if

u 4 v , we also have u ⊕ n ∗ 4 v for any n if [u] ≠ [v ].
Any u ∈ [x] is 4y: x 4 y by assumption. If u 4 x, u 4 y by
transitivity. And if x 4 u but u ∈ [x], we have u = x ⊕ n ∗ for
some n, and so u 4 y by the fact just proved.
Now suppose that v ∈ [y] is 4y, i.e., v ⊕ m ∗ = y for some
standard m. This rules out v 4 x, otherwise y = v ⊕ m ∗ 4 x.
Clearly also, x ≠ v , otherwise x ⊕ m ∗ = v ⊕ m ∗ = y and we would
have [x] = [y]. So, x 4 v . But then also x ⊕ n ∗ 4 v for any n.
Hence, if x 4 u and u ∈ [x], we have u 4 v . If u 4 x then u 4 v by
transitivity.
Lastly, if y 4 v , u 4 v since, as we’ve shown, u 4 y and y 4 v .□
139 6.8. MODELS OF PA

Corollary 6.25. If [x] ≠ [y], [x] ∩ [y] = ∅.

Proof. Suppose z ∈ [x] and x 4 y. Then z 4 u for all u ∈ [y]. If

z ∈ [y], we would have z 4 z . Similarly if y 4 x. □

This means that the blocks themselves can be ordered in a

way that respects 4: [x] 4 [y] iff x 4 y, or, equivalently, if u 4 v
for any u ∈ [x] and v ∈ [y]. Clearly, the standard block [0] is
the least block. It intersects with no non-standard block, and no
two non-standard blocks intersect either. Specifically, you cannot
“reach” a different block by taking repeated successors or prede-
cessors.

Proposition 6.26. If x and y are non-standard, then x 4 x ⊕ y and

x ⊕ y ∉ [x].

Proof. If y is nonstandard, then y ≠ z. PA ⊢ ∀x (y ≠ 0 → x <

(x + y)). Now suppose x ⊕ y ∈ [x]. Since x 4 x ⊕ y, we would have
x ⊕ n ∗ = x ⊕ y. But PA ⊢ ∀x ∀y ∀z ((x + y) = (x + z ) → y = z ) (the
cancellation law for addition). This would mean y = n ∗ for some
standard n; but y is assumed to be non-standard. □

Proposition 6.27. There is no least non-standard block.

Proof. PA ⊢ ∀x ∃y ((y + y) = x ∨ (y + y) ′ = x), i.e., that every x is

divisible by 2 (possibly with remainder 1). If x is non-standard,
so is y. By the preceding proposition, y 4 y ⊕ y and y ⊕ y ∉ [y].
Then also y 4 (y ⊕ y) ∗ and (y ⊕ y) ∗ ∉ [y]. But x = y ⊕ y or
x = (y ⊕ y) ∗ , so y 4 x and y ∉ [x]. □

Proposition 6.28. There is no largest block.

Proof. Exercise. □
140 CHAPTER 6. MODELS OF ARITHMETIC

Proposition 6.29. The ordering of the blocks is dense. That is, if

x 4 y and [x] ≠ [y], then there is a block [z ] distinct from both that
is between them.

Proof. Suppose x 4y. As before, x ⊕ y is divisible by two (possibly

with remainder): there is a z ∈ |M| such that either x ⊕ y = z ⊕ z
or x ⊕ y = (z ⊕ z ) ∗ . The element z is the “average” of x and y,
and x 4 z and z 4 y. □

The non-standard blocks are therefore ordered like the ratio-

nals: they form a countably infinite dense linear ordering without
endpoints. One can show that any two such countably infinite
orderings are isomorphic. It follows that for any two countable
non-standard models M1 and M2 of true arithmetic, their reducts
to the language containing < and = only are isomorphic. Indeed,
an isomorphism h can be defined as follows: the standard parts
of M1 and M2 are isomorphic to the standard model N and hence
to each other. The blocks making up the non-standard part are
themselves ordered like the rationals and therefore isomorphic;
an isomorphism of the blocks can be extended to an isomorphism
within the blocks by matching up arbitrary elements in each, and
then taking the image of the successor of x in M1 to be the suc-
cessor of the image of x in M2 . Note that it does not follow that
𝔐1 and 𝔐2 are isomorphic in the full language of arithmetic (in-
deed, isomorphism is always relative to a language), as there are
non-isomorphic ways to define addition and multiplication over
|M1 | and |M2 |. (This also follows from a famous theorem due
to Vaught that the number of countable models of a complete
theory cannot be 2.)

6.9 Computable Models of Arithmetic

The standard model N has two nice features. Its domain is the
natural numbers N, i.e., its elements are just the kinds of things
we want to talk about using the language of arithmetic, and the
standard numeral n actually picks out n. The other nice feature
141 6.9. COMPUTABLE MODELS OF ARITHMETIC

is that the interpretations of the non-logical symbols of LA are

all computable. The successor, addition, and multiplication func-
tions which serve as ′N , +N , and ×N are computable functions
of numbers. (Computable by Turing machines, or definable by
primitive recursion, say.) And the less-than relation on N, i.e.,
<N , is decidable.
Non-standard models of arithmetical theories such as Q and
PA must contain non-standard elements. Thus their domains typ-
ically include elements in addition to N. However, any countable
structure can be built on any countably infinite set, including N.
So there are also non-standard models with domain N. In such
models M, of course, at least some numbers cannot play the roles
they usually play, since some k must be different from ValM (n)
for all n ∈ N.

Definition 6.30. A structure M for LA is computable iff |M| = N

and ′M , +M , ×M are computable functions and <M is a decidable
relation.

Example 6.31. Recall the structure K from Example 6.18. Its

domain was |K| = N ∪ {a} and interpretations

0K = 0
{︄
x + 1 if x ∈ N
′K (x) =
a if x = a
{︄
x +y if x, y ∈ N
+K (x, y) =
a otherwise
⎧
⎪
⎪ xy if x, y ∈ N
K
⎨
⎪
× (x, y) = 0 if x = 0 or y = 0
⎪
⎪a otherwise
⎪
⎩
<K = {⟨x, y⟩ : x, y ∈ N and x < y } ∪ {⟨x,a⟩ : n ∈ |K|}

But |K| is countably infinite and so is equinumerous with N. For

instance, g : N → |K| with g (0) = a and g (n) = n + 1 for n > 0 is
142 CHAPTER 6. MODELS OF ARITHMETIC

a bijection. We can turn it into an isomorphism between a new

model K ′ of Q and K. In K ′, we have to assign different functions
and relations to the symbols of LA , since different elements of N
play the roles of standard and non-standard numbers.
Specifically, 0 now plays the role of a, not of the smallest
standard number. The smallest standard number is now 1. So
′
we assign 0K = 1. The successor function is also different now:
given a standard number, i.e., an n > 0, it still returns n +1. But 0
′
now plays the role of a, which is its own successor. So ′K (0) = 0.
For addition and multiplication we likewise have
{︄
′ x + y − 1 if x, y > 0
+K (x, y) =
0 otherwise
⎧
⎪
⎪ 1 if x = 1 or y = 1
K′
⎨
⎪
× (x, y) = xy − x − y + 2 if x, y > 1
⎪
⎪0
⎪
otherwise
⎩
′
And we have ⟨x, y⟩ ∈ <K iff x < y and x > 0 and y > 0, or if
y = 0.
All of these functions are computable functions of natural
′
numbers and <K is a decidable relation on N—but they are
not the same functions as successor, addition, and multiplication
′
on N, and <K is not the same relation as < on N.

Example 6.31 shows that Q has computable non-standard

models with domain N. However, the following result shows
that this is not true for models of PA (and thus also for mod-
els of TA).

Theorem 6.32 (Tennenbaum’s Theorem). N is the only com-

putable model of PA.

Summary
A model of arithmetic is a structure for the language LA of
arithmetic. There is one distinguished such model, the standard
143 6.9. COMPUTABLE MODELS OF ARITHMETIC

model N, with |N| = N and interpretations of 0, ′, +, ×, and <

given by 0, the successor, addition, and multiplication functions
on N, and the less-than relation. N is a model of the theories Q
and PA.
More generally, a structure for LA is called standard iff it is
isomorphic to N. Two structures are isomorphic if there is an
isomorphism between them, i.e., a bijective function which pre-
serves the interpretations of constant symbols, function symbols,
and predicate symbols. By the isomorphism theorem, isomor-
phic structures are elementarily equivalent, i.e., they make the
same sentences true. In standard models, the domain is just the
set of values of all the numerals n.
Models of Q and PA that are not isomorphic to N are called
non-standard. In non-standard models, the domain is not ex-
hausted by the values of the numerals. An element x ∈ |M| where
x ≠ ValM (n) for all n ∈ N is called a non-standard element
of M. If M ⊨ Q , non-standard elements must obey the axioms
of Q , e.g., they have unique successors, they can be added and
multiplied, and compared using <. The standard elements of M
are all <M all the non-standard elements. Non-standard models
exist because of the compactness theorem, and for Q they can
relatively easily be given explicitly. Such models can be used
to show that, e.g., Q is not strong enough to prove certain sen-
tences, e.g., Q ⊬ ∀x ∀y (x + y) = (y +x). This is done by defining a
non-standard M in which non-standard elements don’t obey the
law of commutativity.
Non-standard models of PA cannot be so easily specified ex-
plicitly. By showing that PA proves certain sentences, we can in-
vestigate the structure of the non-standard part of a non-standard
model of PA. If a non-standard model M of PA is countable, ev-
ery non-standard element is part of a “block” of non-standard
elements which are ordered like Z by <M . These blocks them-
selves are arranged like Q, i.e., there is no smallest or largest
block, and there is always a block in between any two blocks.
Any countable model is isomorphic to one with domain N.
If the interpretations of ′, +, ×, and < in such a model are com-
144 CHAPTER 6. MODELS OF ARITHMETIC

putable functions, we say it is a computable model. The stan-

dard model N is computable, since the successor, addition, and
multiplication functions and the less-than relation on N are com-
putable. It is possible to define computable non-standard models
of Q , but N is the only computable model of PA. This is Tan-
nenbaum’s Theorem.

Problems
Problem 6.1. Prove Proposition 6.2.

Problem 6.2. Carry out the proof of (b) of Theorem 6.6 in de-
tail. Make sure to note where each of the five properties charac-
terizing isomorphisms of Definition 6.5 is used.

Problem 6.3. Show that for any structure M, if X is a definable

subset of M, and h is an automorphism of M, then X = {h (x) :
x ∈ X } (i.e., X is fixed under h).

Problem 6.4. Show that the converse of Proposition 6.12 is false,

i.e., give an example of a structure M with |M| = {ValM (n) : n ∈
N} that is not isomorphic to N.

Problem 6.5. Recall that Q contains the axioms

∀x ∀y (x ′ = y ′ → x = y) (Q 1 )
′
∀x 0 ≠ x (Q 2 )
′
∀x (x = 0 ∨ ∃y x = y ) (Q 3 )

Give structures M1 , M2 , M3 such that

1. M1 ⊨ Q 1 , M1 ⊨ Q 2 , M1 ⊭ Q 3 ;

2. M2 ⊨ Q 1 , M2 ⊭ Q 2 , M2 ⊨ Q 3 ; and

3. M3 ⊭ Q 1 , M3 ⊨ Q 2 , M3 ⊨ Q 3 ;

Obviously, you just have to specify 0Mi and ′Mi for each.
145 6.9. COMPUTABLE MODELS OF ARITHMETIC

Problem 6.6. Prove that K from Example 6.18 satisifies the re-
maining axioms of Q ,

∀x (x × 0) = 0 (Q 6 )
′
∀x ∀y (x × y ) = ((x × y) + x) (Q 7 )
′
∀x ∀y (x < y ↔ ∃z (z + x) = y) (Q 8 )

Find a sentence only involving ′ true in N but false in K.

Problem 6.7. Expand L of Example 6.19 to include ⊗ and 4

that interpret × and <. Show that your structure satisifies the
remaining axioms of Q ,

∀x (x × 0) = 0 (Q 6 )
′
∀x ∀y (x × y ) = ((x × y) + x) (Q 7 )
′
∀x ∀y (x < y ↔ ∃z (z + x) = y) (Q 8 )

Problem 6.8. In L of Example 6.19, a ∗ = a and b ∗ = b. Is there

a model of Q in which a ∗ = b and b ∗ = a?

Problem 6.9. Find sentences in LA derivable in PA (and hence

true in N) which guarantee the properties of z, ∗, and 4 in Propo-
sition 6.21

Problem 6.10. Show that in a non-standard model of PA, there

is no largest block.

Problem 6.11. Write out a detailed proof of Proposition 6.29.

Which sentence must PA derive in order to guarantee the exis-
tence of z ? Why is x 4 z and z 4 y, and why is [x] ≠ [z ] and
[z ] ≠ [y]?

Problem 6.12. Give a structure L′ with |L′ | = N isomorphic to L

of Example 6.19.
CHAPTER 7

Second-Order
Logic
7.1 Introduction
In first-order logic, we combine the non-logical symbols of a given
language, i.e., its constant symbols, function symbols, and pred-
icate symbols, with the logical symbols to express things about
first-order structures. This is done using the notion of satisfac-
tion, which relates a structure M, together with a variable assign-
ment s , and a formula A: M,s ⊨ A holds iff what A expresses when
its constant symbols, function symbols, and predicate symbols
are interpreted as M says, and its free variables are interpreted
as s says, is true. The interpretation of the identity predicate =
is built into the definition of M,s ⊨ A, as is the interpretation
of ∀ and ∃. The former is always interpreted as the identity rela-
tion on the domain |M| of the structure, and the quantifiers are
always interpreted as ranging over the entire domain. But, cru-
cially, quantification is only allowed over elements of the domain,
and so only object variables are allowed to follow a quantifier.
In second-order logic, both the language and the definition of
satisfaction are extended to include free and bound function and
predicate variables, and quantification over them. These vari-
ables are related to function symbols and predicate symbols the

146
147 7.2. TERMS AND FORMULAS

same way that object variables are related to constant symbols.

They play the same role in the formation of terms and formulas
of second-order logic, and quantification over them is handled in
a similar way. In the standard semantics, the second-order quanti-
fiers range over all possible objects of the right type (n-place func-
tions from |M| to |M| for function variables, n-place relations for
predicate variables). For instance, while ∀v0 (P01 (v0 )∨¬P01 (v0 )) is
a formula in both first- and second-order logic, in the latter we can
also consider ∀V01 ∀v0 (V01 (v0 ) ∨ ¬V01 (v0 )) and ∃V01 ∀v0 (V01 (v0 ) ∨
¬V01 (v0 )). Since these contain no free varaibles, they are sen-
tences of second-order logic. Here, V01 is a second-order 1-place
predicate variable. The allowable interpretations of V01 are the
same that we can assign to a 1-place predicate symbol like P01 ,
i.e., subsets of |M|. Quantification over them then amounts to
saying that ∀v0 (V01 (v0 ) ∨ ¬V01 (v 0 )) holds for all ways of assigning
a subset of |M| as the value of V01 , or for at least one. Since every
set either contains or fails to contain a given object, both are true
in any structure.
Since second-order logic can quantify over subsets of the do-
main as well as functions, it is to be expected that some amount,
at least, of set theory can be carried out in second-order logic.
By “carry out,” we mean that it is possible to express set theo-
retic properties and statements in second-order logic, and is pos-
sible without any special, non-logical vocabulary for sets (e.g.,
the membership predicate symbol of set theory). For instance,
we can define unions and intersections of sets and the subset re-
lationship, but also compare the sizes of sets, and state results
such as Cantor’s Theorem.

7.2 Terms and Formulas

Like in first-order logic, expressions of second-order logic are
built up from a basic vocabulary containing variables, constant
symbols, predicate symbols and sometimes function symbols. From
them, together with logical connectives, quantifiers, and punctu-
148 CHAPTER 7. SECOND-ORDER LOGIC

ation symbols such as parentheses and commas, terms and formu-

las are formed. The difference is that in addition to variables for
objects, second-order logic also contains variables for relations
and functions, and allows quantification over them. So the log-
ical symbols of second-order logic are those of first-order logic,
plus:

1. A countably infinite set of second-order relation variables

of every arity n: V0n , V1n , V2n , . . .

2. A countably infinite set of second-order function variables:

u0n , u1n , u2n , . . .

Just as we use x, y, z as meta-variables for first-order variables

vi , we’ll use X , Y , Z , etc., as metavariables for Vin and u, v , etc.,
as meta-variables for uin .
The non-logical symbols of a second-order language are spec-
ified the same way a first-order language is: by listing its constant
symbols, function symbols, and predicate symbols.
In first-order logic, the identity predicate = is usually in-
cluded. In first-order logic, the non-logical symbols of a lan-
guage L are crucial to allow us to express anything interesting.
There are of course sentences that use no non-logical symbols,
but with only = it is hard to say anything interesting. In second-
order logic, since we have an unlimited supply of relation and
function variables, we can say anything we can say in a first-order
language even without a special supply of non-logical symbols.

Definition 7.1 (Second-order Terms). The set of second-order

terms of L, Trm2 (L), is defined by adding to Definition B.4 the
clause

1. If u is an n-place function variable and t1 , . . . , tn are terms,

then u (t1 , . . . ,tn ) is a term.

So, a second-order term looks just like a first-order term, ex-

cept that where a first-order term contains a function symbol fi n ,
149 7.3. SATISFACTION

a second-order term may contain a function variable uin in its

place.

Definition 7.2 (Second-order formula). The set of second-order

formulas Frm2 (L) of the language L is defined by adding to Def-
inition B.4 the clauses

1. If X is an n-place predicate variable and t1 , . . . , tn are

second-order terms of L, then X (t1 , . . . ,tn ) is an atomic
formula.

2. If A is a formula and u is a function variable, then ∀u A is

a formula.

3. If A is a formula and X is a predicate variable, then ∀X A

is a formula.

4. If A is a formula and u is a function variable, then ∃u A is

a formula.

5. If A is a formula and X is a predicate variable, then ∃X A

is a formula.

7.3 Satisfaction
To define the satisfaction relation M,s ⊨ A for second-order for-
mulas, we have to extend the definitions to cover second-order
variables. The notion of a structure is the same for second-order
logic as it is for first-order logic. There is only a diffence for vari-
able assignments s : these now must not just provide values for
the first-order variables, but also for the second-order variables.

Definition 7.3 (Variable Assignment). A variable assign-

ment s for a structure M is a function which maps each

1. object variable vi to an element of |M|, i.e., s (vi ) ∈ |M|

150 CHAPTER 7. SECOND-ORDER LOGIC

2. n-place relation variable Vi n to an n-place relation on |M|,

i.e., s (Vi n ) ⊆ |M| n ;

3. n-place function variable uin to an n-place function from

|M| to |M|, i.e., s (uin ) : |M| n → |M|;

A structure assigns a value to each constant symbol and func-

tion symbol, and a second-order variable assigns objects and
functions to each object and function variable. Together, they
let us assign a value to every term.

Definition 7.4 (Value of a Term). If t is a term of the lan-

guage L, M is a structure for L, and s is a variable assignment
for M, the value ValsM (t ) is defined as for first-order terms, plus
the following clause:

t ≡ u (t1 , . . . ,tn ):

ValsM (t ) = s (u) (ValsM (t1 ), . . . , ValsM (tn )).

Definition 7.5 (x-Variant). If s is a variable assignment for

a structure M, then any variable assignment s ′ for M which dif-
fers from s at most in what it assigns to x is called an x-variant
of s . If s ′ is an x-variant of s we write s ∼x s ′. (Similarly for
second-order variables X or u.)

Definition 7.6 (Satisfaction). For second-order formulas A,

the definition of satisfaction is like Definition B.23 with the addi-
tion of:

1. A ≡ X n (t1 , . . . ,tn ): M,s ⊨ A iff ⟨ValsM (t1 ), . . . , ValsM (tn )⟩ ∈

s (X n ).

2. A ≡ ∀X B: M,s ⊨ A iff for every X -variant s ′ of s , M,s ′ ⊨ B.

151 7.3. SATISFACTION

3. A ≡ ∃X B: M,s ⊨ A iff there is an X -variant s ′ of s so that

M,s ′ ⊨ B.

4. A ≡ ∀u B: M,s ⊨ A iff for every u-variant s ′ of s , M,s ′ ⊨ B.

5. A ≡ ∃u B: M,s ⊨ A iff there is an u-variant s ′ of s so that

M,s ′ ⊨ B.

Example 7.7. Consider the formula ∀z (X (z ) ↔ ¬Y (z )). It con-

tains no second-order quantifiers, but does contain the second-
order variables X and Y (here understood to be one-place). The
corresponding first-order sentence ∀z (P (z ) ↔ ¬R (z )) says that
whatever falls under the interpretation of P does not fall under
the interpretation of R and vice versa. In a structure, the interpre-
tation of a predicate symbol P is given by the interpretation P M .
But for second-order variables like X and Y , the interpretation is
provided, not by the structure itself, but by a variable assignment.
Since the second-order formula is not a sentence (in includes free
variables X and Y ), it is only satisfied relative to a structure M
together with a variable assignment s .
M,s ⊨ ∀z (X (z ) ↔ ¬Y (z )) whenever the elements of s (X ) are
not elements of s (Y ), and vice versa, i.e., iff s (Y ) = |M| \ s (X ).
So for instance, take |M| = {1, 2, 3}. Since no predicate symbols,
function symbols, or constant symbols are involved, the domain
of M is all that is relevant. Now for s 1 (X ) = {1, 2} and s 1 (Y ) =
{3}, we have M,s1 ⊨ ∀z (X (z ) ↔ ¬Y (z )).
By contrast, if we have s2 (X ) = {1, 2} and s2 (Y ) = {2, 3},
M,s 2 ⊭ ∀z (X (z ) ↔ ¬Y (z )). That’s because there is a z -variant
s 2′ of s 2 with s 2′ (z ) = 2 where M,s 2′ ⊨ X (z ) (since 2 ∈ s 2′ (X )) but
M,s 2′ ⊭ ¬Y (z ) (since also s 2′ (z ) ∈ s 2′ (Y )).

Example 7.8. M,s ⊨ ∃Y (∃y Y (y) ∧ ∀z (X (z ) ↔ ¬Y (z ))) if there

is an s ′ ∼Y s such that M,s ′ ⊨ (∃y Y (y) ∧ ∀z (X (z ) ↔ ¬Y (z ))).
And that is the case iff s ′ (Y ) ≠ ∅ (so that M,s ′ ⊨ ∃y Y (y)) and,
as in the previous example, s ′ (Y ) = |M| \ s ′ (X ). In other words,
M,s ⊨ ∃Y (∃y Y (y) ∧ ∀z (X (z ) ↔ ¬Y (z ))) iff |M| \ s (X ) is non-
152 CHAPTER 7. SECOND-ORDER LOGIC

empty, i.e., s (X ) ≠ |M|. So, the formula is satisfied, e.g., if |M| =

{1, 2, 3} and s (X ) = {1, 2}, but not if s (X ) = {1, 2, 3} = |M|.
Since the formula is not satisfied whenever s (X ) = |M|, the
sentence

∀X ∃Y (∃y Y (y) ∧ ∀z (X (z ) ↔ ¬Y (z )))

is never satisfied: For any structure M, the assignment s (X ) =

|M| will make the sentence false. On the other hand, the sentence

∃X ∃Y (∃y Y (y) ∧ ∀z (X (z ) ↔ ¬Y (z )))

is satisfied relative to any assignment s , since we can always find

an X -variant s ′ of s with s ′ (X ) ≠ |M|.

7.4 Semantic Notions

The central logical notions of validity, entailment, and satisfiability
are defined the same way for second-order logic as they are for
first-order logic, except that the underlying satisfaction relation
is now that for second-order formulas. A second-order sentence,
of course, is a formula in which all variables, including predicate
and function variables, are bound.

Definition 7.9 (Validity). A sentence A is valid, ⊨ A, iff M ⊨ A

for every structure M.

Definition 7.10 (Entailment). A set of sentences 𝛤 entails a

sentence A, 𝛤 ⊨ A, iff for every structure M with M ⊨ 𝛤, M ⊨ A.

Definition 7.11 (Satisfiability). A set of sentences 𝛤 is satisfi-

able if M ⊨ 𝛤 for some structure M. If 𝛤 is not satisfiable it is
called unsatisfiable.
153 7.5. EXPRESSIVE POWER

7.5 Expressive Power

Quantification over second-order variables is responsible for an
immense increase in the expressive power of the language over
that of first-order logic. Second-order existential quantification
lets us say that functions or relations with certain properties ex-
ists. In first-order logic, the only way to do that is to specify a
non-logical symbol (i.e., a function symbol or predicate symbol)
for this purpose. Second-order universal quantification lets us say
that all subsets of, relations on, or functions from the domain to
the domain have a property. In first-order logic, we can only say
that the subsets, relations, or functions assigned to one of the
non-logical symbols of the language have a property. And when
we say that subsets, relations, functions exist that have a property,
or that all of them have it, we can use second-order quantification
in specifying this property as well. This lets us define relations
not definable in first-order logic, and express properties of the
domain not expressible in first-order logic.

Definition 7.12. If M is a structure for a language L, a rela-

tion R ⊆ |M| 2 is definable in L if there is some formula AR (x, y)
with only the variables x and y free, such that R (a,b) holds (i.e.,
⟨a,b⟩ ∈ R) iff M,s ⊨ AR (x, y) for s (x) = a and s (y) = b.

Example 7.13. In first-order logic we can define the identity re-

lation Id |M | (i.e., {⟨a,a⟩ : a ∈ |M|}) by the formula x = y. In
second-order logic, we can define this relation without =. For if
a and b are the same element of |M|, then they are elements of
the same subsets of |M| (since sets are determined by their el-
ements). Conversely, if a and b are different, then they are not
elements of the same subsets: e.g., a ∈ {a} but b ∉ {a} if a ≠ b.
So “being elements of the same subsets of |M|” is a relation that
holds of a and b iff a = b. It is a relation that can be expressed in
second-order logic, since we can quantify over all subsets of |M|.
Hence, the following formula defines Id |M | :
∀X (X (x) ↔ X (y))
154 CHAPTER 7. SECOND-ORDER LOGIC

Example 7.14. If R is a two-place predicate symbol, R M is a two-

place relation on |M|. Perhaps somewhat confusingly, we’ll use R
as the predicate symbol for R and for the relation R M itself. The
transitive closure R ∗ of R is the relation that holds between a and b
iff for some c 1 , . . . , c k , R (a,c 1 ), R (c 1 ,c 2 ), . . . , R (c k ,b) holds. This
includes the case if k = 0, i.e., if R (a,b) holds, so does R ∗ (a,b).
This means that R ⊆ R ∗ . In fact, R ∗ is the smallest relation that
includes R and that is transitive. We can say in second-order
logic that X is a transitive relation that includes R:

BR (X ) ≡ ∀x ∀y (R (x, y) → X (x, y)) ∧

∀x ∀y ∀z ((X (x, y) ∧ X (y, z )) → X (x, z )).

The first conjunct says that R ⊆ X and the second that X is

transitive.
To say that X is the smallest such relation is to say that it is
itself included in every relation that includes R and is transitive.
So we can define the transitive closure of R by the formula

R ∗ (X ) ≡ BR (X ) ∧ ∀Y (BR (Y ) → ∀x ∀y (X (x, y) → Y (x, y))).

We have M,s ⊨ R ∗ (X ) iff s (X ) = R ∗ . The transitive closure of R

cannot be expressed in first-order logic.

7.6 Describing Infinite and Countable

Domains
A set M is (Dedekind) infinite iff there is an injective func-
tion f : M → M which is not surjective, i.e., with dom( f ) ≠ M .
In first-order logic, we can consider a one-place function sym-
bol f and say that the function f M assigned to it in a structure M
is injective and ran( f ) ≠ |M|:

∀x ∀y ( f (x) = f (y) → x = y) ∧ ∃y ∀x y ≠ f (x).

If M satisfies this sentence, f M : |M| → |M| is injective, and so

|M| must be infinite. If |M| is infinite, and hence such a function
155 7.6. DESCRIBING INFINITE AND COUNTABLE DOMAINS

exists, we can let f M be that function and M will satisfy the

sentence. However, this requires that our language contains the
non-logical symbol f we use for this purpose. In second-order
logic, we can simply say that such a function exists. This no-longer
requires f , and we obtain the sentence in pure second-order logic

Inf ≡ ∃u (∀x ∀y (u (x) = u (y) → x = y) ∧ ∃y ∀x y ≠ u (x)).

M ⊨ Inf iff |M| is infinite. We can then define Fin ≡ ¬Inf; M ⊨ Fin
iff |M| is finite. No single sentence of pure first-order logic can
express that the domain is infinite although an infinite set of them
can. There is no set of sentences of pure first-order logic that is
satisfied in a structure iff its domain is finite.
Proposition 7.15. M ⊨ Inf iff |M| is infinite.

Proof. M ⊨ Inf iff M,s ⊨ ∀x ∀y (u (x) = u (y) → x = y) ∧ ∃y ∀x y ≠

u (x) for some s . If it does, s (u) is an injective function, and
some y ∈ |M| is not in the domain of s (u). Conversely, if there is
an injective f : |M| → |M| with dom( f ) ≠ |M|, then s (u) = f is
such a variable assignment. □

A set M is countable if there is an enumeration

m 0 ,m1 ,m 2 , . . .

of its elements (without repetitions but possibly finite). Such an

enumeration exists iff there is an element z ∈ M and a function
f : M → M such that z , f (z ), f ( f (z )), . . . , are all the elements
of M . For if the enumeration exists, z = m 0 and f (mk ) = mk +1 (or
f (mk ) = mk if mk is the last element of the enumeration) are the
requisite element and function. On the other hand, if such a z
and f exist, then z , f (z ), f ( f (z )), . . . , is an enumeration of M ,
and M is countable. We can express the existence of z and f in
second-order logic to produce a sentence true in a structure iff
the structure is countable:

Count ≡ ∃z ∃u ∀X ((X (z ) ∧ ∀x (X (x) → X (u (x)))) → ∀x X (x))

156 CHAPTER 7. SECOND-ORDER LOGIC

Proposition 7.16. M ⊨ Count iff |M| is countable.

Proof. Suppose |M| is countable, and let m 0 , m 1 , . . . , be an enu-

meration. By removing repetions we can guarantee that no mk ap-
pears twice. Define f (mk ) = mk +1 and let s (z ) = m0 and s (u) = f .
We show that
M,s ⊨ ∀X ((X (z ) ∧ ∀x (X (x) → X (u (x)))) → ∀x X (x))
Suppose s ′ ∼X s is arbitrary, and let M = s ′ (X ). Suppose further
that M,s ′ ⊨ (X (z ) ∧ ∀x (X (x) → X (u (x)))). Then s ′ (z ) ∈ M
and whenever x ∈ M , also s ′ (u) (x) ∈ M . In other words, since
s ′ ∼X s , m0 ∈ M and if x ∈ M then f (x) ∈ M , so m 0 ∈ M ,
m 1 = f (m 0 ) ∈ M , m 2 = f ( f (m0 )) ∈ M , etc. Thus, M = |M|, and
so M,s ′ ⊨ ∀x X (x). Since s ′ was an arbitrary X -variant of s , we
are done: M ⊨ Count.
Now assume that M ⊨ Count, i.e.,
M,s ⊨ ∀X ((X (z ) ∧ ∀x (X (x) → X (u (x)))) → ∀x X (x))
for some s . Let m = s (z ) and f = s (u) and consider M =
{m, f (m), f ( f (m)), . . . }. Let s ′ be the X -variant of s with s (X ) =
M . Then
M,s ′ ⊨ (X (z ) ∧ ∀x (X (x) → X (u (x)))) → ∀x X (x)
by assumption. Also, M,s ′ ⊨ X (z ) since s ′ (X ) = M ∋ m = s ′ (z ),
and also M,s ′ ⊨ ∀x (X (x) → X (u (x))) since whenever x ∈ M
also f (x) ∈ M . So, since both antecedent and conditional are
satisfied, the consequent must also be: M,s ′ ⊨ ∀x X (x). But that
means that M = |M|, and so |M| is countable since M is, by
definition. □

7.7 Second-order Arithmetic

Recall that the theory PA of Peano arithmetic includes the eight
axioms of Q ,
∀x x ′ ≠ 0
157 7.7. SECOND-ORDER ARITHMETIC

∀x ∀y (x ′ = y ′ → x = y)
∀x (x = 0 ∨ ∃y x = y ′)
∀x (x + 0) = x
∀x ∀y (x + y ′) = (x + y) ′
∀x (x × 0) = 0
∀x ∀y (x × y ′) = ((x × y) + x)
∀x ∀y (x < y ↔ ∃z (z ′ + x) = y)

plus all sentences of the form

(A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x).

The latter is a “schema,” i.e., a pattern that generates infinitely
many sentences of the language of arithmetic, one for each for-
mula A(x). We call this schema the (first-order) axiom schema of
induction. In second-order Peano arithmetic PA2 , induction can be
stated as a single sentence. PA2 consists of the first eight axioms
above plus the (second-order) induction axiom:
∀X (X (0) ∧ ∀x (X (x) → X (x ′))) → ∀x X (x).
It says that if a subset X of the domain contains 0M and with
any x ∈ |M| also contains ′M (x) (i.e., it is “closed under succes-
sor”) it contains everything in the domain (i.e., X = |M|).
The induction axiom guarantees that any structure satisfying
it contains only those elements of |M| the axioms require to be
there, i.e., the values of n for n ∈ N. A model of PA2 contains
no non-standard numbers.

Theorem 7.17. If M ⊨ PA2 then |M| = {ValM (n) : n ∈ N}.

Proof. Let N = {ValM (n) : n ∈ N}, and suppose M ⊨ PA2 . Of

course, for any n ∈ N, ValM (n) ∈ |M|, so N ⊆ |M|.
Now for inclusion in the other direction. Consider a variable
assignment s with s (X ) = N . By assumption,
M ⊨ ∀X (X (0) ∧ ∀x (X (x) → X (x ′))) → ∀x X (x), thus
158 CHAPTER 7. SECOND-ORDER LOGIC

M,s ⊨ (X (0) ∧ ∀x (X (x) → X (x ′))) → ∀x X (x).

Consider the antecedent of this conditional. ValM (0) ∈ N , and

so M,s ⊨ X (0). The second conjunct, ∀x (X (x) → X (x ′)) is also
satisfied. For suppose x ∈ N . By definition of N , x = ValM (n) for
some n. That gives ′M (x) = ValM (n + 1) ∈ N . So, ′M (x) ∈ N .
We have that M,s ⊨ X (0) ∧∀x (X (x) →X (x ′)). Consequently,
M,s ⊨ ∀x X (x). But that means that for every x ∈ |M| we have
x ∈ s (X ) = N . So, |M| ⊆ N . □

Corollary 7.18. Any two models of PA2 are isomorphic.

Proof. By Theorem 7.17, the domain of any model of PA2 is ex-

hausted by ValM (n). Any such model is also a model of Q . By
Proposition 6.13, any such model is standard, i.e., isomorphic
to N. □

Above we defined PA2 as the theory that contains the first

eight arithmetical axioms plus the second-order induction axiom.
In fact, thanks to the expressive power of second-order logic, only
the first two of the arithmetical axioms plus induction are needed
for second-order Peano arithmetic.

Proposition 7.19. Let PA2† be the second-order theory containing the

first two arithmetical axioms (the successor axioms) and the second-order
induction axiom. Then ≤, +, and × are definable in PA2† .

Proof. To show that ≤ is definable, we have to find a for-

mula A ≤ (x, y) such that N ⊨ A ≤ (n,m) iff n ≤ m. Consider the
formula
B (x,Y ) ≡ Y (x) ∧ ∀y (Y (y) → Y (y ′))
Clearly, B (n,Y ) is satisfied by a set Y ⊆ N iff {m : n ≤ m} ⊆ Y ,
so we can take A ≤ (x, y) ≡ ∀Y (B (x,Y ) → Y (y)). □
159 7.8. SECOND-ORDER LOGIC IS NOT AXIOMATIZABLE

Corollary 7.20. M ⊨ PA2 iff M ⊨ PA2† .

Proof. Immediate from Proposition 7.19. □

7.8 Second-order Logic is not

Axiomatizable
Theorem 7.21. Second-order logic is undecidable.

Proof. A first-order sentence is valid in first-order logic iff it is

valid in second-order logic, and first-order logic is undecidable.□

Theorem 7.22. There is no sound and complete proof system for

second-order logic.

Proof. Let A be a sentence in the language of arithmetic. N ⊨ A

iff PA2 ⊨ A. Let P be the conjunction of the nine axioms of PA2 .
PA2 ⊨ A iff ⊨ P → A, i.e., M ⊨ P → A . Now consider the
sentence ∀z ∀u ∀u ′ ∀u ′′ ∀L (P ′ → A ′) resulting by replacing 0 by
z , ′ by the one-place function variable u, + and × by the two-
place function-variables u ′ and u ′′, respectively, and < by the
two-place relation variable L and universally quantifying. It is a
valid sentence of pure second-order logic iff the original sentence
was valid iff PA2 ⊨ A iff N ⊨ A. Thus if there were a sound and
complete proof system for second-order logic, we could use it to
define a computable enumeration f : N → Sent(LA ) of the sen-
tences true in N. This function would be representable in Q by
some first-order formula B f (x, y). Then the formula ∃x B f (x, y)
would define the set of true first-order sentences of N, contradict-
ing Tarski’s Theorem. □

7.9 Second-order Logic is not Compact

Call a set of sentences 𝛤 finitely satisfiable if every one of its finite
subsets is satisfiable. First-order logic has the property that if a
160 CHAPTER 7. SECOND-ORDER LOGIC

set of sentences 𝛤 is finitely satisfiable, it is satisfiable. This prop-

erty is called compactness. It has an equivalent version involving
entailment: if 𝛤 ⊨ A, then already 𝛤0 ⊨ A for some finite sub-
set 𝛤0 ⊆ 𝛤. In this version it is an immediate corollary of the
completeness theorem: for if 𝛤 ⊨ A, by completeness 𝛤 ⊢ A. But
a derivation can only make use of finitely many sentences of 𝛤.
Compactness is not true for second-order logic. There are
sets of second-order sentences that are finitely satisfiable but not
satisfiable, and that entail some A without a finite subset entail-
ing A.

Theorem 7.23. Second-order logic is not compact.

Proof. Recall that

Inf ≡ ∃u (∀x ∀y (u (x) = u (y) → x = y) ∧ ∃y ∀x y ≠ u (x))

is satisfied in a structure iff its domain is infinite. Let A ≥n be

a sentence that asserts that the domain has at least n elements,
e.g.,

A ≥n ≡ ∃x 1 . . . ∃x n (x 1 ≠ x 2 ∧ x 1 ≠ x 3 ∧ · · · ∧ x n−1 ≠ x n ).

Consider the set of sentences

𝛤 = {¬Inf,A ≥1 ,A ≥2 ,A ≥3 , . . . }.

It is finitely satisfiable, since for any finite subset 𝛤0 ⊆ 𝛤 there is

some k so that A ≥k ∈ 𝛤 but no A ≥n ∈ 𝛤 for n > k . If |M| has k
elements, M ⊨ 𝛤0 . But, 𝛤 is not satisfiable: if M ⊨ ¬Inf, |M| must
be finite, say, of size k . Then M ⊭ A ≥k +1 . □

7.10 The Löwenheim-Skolem Theorem

Fails for Second-order Logic
The (Downward) Löwenheim-Skolem Theorem states that every
set of sentences with an infinite model has a countable model. It,
161 7.11. COMPARING SETS

too, is a consequence of the completeneness theorem: the proof

of completeness generates a model for any consistent set of sen-
tences, and that model is countable. There is also an Upward
Löwenheim-Skolem Theorem, which guarantees that if a set of
sentences has a countably infinite model it also has an uncount-
able model. Both theorems fail in second-order logic.

Theorem 7.24. The Löwenheim-Skolem Theorem fails for second-

order logic: There are sentences with infinite models but no countable
models.

Proof. Recall that

Count ≡ ∃z ∃u ∀X ((X (z ) ∧ ∀x (X (x) → X (u (x)))) → ∀x X (x))

is true in a structure M iff |M| is countable, so ¬Count is true

in M iff |M| is uncountable. There are such structures—take any
uncountable set as the domain, e.g., ℘(N) or R. So ¬Count has
infinite models but no countable models. □

Theorem 7.25. There are sentences with countably infinite but no

uncountable models.

Proof. Count ∧ Inf is true in N but not in any structure M with

|M| uncountable. □

7.11 Comparing Sets

Proposition 7.26. The formula ∀x (X (x)→Y (x)) defines the subset
relation, i.e., M,s ⊨ ∀x (X (x) → Y (x)) iff s (X ) ⊆ s (Y ).

Proposition 7.27. The formula ∀x (X (x) ↔Y (x)) defines the iden-

tity relation on sets, i.e., M,s ⊨ ∀x (X (x) ↔ Y (x)) iff s (X ) = s (Y ).
162 CHAPTER 7. SECOND-ORDER LOGIC

Proposition 7.28. The formula ∃x X (x) defines the property of being

non-empty, i.e., M,s ⊨ ∃x X (x) iff s (X ) ≠ ∅.

A set X is no larger than a set Y , X ⪯ Y , iff there is an in-

jective function f : X → Y . Since we can express that a function
is injective, and also that its values for arguments in X are in Y ,
we can also define the relation of being no larger than on subsets
of the domain.

Proposition 7.29. The formula

∃u (∀x (X (x) → Y (u (x))) ∧ ∀x ∀y (u (x) = u (y) → x = y))

defines the relation of being no larger than.

Two sets are the same size, or “equinumerous,” X ≈ Y , iff

there is a bijective function f : X → Y .

Proposition 7.30. The formula

∃u (∀x (X (x) → Y (u (x))) ∧

∀x ∀y (u (x) = u (y) → x = y) ∧
∀y (Y (y) → ∃x (X (x) ∧ y = u (x))))

defines the relation of being equinumerous with.

We will abbreviate these formulas, respectively, as X ⊆ Y ,

X = Y , X ≠ ∅, X ⪯ Y , and X ≈ Y . (This may be slightly con-
fusing, since we use the same notation when we speak informally
about sets X and Y —but here the notation is an abbreviation
for formulas in second-order logic involving one-place relation
variables X and Y .)
163 7.12. CARDINALITIES OF SETS

Proposition 7.31. The sentence ∀X ∀Y ((X ⪯ Y ∧Y ⪯ X ) → X ≈

Y ) is valid.

Proof. The sentence is satisfied in a structure M if, for any subsets

X ⊆ |M| and Y ⊆ |M|, if X ⪯ Y and Y ⪯ X then X ≈ Y .
But this holds for any sets X and Y —it is the Schröder-Bernstein
Theorem. □

7.12 Cardinalities of Sets

Just as we can express that the domain is finite or infinite, count-
able or uncountable, we can define the property of a subset of |M|
being finite or infinite, countable or uncountable.

Proposition 7.32. The formula Inf(X ) ≡

∃u (∀x ∀y (u (x) = u (y) → x = y) ∧

∃y (X (y) ∧ ∀x (X (x) → y ≠ u (x)))

is satisfied with respect to a variable assignment s iff s (X ) is infinite.

Proposition 7.33. The formula Count(X ) ≡

∃z ∃u (X (z ) ∧ ∀x (X (x) → X (u (x))) ∧
∀Y ((Y (z ) ∧ ∀x (Y (x) → Y (u (x)))) → X = Y ))

is satisfied with respect to a variable assignment s iff s (X ) is countable.

We know from Cantor’s Theorem that there are uncountable

sets, and in fact, that there are infinitely many different levels of
infinite sizes. Set theory develops an entire arithmetic of sizes of
sets, and assigns infinite cardinal numbers to sets. The natural
numbers serve as the cardinal numbers measuring the sizes of
finite sets. The cardinality of countably infinite sets is the first
infinite cardinality, called ℵ0 (“aleph-nought” or “aleph-zero”).
164 CHAPTER 7. SECOND-ORDER LOGIC

The next infinite size is ℵ1 . It is the smallest size a set can be

without being countable (i.e., of size ℵ0 ). We can define “X has
size ℵ0 ” as Aleph0 (X ) ↔ Inf(X ) ∧ Count(X ). X has size ℵ1 iff
all its subsets are finite or have size ℵ0 , but is not itself of size ℵ0 .
Hence we can express this by the formula Aleph1 (X ) ≡ ∀Y (Y ⊆
X → (¬Inf(Y ) ∨ Aleph0 (Y ))) ∧ ¬Aleph0 (X ). Being of size ℵ2 is
defined similarly, etc.
There is one size of special interest, the so-called cardinality
of the continuum. It is the size of ℘(N), or, equivalently, the size
of R. That a set is the size of the continuum can also be expressed
in second-order logic, but requires a bit more work.

7.13 The Power of the Continuum

In second-order logic we can quantify over subsets of the domain,
but not over sets of subsets of the domain. To do this directly,
we would need third-order logic. For instance, if we wanted to
state Cantor’s Theorem that there is no injective function from
the power set of a set to the set itself, we might try to formulate
it as “for every set X , and every set P , if P is the power set
of X , then not P ⪯ X ”. And to say that P is the power set of X
would require formalizing that the elements of P are all and only
the subsets of X , so something like ∀Y (P (Y ) ↔ Y ⊆ X ). The
problem lies in P (Y ): that is not a formula of second-order logic,
since only terms can be arguments to one-place relation variables
like P .
We can, however, simulate quantification over sets of sets, if
the domain is large enough. The idea is to make use of the fact
that two-place relations R relates elements of the domain to el-
ements of the domain. Given such an R, we can collect all the
elements to which some x is R-related: {y ∈ |M| : R (x, y)} is the
set “coded by” x. Conversely, if Z ⊆ ℘(|M|) is some collection
of subsets of |M|, and there are at least as many elements of |M|
as there are sets in Z , then there is also a relation R ⊆ |M| 2 such
that every Y ∈ Z is coded by some x using R.
165 7.13. THE POWER OF THE CONTINUUM

Definition 7.34. If R ⊆ |M| 2 , then x R-codes {y ∈ |M| : R (x, y)}.

If an element x ∈ |M| R-codes a set Z ⊆ |M|, then a set Y ⊆

|M| codes a set of sets, namely the sets coded by the elements
of Y . So a set Y can R-code ℘(X ). It does so iff for every Z ⊆ X ,
some x ∈ Y R-codes Z , and every x ∈ Y R-codes a Z ⊆ X .

Proposition 7.35. The formula

Codes(x,R,Z ) ≡ ∀y (Z (y) ↔ R (x, y))

expresses that s (x) s (R)-codes s (Z ). The formula

Pow(Y,R,X ) ≡
∀Z (Z ⊆ X → ∃x (Y (x) ∧ Codes(x,R,Z ))) ∧
∀x (Y (x) → ∀Z (Codes(x,R,Z ) → Z ⊆ X )

expresses that s (Y ) s (R)-codes the power set of s (X ), i.e., the elements

of s (Y ) s (R)-code exactly the subsets of s (X ).

With this trick, we can express statements about the power

set by quantifying over the codes of subsets rather than the sub-
sets themselves. For instance, Cantor’s Theorem can now be ex-
pressed by saying that there is no injective function from the do-
main of any relation that codes the power set of X to X itself.

Proposition 7.36. The sentence

∀X ∀Y ∀R (Pow(Y,R,X )→
¬∃u (∀x ∀y (u (x) = u (y) → x = y) ∧
∀x (Y (x) → X (u (x)))))

is valid.

The power set of a countably infinite set is uncountable, and

so its cardinality is larger than that of any countably infinite set
166 CHAPTER 7. SECOND-ORDER LOGIC

(which is ℵ0 ). The size of ℘(N) is called the “power of the contin-

uum,” since it is the same size as the points on the real number
line, R. If the domain is large enough to code the power set of
a countably infinite set, we can express that a set is the size of
the continuum by saying that it is equinumerous with any set Y
that codes the power set of set X of size ℵ0 . (If the domain is
not large enough, i.e., it contains no subset equinumerous with R,
then there can also be no relation that codes ℘(X ).)

Proposition 7.37. If R ⪯ |M|, then the formula

Cont(Y ) ≡ ∃X ∃R ((Aleph0 (X ) ∧ Pow(Y,R,X ))∧

∀x ∀y ((Y (x) ∧ Y (y) ∧ ∀z R (x, z ) ↔ R (y, z )) → x = y))

expresses that s (Y ) ≈ R.

Proof. Pow(Y,R,X ) expresses that s (Y ) s (R)-codes the power set

of s (X ), which Aleph0 (X ) says is countable. So s (Y ) is at least
as large as the power of the continuum, although it may be larger
(if multiple elements of s (Y ) code the same subset of X ). This
is ruled out be the last conjunct, which requires the association
between elements of s (Y ) and subsets of s (Z ) via s (R) to be
injective. □

Proposition 7.38. |M| ≈ R iff

M ⊨ ∃X ∃Y ∃R (Aleph0 (X ) ∧ Pow(Y,R,X )∧
∃u (∀x ∀y (u (x) = u (y) → x = y) ∧
∀y (Y (y) → ∃x y = u (x)))).

The Continuum Hypothesis is the statement that the size of

the continuum is the first uncountable cardinality, i.e, that ℘(N)
has size ℵ1 .
167 7.13. THE POWER OF THE CONTINUUM

Proposition 7.39. The Continuum Hypothesis is true iff

CH ≡ ∀X (Aleph1 (X ) ↔ Cont(X ))

is valid.

Note that it isn’t true that ¬CH is valid iff the Continuum
Hypothesis is false. In a countable domain, there are no subsets
of size ℵ1 and also no subsets of the size of the continuum, so
CH is always true in a countable domain. However, we can give
a different sentence that is valid iff the Continuum Hypothesis is
false:

Proposition 7.40. The Continuum Hypothesis is false iff

NCH ≡ ∀X (Cont(X ) → ∃Y (Y ⊆ X ∧ ¬Count(Y ) ∧ ¬X ≈ Y ))

is valid.

Summary
Second-order logic is an extension of first-order logic by vari-
ables for relations and functions, which can be quantified. Struc-
tures for second-order logic are just like first-order structures and
give the interpretations of all non-logical symbols of the language.
Variable assignments, however, also assign relations and func-
tions on the domain to the second-order variables. The satisfac-
tion relation is defined for second-order formulas just like in the
first-order case, but extended to deal with second-order variables
and quantifiers.
Second-order quantifiers make second-order logic more ex-
pressive than first-order logic. For instance, the identity rela-
tion on the domain of a structure can be defined without =, by
∀X (X (x) ↔ X (y)). Second-order logic can express the transi-
tive closure of a relation, which is not expressible in first-order
logic. Second-order quantifiers can also express properties of
168 CHAPTER 7. SECOND-ORDER LOGIC

the domain, that it is finite or infinite, countable or uncount-

able. This means that, e.g., there is a second-order sentence Inf
such that M ⊨ Inf iff |M| is infinite. Importantly, these are pure
second-order sentences, i.e., they contain no non-logical symbols.
Because of the compactness and Löwenheim-Skolem theorems,
there are no first-order sentences that have these properties. It
also shows that the compactness and Löwenheim-Skolem the-
orems fail for second-order logic.
Second-order quantification also makes it possible to replace
first-order schemas by single sentences. For instance, second-
order arithmetic PA2 is comprised of the axioms of Q plus the
single induction axiom

∀X ((X (0) ∧ ∀x (X (x) → X (x ′))) → ∀x X (x)).

In contrast to first-order PA, all second-order models of PA2 are

isomorphic to the standard model. In other words, PA2 has no
non-standard models.
Since second-order logic includes first-order logic, it is unde-
cidable. First-order logic is at least axiomatizable, i.e., it has a
sound and complete proof system. Second-order logic does not, it
is not axiomatizable. Thus, the set of validities of second-order
logic is highly non-computable. In fact, pure second-order logic
can express set-theoretic claims like the continuum hypothesis,
which are independent of set theory.

Problems
Problem 7.1. Show that ∀X (X (x) → X (y)) (note: → not ↔!)
defines Id |M | .

Problem 7.2. The sentence Inf ∧ Count is true in all and only
countably infinite domains. Adjust the definition of Count so
that it becomes a different sentence that directly expresses that
the domain is countably infinite, and prove that it does.

Problem 7.3. Complete the proof of Proposition 7.19.

169 7.13. THE POWER OF THE CONTINUUM

Problem 7.4. Give an example of a set 𝛤 and a sentence A so

that 𝛤 ⊨ A but for every finite subset 𝛤0 ⊆ 𝛤, 𝛤0 ⊭ A.
CHAPTER 8

The Lambda
Calculus
8.1 Overview
The lambda calculus was originally designed by Alonzo Church
in the early 1930s as a basis for constructive logic, and not as
a model of the computable functions. But it was soon shown to
be equivalent to other definitions of computability, such as the
Turing computable functions and the partial recursive functions.
The fact that this initially came as a small surprise makes the
characterization all the more interesting.
Lambda notation is a convenient way of referring to a func-
tion directly by a symbolic expression which defines it, instead of
defining a name for it. Instead of saying “let f be the function
defined by f (x) = x + 3,” one can say, “let f be the function
𝜆x . (x + 3).” In other words, 𝜆x . (x + 3) is just a name for the
function that adds three to its argument. In this expression, x is
a dummy variable, or a placeholder: the same function can just
as well be denoted by 𝜆 y . (y + 3). The notation works even with
other parameters around. For example, suppose g (x, y) is a func-
tion of two variables, and k is a natural number. Then 𝜆x . g (x,k )
is the function which maps any x to g (x,k ).
This way of defining a function from a symbolic expression is

170
171 8.1. OVERVIEW

known as lambda abstraction. The flip side of lambda abstraction

is application: assuming one has a function f (say, defined on
the natural numbers), one can apply it to any value, like 2. In
conventional notation, of course, we write f (2) for the result.
What happens when you combine lambda abstraction with
application? Then the resulting expression can be simplified, by
“plugging” the applicand in for the abstracted variable. For ex-
ample,
(𝜆x . (x + 3)) (2)
can be simplified to 2 + 3.
Up to this point, we have done nothing but introduce new
notations for conventional notions. The lambda calculus, how-
ever, represents a more radical departure from the set-theoretic
viewpoint. In this framework:

1. Everything denotes a function.

2. Functions can be defined using lambda abstraction.

3. Anything can be applied to anything else.

For example, if F is a term in the lambda calculus, F (F ) is always

assumed to be meaningful. This liberal framework is known as
the untyped lambda calculus, where “untyped” means “no restric-
tion on what can be applied to what.”
There is also a typed lambda calculus, which is an important
variation on the untyped version. Although in many ways the
typed lambda calculus is similar to the untyped one, it is much
easier to reconcile with a classical set-theoretic framework, and
has some very different properties.
Research on the lambda calculus has proved to be central in
theoretical computer science, and in the design of programming
languages. LISP, designed by John McCarthy in the 1950s, is an
early example of a language that was influenced by these ideas.
172 CHAPTER 8. THE LAMBDA CALCULUS

8.2 The Syntax of the Lambda Calculus

One starts with a sequence of variables x, y, z , . . . and some con-
stant symbols a, b, c , . . . . The set of terms is defined inductively,
as follows:

1. Each variable is a term.

2. Each constant is a term.

3. If M and N are terms, so is (M N ).

4. If M is a term and x is a variable, then (𝜆x . M ) is a term.

The system without any constants at all is called the pure lambda
calculus.
We will follow a few notational conventions:

Convention 8.1. 1. When parentheses are left out, application

takes place from left to right. For example, if M , N , P , and
Q are terms, then M N P Q abbreviates (((M N )P )Q ).

2. Again, when parentheses are left out, lambda abstraction

is to be given the widest scope possible. From example,
𝜆x . M N P is read (𝜆x . ((M N )P )).

3. A lambda can be used to abstract multiple variables. For

example, 𝜆xyz . M is short for 𝜆x . 𝜆 y . 𝜆 z . M .

For example,
𝜆xy . xxyx𝜆 z . xz
abbreviates
𝜆x . 𝜆 y . ((((xx)y)x) (𝜆 z . (xz ))).
You should memorize these conventions. They will drive you
crazy at first, but you will get used to them, and after a while
they will drive you less crazy than having to deal with a morass
of parentheses.
173 8.3. REDUCTION OF LAMBDA TERMS

Two terms that differ only in the names of the bound variables
are called 𝛼-equivalent; for example, 𝜆x . x and 𝜆 y . y. It will be
convenient to think of these as being the “same” term; in other
words, when we say that M and N are the same, we also mean
“up to renamings of the bound variables.” Variables that are in
the scope of a 𝜆 are called “bound”, while others are called “free.”
There are no free variables in the previous example; but in

(𝜆 z . yz )x

y and x are free, and z is bound.

8.3 Reduction of Lambda Terms

What can one do with lambda terms? Simplify them. If M and N
are any lambda terms and x is any variable, we can use M [N /x]
to denote the result of substituting N for x in M , after renam-
ing any bound variables of M that would interfere with the free
variables of N after the substitution. For example,

(𝜆w . xxw) [yyz /x] = 𝜆w . (yyz ) (yyz )w .

Alternative notations for substitution are [N /x]M , [x/N ]M ,

and also M [x/N ]. Beware!
Intuitively, (𝜆x . M )N and M [N /x] have the same meaning;
the act of replacing the first term by the second is called 𝛽 -
contraction. (𝜆x . M )N is called a redex and M [N /x] its contrac-
tum. Generally, if it is possible to change a term P to P ′ by
𝛽 -contraction of some subterm, we say that P 𝛽 -reduces to P ′ in
one step, and write P →− P ′. If from P we can obtain P ′ with some
number of one-step reductions (possibly none), then P 𝛽 -reduces
to P ′; in symbols, P →−→ P ′. A term that cannot be 𝛽 -reduced any
further is called 𝛽 -irreducible, or 𝛽 -normal. We will say “reduces”
instead of “𝛽 -reduces,” etc., when the context is clear.
Let us consider some examples.
174 CHAPTER 8. THE LAMBDA CALCULUS

1. We have

(𝜆x . xxy)𝜆 z . z →
− (𝜆 z . z ) (𝜆 z . z )y
→
− (𝜆 z . z )y
→
− y.

2. “Simplifying” a term can make it more complex:

(𝜆x . xxy) (𝜆x . xxy) →

− (𝜆x . xxy) (𝜆x . xxy)y
→
− (𝜆x . xxy) (𝜆x . xxy)yy
→
− ...

3. It can also leave a term unchanged:

(𝜆x . xx) (𝜆x . xx) →

− (𝜆x . xx) (𝜆x . xx).

4. Also, some terms can be reduced in more than one way;

for example,

(𝜆x . (𝜆 y . yx)z )v →
− (𝜆 y . yv )z

by contracting the outermost application; and

(𝜆x . (𝜆 y . yx)z )v →
− (𝜆x . zx)v

by contracting the innermost one. Note, in this case, how-

ever, that both terms further reduce to the same term, zv .

The final outcome in the last example is not a coincidence,

but rather illustrates a deep and important property of the
lambda calculus, known as the “Church-Rosser property.”

8.4 The Church-Rosser Property

175 8.5. CURRYING

Theorem 8.2. Let M , N 1 , and N2 be terms, such that M →

−
→ N1 and
M→−
→ N 2 . Then there is a term P such that N1 →
−
→ P and N2 →−
→ P.

Corollary 8.3. Suppose M can be reduced to normal form. Then this

normal form is unique.

Proof. If M →
→ N 1 and M →
− → N 2 , by the previous theorem there
−
is a term P such that N 1 and N 2 both reduce to P . If N1 and N 2
are both in normal form, this can only happen if N 1 ≡ P ≡ N 2 . □

Finally, we will say that two terms M and N are 𝛽 -equivalent,

or just equivalent, if they reduce to a common term; in other
words, if there is some P such that M → → P and N →
− → P . This is
−
𝛽 𝛽
written M = N . Using Theorem 8.2, you can check that = is an
equivalence relation, with the additional property that for every
𝛽
M and N , if M →
→ N or N →
− → M , then M = N . (In fact, one
−
𝛽
can show that = is the smallest equivalence relation having this
property.)

8.5 Currying
A 𝜆 -abstract 𝜆x . M represents a function of one argument, which
is quite a limitation when we want to define function accepting
multiple arguments. One way to do this would be by extending
the 𝜆 -calculus to allow the formation of pairs, triples, etc., in
which case, say, a three-place function 𝜆x . M would expect its
argument to be a triple. However, it is more convenient to do
this by Currying.
Let’s consider an example. If we want to define a function that
accepts two arguments and returns the first, we write 𝜆x . 𝜆 y . x,
which literally is a function that accepts an argument and returns
a function that accepts another argument and returns the first
argument while it drops the second. Let’s see what happens when
176 CHAPTER 8. THE LAMBDA CALCULUS

we apply it to two arguments:

𝛽
(𝜆x . 𝜆 y . x)M N −
→ (𝜆 y . M )N
𝛽
−
→M
In general, to write a function with parameters x 1 , . . . , x n
defined by some term N , we can write 𝜆x 1 . 𝜆x 2 . . . . 𝜆x n . N . If we
apply n arguments to it we get:
𝛽
(𝜆x 1 . 𝜆x 2 . . . . 𝜆x n . N )M1 . . . Mn −
→
𝛽
−
→ ((𝜆x 2 . . . . 𝜆x n . N ) [M1 /x 1 ])M2 . . . Mn
≡ (𝜆x 2 . . . . 𝜆x n . N [M 1 /x 1 ])M 2 . . . Mn
..
.
𝛽
−
→ P [M1 /x 1 ] . . . [Mn /x n ]
The last line literally means substituting Mi for x i in the body
of the function definition, which is exactly what we want when
applying multiple arguments to a function.

8.6 Lambda Definability

At first glance, the lambda calculus is just a very abstract cal-
culus of expressions that represent functions and applications of
them to others. Nothing in the syntax of the lambda calculus
suggests that these are functions of particular kinds of objects, in
particular, the syntax includes no mention of natural numbers.
Its basic operations—application and lambda abstractions—are
operations that apply to any function, not just functions on nat-
ural numbers.
Nevertheless, with some ingenuity, it is possible to define
arithmetical functions, i.e., functions on the natural numbers, in
the lambda calculus. To do this, we define, for each natural num-
ber n ∈ N, a special 𝜆 -term n, the Church numeral for n. (Church
numerals are named for Alonzo Church.)
177 8.6. LAMBDA DEFINABILITY

Definition 8.4. If n ∈ N, the corresponding Church numeral n

represents n:
n ≡ 𝜆 f x . f n (x)
Here, f n (x) stands for the result of applying f to x n times. For
example, 0 is 𝜆 f x . x, and 3 is 𝜆 f x . f ( f ( f x)).

The Church numeral n is encoded as a lambda term which

represents a function accepting two arguments f and x, and re-
turns f n (x). Church numerals are evidently in normal form.
A represention of natural numbers in the lambda calculus is
only useful, of course, if we can compute with them. Computing
with Church numerals in the lambda calculus means applying a
𝜆 -term F to such a Church numeral, and reducing the combined
term F n to a normal form. If it always reduces to a normal form,
and the normal form is always a Church numeral m, we can think
of the output of the computation as being the number m. We
can then think of F as defining a function f : N → N, namely
the function such that f (n) = m iff F n → → m. Because of the
−
Church-Rosser property, normal forms are unique if they exist.
So if F n → → m, there can be no other term in normal form, in
−
particular no other Church numeral, that F n reduces to.
Conversely, given a function f : N → N, we can ask if there
is a term F that defines f in this way. In that case we say that F
𝜆 -defines f , and that f is 𝜆 -definable. We can generalize this to
many-place and partial functions.
178 CHAPTER 8. THE LAMBDA CALCULUS

Definition 8.5. Suppose f : Nk → N. We say that a lambda

term F 𝜆 -defines f if for all n 0 , . . . , nk −1 ,

F n 0 m 1 . . . nk −1 →
−
→ f (n0 ,n 1 , . . . ,nk −1 )

if f (n 0 , . . . ,nk −1 ) is defined, and F n 0 n 1 . . . nk −1 has no normal

form otherwise.

A very simple example are the constant functions. The term

C k ≡ 𝜆x . k 𝜆 -defines the function c k : N → N such that c (n) = k .
For C k n ≡ (𝜆x . k )n →− k for any n. The identity function is 𝜆 -
defined by 𝜆x . x. More complex functions are of course harder
to define, and often require a lot of ingenuity. So it is perhaps
surprising that every computable function is 𝜆 -definable. The
converse is also true: if a function is 𝜆 -definable, it is computable.

8.7 𝜆 -Definable Arithmetical Functions

Proposition 8.6. The successor function succ is 𝜆 -definable.

Proof. A term that 𝜆 -defines the successor function is

Succ ≡ 𝜆a. 𝜆 f x . f (a f x).

Succ is a function that accepts as argument a number a, and

evaluates to another function, 𝜆 f x . f (a f x). That function is not
itself a Church numeral. However, if the argument a is a Church
numeral, it reduces to one. Consider:

(𝜆a. 𝜆 f x . f (a f x)) n →
− 𝜆 f x . f (n f x).

The embedded term n f x is a redex, since n is 𝜆 f x . f n x. So

− f n x and so, for the entire term we have
nf x →

→ 𝜆 f x . f ( f n (x)),
Succ n →
−

i.e., n + 1. □
179 8.7. 𝜆 -DEFINABLE ARITHMETICAL FUNCTIONS

Proposition 8.7. The addition function add is 𝜆 -definable.

Proof. Addition is 𝜆 -defined by the terms

Add ≡ 𝜆ab . 𝜆 f x . a f (b f x)

or, alternatively,

Add′ ≡ 𝜆ab . a Succ b .

The first addition works as follows: Add first accept two numbers
a and b. The result is a function that accepts f and x and returns
a f (b f x). If a and b are Church numerals n and m, this reduces
to f n+m (x), which is identical to f n ( f m (x)). Or, slowly:

(𝜆ab . 𝜆 f x . a f (b f x))n m →
− 𝜆 f x . n f (m f x)
− 𝜆 f x . n f ( f m x)
→
− 𝜆 f x . f n ( f m x) ≡ n + m.
→

The second representation of addition Add′ works differently:

Applied to two Church numerals n and m,

Add′n m →
− n Succ m.

But n f x always reduces to f n (x). So,

→ Succn (m).
n Succ m →
−

And since Succ 𝜆 -defines the successor function, and the succes-
sor function applied n times to m gives n + m, this in turn reduces
to n + m. □

Proposition 8.8. Multiplication is 𝜆 -definable by the term

Mult ≡ 𝜆ab . 𝜆 f x . a (b f )x
180 CHAPTER 8. THE LAMBDA CALCULUS

Proof. To see how this works, suppose we apply Mult to Church

numerals n and m: Mult n m reduces to 𝜆 f x . n (m f )x. The term
m f defines a function which applies f to its argument m times.
Consequently, n (m f )x applies the function “apply f m times”
itself n times to x. In other words, we apply f to x, n · m times.
But the resulting normal term is just the Church numeral nm. □

We can actually simplify this term further by 𝜂-reduction:

Mult ≡ 𝜆ab . 𝜆 f . a (b f ).

The definition of exponentiation as a 𝜆 -term is surprisingly

simple:
Exp ≡ 𝜆be . eb .
The first argument b is the base and the second e is the exponent.
Intuitively, e f is f e by our encoding of numbers. If you find it
hard to understand, we can still define exponentiation also by
iterated multiplication:

Exp′ ≡ 𝜆be . e (Mult b)1.

Predecessor and subtraction on Church numeral is not as sim-

ple as we might think: it requires encoding of pairs.

8.8 Pairs and Predecessor

Definition 8.9. The pair of M and N (written ⟨M ,N ⟩) is defined
as follows:
⟨M ,N ⟩ ≡ 𝜆 f . f M N .

Intuitively it is a function that accepts a function, and applies

that function to the two elements of the pair. Following this idea
we have this constructor, which takes two terms and returns the
pair containing them:

Pair ≡ 𝜆mn. 𝜆 f . f mn
181 8.9. TRUTH VALUES AND RELATIONS

Given a pair, we also want to recover its elements. For this we

need two access functions, which accept a pair as argument and
return the first or second elements in it:
Fst ≡ 𝜆 p. p (𝜆mn. m)
Snd ≡ 𝜆 p. p (𝜆mn. n)
Now with pairs we can 𝜆 -define the predecessor function:
Pred ≡ 𝜆n. Fst(n (𝜆 p. ⟨Snd p, Succ(Snd p)⟩)⟨0, 0⟩)
Remember that n f x reduces to f n (x); in this case f is a function
that accepts a pair p and returns a new pair containing the second
component of p and the successor of the second component; x is
the pair ⟨0, 0⟩. Thus, the result is ⟨0, 0⟩ for n = 0, and ⟨n − 1,n⟩
otherwise. Pred then returns the first component of the result.
Subtraction can be defined as Pred applied to a, b times:
Sub ≡ 𝜆ab . bPred a.

8.9 Truth Values and Relations

We can encode truth values in the pure lambda calculus as fol-
lows:
true ≡ 𝜆x . 𝜆 y . x
false ≡ 𝜆x . 𝜆 y . y
Truth values are represented as selectors, i.e., functions that
accept two arguments and returning one of them. The truth value
true selects its first argument, and false its second. For example,
true M N always reduces to M , while false M N always reduces
to N .

Definition 8.10. We call a relation R ⊆ Nn 𝜆 -definable if there

is a term R such that
𝛽
→ true
R n 1 . . . nk −
→
182 CHAPTER 8. THE LAMBDA CALCULUS

whenever R (n 1 , . . . ,nk ) and

𝛽
→ false
R n 1 . . . nk −
→

otherwise.

For instance, the relation IsZero = {0} which holds of 0 and

0 only, is 𝜆 -definable by

IsZero ≡ 𝜆n. n (𝜆x . false) true.

How does it work? Since Church numerals are defined as iter-

ators (functions which apply their first argument n times to the
second), we set the initial value to be true, and for every step of
iteration, we return false regardless of the result of the last iter-
ation. This step will be applied to the initial value n times, and
the result will be true if and only if the step is not applied at all,
i.e., when n = 0.
On the basis of this representation of truth values, we can
further define some truth functions. Here are two, the represen-
tations of negation and conjunction:

Not ≡ 𝜆x . x false true

And ≡ 𝜆x . 𝜆 y . xy false

The function “Not” accepts one argument, and returns true if the
argument is false, and false if the argument is true. The function
“And” accepts two truth values as arguments, and should return
true iff both arguments are true. Truth values are represented
as selectors (described above), so when x is a truth value and is
applied to two arguments, the result will be the first argument
if x is true and the second argument otherwise. Now And takes
its two arguments x and y, and in return passes y and false to
its first argument x. Assuming x is a truth value, the result will
evaluate to y if x is true, and to false if x is false, which is just
what is desired.
183 8.10. PRIMITIVE RECURSIVE FUNCTIONS ARE 𝜆 -DEFINABLE

Note that we assume here that only truth values are used as
arguments to And. If it is passed other terms, the result (i.e., the
normal form, if it exists) may well not be a truth value.

8.10 Primitive Recursive Functions are

𝜆 -Definable
Recall that the primitive recursive functions are those that can be
defined from the basic functions zero, succ, and Pin by composi-
tion and primitive recursion.

Lemma 8.11. The basic primitive recursive functions zero, succ, and
projections Pin are 𝜆 -definable.

Proof. They are 𝜆 -defined by the following terms:

Zero ≡ 𝜆a. 𝜆 f x . x
Succ ≡ 𝜆a. 𝜆 f x . f (a f x)
Projin ≡ 𝜆x 0 . . . x n−1 . x i □

Lemma 8.12. Suppose the k -ary function f , and n-ary functions

g 0 , . . . , g k −1 , are 𝜆 -definable by terms F , G 0 , . . . , G k , and h is defined
from them by composition. Then H is 𝜆 -definable.

Proof. h can be 𝜆 -defined by the term

H ≡ 𝜆x 0 . . . x n−1 . F (G 0 x 0 . . . x n−1 ) . . . (G k −1 x 0 . . . x n−1 )

We leave verification of this fact as an exercise. □

Note that Lemma 8.12 did not require that f and g 0 , . . . , g k −1

are primitive recursive; it is only required that they are total and
𝜆 -definable.
184 CHAPTER 8. THE LAMBDA CALCULUS

Lemma 8.13. Suppose f is an n-ary function and g is an n + 2-ary

function, they are 𝜆 -definable by terms F and G , and the function h is
defined from f and g by primitive recursion. Then h is also 𝜆 -definable.

Proof. Recall that h is defined by

h (x 1 , . . . ,x n , 0) = f (x 1 , . . . ,x n )
h (x 1 , . . . ,x n , y + 1) = h (x 1 , . . . ,x n , y,h (x 1 , . . . ,x n , y)).
Informally speaking, the primitive recursive definition iterates
the application of the function h y times and applies it to
f (x 1 , . . . ,x n ). This is reminiscent of the definition of Church nu-
merals, which is also defined as a iterator.
For simplicity, we give the definition and proof for a single
additional argument x. The function h is 𝜆 -defined by:
H ≡𝜆x . 𝜆 y . Snd(yD ⟨0,F x⟩)

where

D ≡𝜆 p. ⟨Succ(Fst p), (G x (Fst p) (Snd p))⟩

The iteration state we maintain is a pair, the first of which is the
current y and the second is the corresponding value of h. For
every step of iteration we create a pair of new values of y and
h; after the iteration is done we return the second part of the
pair and that’s the final h value. We now prove this is indeed a
representation of primitive recursion.
We want to prove that for any n and m, H n m → −→ h (n,m).
To do this we first show that if D n ≡ D [n/x], then D nm ⟨0,F n⟩ → −
→
⟨m,h (n,m)⟩ We proceed by induction on m.
If m = 0, we want D n0 ⟨0,F n⟩ → → ⟨0,h (n, 0)⟩. But D n0 ⟨0,F n⟩
−
just is ⟨0,F n⟩. Since F 𝜆 -defines f , this reduces to ⟨0, f (n)⟩, and
since f (n) = h (n, 0), this is ⟨0,h (n, 0)⟩
Now suppose that D nm ⟨0,F n⟩ → −→ ⟨m,h (n,m)⟩. We want to
show that D nm+1 ⟨0,F n⟩ → → ⟨m + 1,h (n,m + 1)⟩.
−
D nm+1 ⟨0,F n⟩ ≡ D n (D nm ⟨0,F n⟩)
185 8.11. FIXPOINTS

→
−
→ D n ⟨m,h (n,m)⟩ (by IH)
≡ (𝜆 p. ⟨Succ(Fst p), (G n (Fst p) (Snd p))⟩)⟨m,h (n,m)⟩
→
− ⟨Succ(Fst ⟨m,h (n,m)⟩),
(G n (Fst ⟨m,h (n,m)⟩) (Snd ⟨m,h (n,m)⟩))⟩
→
−
→ ⟨Succ m, (G n m h (n,m))⟩
→ ⟨m + 1, g (n,m,h (n,m))⟩
→
−

Since g (n,m,h (n,m)) = h (n,m + 1), we are done.

Finally, consider

H n m ≡ 𝜆x . 𝜆 y . Snd(y (𝜆 p.⟨Succ(Fst p), (G x (Fst p) (Snd p))⟩)⟨0,F x⟩)

nm
→ Snd(m (𝜆 p.⟨Succ(Fst p), (G n (Fst p) (Snd p))⟩) ⟨0,F n⟩)
→
−
⏞ˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉ⏟⏟ˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉ⏞
Dn

≡ Snd(m D n ⟨0,F n⟩)

→ Snd (D nm ⟨0,F n⟩)
→
−
→ Snd ⟨m,h (n,m)⟩
→
−
→
−
→ h (n,m). □

Proposition 8.14. Every primitive recursive function is 𝜆 -definable.

Proof. By Lemma 8.11, all basic functions are 𝜆 -definable, and

by Lemma 8.12 and Lemma 8.13, the 𝜆 -definable functions are
closed under composition and primitive recursion. □

8.11 Fixpoints
Suppose we wanted to define the factorial function by recursion
as a term Fac with the following property:

Fac ≡ 𝜆n. IsZero n 1(Mult n (Fac(Pred n)))

186 CHAPTER 8. THE LAMBDA CALCULUS

That is, the factorial of n is 1 if n = 0, and n times the factorial

of n − 1 otherwise. Of course, we cannot define the term Fac
this way since Fac itself occurs in the right-hand side. Such recur-
sive definitions involving self-reference are not part of the lambda
calculus. Defining a term, e.g., by
Mult ≡ 𝜆ab . a (Add a)0
only involves previously defined terms in the right-hand side, such
as Add. We can always remove Add by replacing it with its defin-
ing term. This would give the term Mult as a pure lambda term; if
Add itself involved defined terms (as, e.g., Add′ does), we could
continue this process and finally arrive at a pure lambda term.
However this is not true in the case of recursive definitions
like the one of Fac above. If we replace the occurrence of Fac on
the right-hand side with the definition of Fac itself, we get:
Fac ≡ 𝜆n. IsZero n 1
(Mult n ((𝜆n. IsZero n 1 (Mult n (Fac(Pred n)))) (Pred n)))
and we still haven’t gotten rid of Fac on the right-hand side.
Clearly, if we repeat this process, the definition keeps growing
longer and the process never results in a pure lambda term. Thus
this way of defining factorial (or more generally recursive func-
tions) is not feasible.
The recursive definition does tell us something, though: If f
were a term representing the factorial function, then the term
Fac′ ≡ 𝜆 g . 𝜆n. IsZero n 1 (Mult n (g (Predn)))
applied to the term f , i.e., Fac′ f , also represents the factorial
function. That is, if we regard Fac′ as a function accepting a
function and returning a function, the value of Fac′ f is just f ,
provided f is the factorial. A function f with the property that
𝛽
Fac′ f = f is called a fixpoint of Fac′. So, the factorial is a fixpoint
of Fac′.
There are terms in the lambda calculus that compute the fix-
points of a given term, and these terms can then be used to turn
a term like Fac′ into the definition of the factorial.
187 8.11. FIXPOINTS

Definition 8.15. The Y-combinator is the term:

Y ≡ (𝜆ux . x (uux)) (𝜆ux . x (uux)).

Theorem 8.16. Y has the property that Y g →

−
→ g (Y g ) for any term
g . Thus, Y g is always a fixpoint of g .

Proof. Let’s abbreviate (𝜆ux . x (uux)) by U , so that Y ≡ U U .

Then

Y g ≡ (𝜆ux . x (uux))U g
→
−
→ (𝜆x . x (U U x))g
→
−
→ g (U U g ) ≡ g (Y g ).
𝛽
Since g (Y g ) and Y g both reduce to g (Y g ), g (Y g ) = Y g , so Y g
is a fixpoint of g . □

Of course, since Y g is a redex, the reduction can continue

indefinitely:

Yg→
−
→ g (Y g )
→
−
→ g (g (Y g ))
→
−
→ g (g (g (Y g )))
...

So we can think of Y g as g applied to itself infinitely many times.

If we apply g to it one additional time, we—so to speak—aren’t
doing anything extra; g applied to g applied infinitely many times
to Y g is still g applied to Y g infinitely many times.
Note that the above sequence of 𝛽 -reduction steps starting
with Y g is infinite. So if we apply Y g to some term, i.e., consider
(Y g )N , that term will also reduce to infinitely many different
terms, namely (g (Y g ))N , (g (g (Y g )))N , . . . . It is nevertheless
possible that some other sequence of reduction steps does termi-
nate in a normal form.
188 CHAPTER 8. THE LAMBDA CALCULUS

Take the factorial for instance. Define Fac as Y Fac′ (i.e., a

fixpoint of Fac′). Then:

→ Y Fac′ 3
Fac 3 →
−
→ Fac′ (Y Fac′) 3
→
−
≡ (𝜆x . 𝜆n. IsZero n 1 (Mult n (x (Pred n)))) Fac 3
→ IsZero 3 1 (Mult 3 (Fac(Pred 3)))
→
−
→ Mult 3 (Fac 2).
→
−

Similarly,

Fac 2 →
→ Mult 2 (Fac 1)
−
Fac 1 →
→ Mult 1 (Fac 0)
−

but

→ Fac′ (Y Fac′) 0
Fac 0 →
−
≡ (𝜆x . 𝜆n. IsZero n 1 (Mult n (x (Pred n)))) Fac 0
→ IsZero 0 1 (Mult 0 (Fac(Pred 0))).
→
−
→ 1.
→
−

So together

Fac 3 →
→ Mult 3 (Mult 2 (Mult 1 1)).
−

What goes for Fac′ goes for any recursive definition. Suppose
we have a recursive equation
𝛽
g x1 . . . xn = N

where N may contain g and x 1 , . . . , x n . Then there is always a

term G ≡ (Y 𝜆 g . 𝜆x 1 . . . x n . N ) such that

𝛽
G x 1 . . . x n = N [G /g ].
189 8.12. MINIMIZATION

For by the fixpoint theorem,

G ≡ (Y 𝜆 g . 𝜆x 1 . . . x n . N ) →
−
→ 𝜆 g . 𝜆x 1 . . . x n . N (Y 𝜆 g . 𝜆x 1 . . . x n . N )
≡ (𝜆 g . 𝜆x 1 . . . x n . N ) G

and consequently

G x1 . . . xn →
−
→ (𝜆 g . 𝜆x 1 . . . x n . N ) G x 1 . . . x n
→
−
→ (𝜆x 1 . . . x n . N [G /g ]) x 1 . . . x n
→
−
→ N [G /g ].

The Y combinator of Definition 8.15 is due to Alan Tur-

ing. Alonzo Church had proposed a different version which we’ll
call YC :
YC ≡ 𝜆 g . (𝜆x . g (xx)) (𝜆x . g (xx)).
𝛽
Church’s combinator is a bit weaker than Turing’s in that Y g =
𝛽
g (Y g ) but not Y g −
→→ g (Y g ). Let V be the term 𝜆x . g (xx), so
that YC ≡ 𝜆 g .V V . Then

→ g (V V ) and thus
V V ≡ (𝜆x . g (xx))V →
−
YC g ≡ (𝜆 g .V V )g →
− → g (V V ), but also
→ VV →
−
g (YC g ) ≡ g ((𝜆 g .V V )g ) →
−
→ g (V V ).

In other words, YC g and g (YC g ) reduce to a common term

𝛽
g (V V ); so YC g = g (YC g ). This is often enough for applications.

8.12 Minimization
The general recursive functions are those that can be obtained
from the basic functions zero, succ, Pin by composition, primitive
recursion, and regular minimization. To show that all general re-
cursive functions are 𝜆 -definable we have to show that any func-
tion defined by regular minimization from a 𝜆 -definable function
is itself 𝜆 -definable.
190 CHAPTER 8. THE LAMBDA CALCULUS

Lemma 8.17. If f (x 1 , . . . ,x k , y) is regular and 𝜆 -definable, then

g defined by

g (x 1 , . . . ,x k ) = 𝜇y f (x 1 , . . . ,x k , y) = 0

is also 𝜆 -definable.

Proof. Suppose the lambda term F 𝜆 -defines the regular function

f (x⃗ , y). To 𝜆 -define h we use a search function and a fixpoint
combinator:
Search ≡ 𝜆 g . 𝜆 f x⃗ y . IsZero( f x⃗ y) y (g x⃗ (Succ y)
H ≡ 𝜆x⃗ . (Y Search)F x⃗ 0,
where Y is any fixpoint combinator. Informally speaking,
Search is a self-referencing function: starting with y, test
whether f x⃗ y is zero: if so, return y, otherwise call itself with
Succ y. Thus (Y Search)F n 1 . . . nk 0 returns the least m for which
f (n 1 , . . . ,nk ,m) = 0.
Specifically, observe that
(Y Search)F n 1 . . . nk m →
−
→m

if f (n 1 , . . . ,nk ,m) = 0, or

→ (Y Search)F n1 . . . nk m + 1
→
−

otherwise. Since f is regular, f (n 1 , . . . ,nk , y) = 0 for some y, and

so

(Y Search)F n1 . . . nk 0 →
−
→ h (n 1 , . . . ,nk ). □

Proposition 8.18. Every general recursive function is 𝜆 -definable.

Proof. By Lemma 8.11, all basic functions are 𝜆 -definable, and

by Lemma 8.12, Lemma 8.13, and Lemma 8.17, the 𝜆 -definable
functions are closed under composition, primitive recursion, and
regular minimization. □
191 8.13. PARTIAL RECURSIVE FUNCTIONS ARE 𝜆 -DEFINABLE

8.13 Partial Recursive Functions are

𝜆 -Definable
Partial recursive functions are those obtained from the basic func-
tions by composition, primitive recursion, and unbounded mini-
mization. They differ from general recursive function in that the
functions used in unbounded search are not required to be reg-
ular. Not requiring regularity means that functions defined by
minimization may sometimes not be defined.
At first glance it might seem that the same methods used
to show that the (total) general recursive functions are all 𝜆 -
definable can be used to prove that all partial recursive func-
tions are 𝜆 -definable. For instance, the composition of f with
g is 𝜆 -defined by 𝜆x . F (G x) if f and g are 𝜆 -defined by terms
F and G , respectively. However, when the functions are partial,
this is problematic. When g (x) is undefined, meaning G x has no
normal form. In most cases this means that F (G x) has no nor-
mal forms either, which is what we want. But consider when F is
𝜆x . 𝜆 y . y, in which case F (G x) does have a normal form (𝜆 y . y).
This problem is not insurmountable, and there are ways to 𝜆 -
define all partial recursive functions in such a way that undefined
values are represented by terms without a normal form. These
ways are, however, somewhat more complicated and less intuitive
than the approach we have taken for general recursive functions.
We record the theorem here without proof:

Theorem 8.19. All partial recursive functions are 𝜆 -definable.

8.14 𝜆 -Definable Functions are Recursive

Not only are all partial recursive functions 𝜆 -definable, the con-
verse is true, too. That is, all 𝜆 -definable functions are partial
recursive.
192 CHAPTER 8. THE LAMBDA CALCULUS

Theorem 8.20. If a partial function f is 𝜆 -definable, it is partial

recursive.
Proof. We only sketch the proof. First, we arithmetize 𝜆 -terms,
i.e., systematially assign Gödel numbers to 𝜆 -terms as using the
usual power-of-primes coding of sequences. Then we define a par-
tial recursive function normalize(t ) operating on the Gödel num-
ber t of a lambda term as argument, and which returns the Gödel
number of the normal form if it has one, or is undefined other-
wise. Then define two partial recursive functions toChurch and
fromChurch that maps natural numbers to and from the Gödel
numbers of the corresponding Church numeral.
Using these recursive functions, we can define the function f
as a partial recursive function. There is a lambda term F
that 𝜆 -defines f . To compute f (n1 , . . . ,nk ), first obtain the
Gödel numbers of the corresponding Church numerals using
toChurch(ni ), append these to # F # to obtain the Gödel number
of the term F n 1 . . . nk . Now use normalize on this Gödel number.
If f (n 1 , . . . ,nk ) is defined, F n1 . . . nk has a normal form (which
must be a Church numeral), and otherwise it has no normal form
(and so
normalize( # F n 1 . . . nk # )
is undefined). Finally, use fromChurch on the Gödel number of
the normalized term. □

Problems
Problem 8.1. The term
Succ′ ≡ 𝜆n. 𝜆 f x . n f ( f x)
𝜆 -defines the successor function. Explain why.

Problem 8.2. Multiplication can be 𝜆 -defined by the term

Mult′ ≡ 𝜆ab . a (Add a)0.
Explain why this works.
193 8.14. 𝜆 -DEFINABLE FUNCTIONS ARE RECURSIVE

Problem 8.3. Explain why the access functions Fst and Snd
work.

Problem 8.4. Define the functions Or and Xor representing the

truth functions of inclusive and exclusive disjunction using the
encoding of truth values as 𝜆 -terms.

Problem 8.5. Complete the proof of Lemma 8.12 by showing

that H n 0 . . . nn−1 →
−
→ h (n 0 , . . . ,nn−1 ).
APPENDIX A

Derivations in
Arithmetic
Theories
When we showed that all general recursive functions are repre-
sentable in Q , and in the proofs of the incompleteness theorems,
we claimed that various things are provable in Q and PA. The
proofs of these claims, however, just gave the arguments infor-
mally without exhibiting actual derivations in natural deduction.
We provide some of these derivations in this capter.
For instance, in Lemma 4.15 we proved that, for all n and
m ∈ N, Q ⊢ (n + m) = n + m. We did this by induction on m.

Proof of Lemma 4.15. Base case: m = 0. Then what has to be

proved is that, for all n, Q ⊢ n + 0 = n + 0. Since 0 is just 0 and
n + 0 is n, this amounts to showing that Q ⊢ (n + 0) = n. The
derivation
∀x (x + 0) = x
∀Elim
(n + 0) = n

is a natural deduction derivation of (n + 0) = n with one undis-

charged assumption, and that undischarged assumption is an ax-

194
195

iom of Q .
Inductive step: Suppose that, for any n, Q ⊢ (n + m) = n + m
(say, by a derivation 𝛿n,m ). We have to show that also Q ⊢ (n +
m + 1) = n + m + 1. Note that m + 1 ≡ m ′, and that n + m + 1 ≡
n + m ′. So we are looking for a derivation of (n + m ′) = n + m ′
from the axioms of Q . Our derivation may use the derivation
𝛿n,m which exists by inductive hypothesis.

∀x ∀y (x + y ′) = (x + y) ′
𝛿n,m ∀Elim
∀y (n + y ′) = (n + y) ′
∀Elim
(n + m) = n + m (n + m ′) = (n + m) ′
=Elim
(n + m ′) = n + m ′

In the last =Elim inference, we replace the subterm n + m of the

right side (n + m) ′ of the right premise by the term n + m. □

In Lemma 4.22, we showed that Q ⊢ ∀x ¬x < 0. What does

an actual derivation look like?

Proof of Lemma 4.22. To prove a universal claim like this, we use

∀Intro, which requires a derivation of ¬a < 0. Looking at axiom
Q 8 , this means proving ¬∃z (z ′ + a) = 0. Specifically, if we had a
proof of the latter, Q 8 would allow us to prove the former (recall
that A ↔ B is short for (A → B) ∧ (B → A).
∀x ∀y (x < y ↔ ∃z (z ′ + x) = y)
∀Elim
∀y (a < y ↔ ∃z (z ′ + a) = y)
∀Elim
a < 0 ↔ ∃z (z ′ + a) = 0
∧Elim
a < 0 → ∃z (z ′ + a) = 0 [a < 0] 1
′ ′ →Elim
¬∃z (z + a) = 0 ∃z (z + a) = 0
⊥ ¬Elim
1
¬a < 0 ¬Intro

This is a derivation of ¬a < 0 from ¬∃z (z ′ + a) = 0 (and Q 8 );

let’s call it 𝛿1 .
Now how do we prove ¬∃z (z ′ +a) = 0 from the axioms of Q ?
To prove a negated claim like this, we’d need a derivation of the
form
196 APPENDIX A. DERIVATIONS IN ARITHMETIC THEORIES

[∃z (z ′ + a) = 0] 2

⊥
2 ¬Intro
¬∃z (z ′ + a) = 0

To get a contradiction from an existential claim, we introduce

a constant b for the existentially quantified variable z and use
∃Elim:
[(b ′ + a) = 0] 3
𝛿2
[∃z (z ′ + a) = 0] 2 ⊥
3
⊥ ∃Elim
2 ′ ¬Intro
¬∃z (z + a) = 0

Now the task is to fill in 𝛿2 , i.e., prove ⊥ from (b ′ + a) = 0 and

the axioms of Q . Q 2 says that 0 can’t be the successor of some
number, so one way of doing that would be to show that (b ′ + a)
is equal to the successor of some number. Since that expression
itself is a sum, the axioms for addition must come into play. If
a = 0, Q 5 would tell us that (b ′ +a) = b ′, i.e., b ′ +a is the successor
of some number, namely of b. On the other hand, if a = c ′ for
some c , then (b ′ + a) = (b ′ + c ′) by =Elim, and (b ′ + c ′) = (b ′ + c ) ′
by Q 6 . So again, b ′ + a is the successor of a number—in this case,
b ′ +c . So the strategy is to divide the task into these two cases. We
also have to verify that Q proves that one of these cases holds,
i.e., Q ⊢ a = 0 ∨ ∃y (a = y ′), but this follows directly from Q 3 by
∀Elim. Here are the two cases:
Case 1: Prove ⊥ from a = 0 and (b ′ + a) = 0 (and axioms Q 2 ,
Q 5 ):
a=0 (b ′ + a) = 0
=Elim
∀x (x + 0) = x (b ′ + 0) = 0
∀Elim
∀x ¬0 = x′ (b ′ + 0) = b ′ 0 = (b ′ + 0)
∀Elim =Elim
¬0 = b ′ 0=b ′
⊥ ¬Elim
197

Call this derivation 𝛿3 . (We’ve abbreviated the derivation of 0 =

(b ′ + 0) from (b ′ + 0) = 0 by a double inference line.)
Case 2: Prove ⊥ from ∃y a = y ′ and (b ′ + a) = 0 (and axioms
Q 2 , Q 6 ). We first show how to derive ⊥ from a = c ′ and (b ′+a) = 0.

∀x ∀y (x + y ′ ) = (x + y) ′
∀Elim
a = c ′ (b ′ + a) = 0 ∀y (b ′ + y ′ ) = (b ′ + y) ′
=Elim ∀Elim
∀x ¬0 = x ′ (b ′ + c ′ ) = 0 (b ′ + c ′ ) = (b ′ + c ) ′
′ ′ ∀Elim =Elim
¬0 = (b + c ) 0 = (b + c ) ′
′
⊥ ¬Elim

Call this 𝛿4 . We get the required derivation 𝛿5 by applying ∃Elim

and discharging the assumption a = c ′:

[a = c ′] 6 (b ′ + a) = 0
𝛿4
∃y a = y ′ ⊥
6
⊥ ∃Elim

Putting everything together, the full proof looks like this:

[a = 0] 7 [∃y a = y ′ ] 7 ⎫
⎪
⎪
⎪
′ 3 ′ 3 ⎪
[(b + a) = 0] [(b + a) = 0] ⎪
⎪
∀x (x = 0 ∨ ⎪
⎪
⎪
⎪
∃y (a = y ′ ))
⎬
⎪
𝛿3 𝛿5 𝛿2
∀Elim ⎪
a =0∨
⎪
⎪
⎪
⊥ ⊥
⎪
⎪
∃y (a = y ′ )
⎪
⎪
⎪
⎪
[∃z (z ′ + a) = 0] 2 7
⊥ ∨Elim ⎪
⎭
3
⊥ ∃Elim
2 ¬Intro
¬∃z (z ′ + a) = 0

𝛿1

¬a < 0
∀Intro
∀x ¬x < 0
□
198 APPENDIX A. DERIVATIONS IN ARITHMETIC THEORIES

In the proof of Theorem 5.7, we defined RProv (y) as

∃x ( Prf (x, y) ∧ ∀z (z < x → ¬Ref (z , y))).

Prf (x, y) is the formula representing the proof relation of T (a

consistent, axiomatizable extension of Q ) in Q , and Ref (z , y) is
the formula representing the refutation relation. That means that
if n is the Gödel number of a proof of A, then Q ⊢ Prf (n, ⌜A⌝),
and otherwise Q ⊢ ¬Prf (n, ⌜A⌝). Similarly, if n is the Gödel
number of a proof of ¬A, then Q ⊢ Ref (n, ⌜A⌝), and otherwise
Q ⊢ ¬Ref (n, ⌜A⌝). We use the Diagonal Lemma to find a sen-
tence R such that Q ⊢ R ↔ ¬RProv (⌜R⌝). Rosser’s Theorem
states that T ⊬ R and T ⊬ ¬R. Both claims were proved indi-
rectly: we show that if T ⊢ R, T is inconsistent, i.e., T ⊢ ⊥, and
the same if T ⊢ ¬R.

Proof of Theorem 5.7. First we prove something things about <. By

Lemma 4.23, we know that Q ⊢ ∀x (x < n + 1 → (x = 0 ∨ · · · ∨ x =
n)) for every n. So of course also (if n > 1), Q ⊢ ∀x (x < n → (x =
0∨· · ·∨x = n − 1)). We can use this to derive a = 0∨· · ·∨a = n − 1
from a < n:

∀x (x < n → (x = 0 ∨ · · · ∨ x = n − 1))
∀Elim
a<n a < n → (a = 0 ∨ · · · ∨ a = n − 1)
→Elim
a = 0∨···∨a = n −1

Let’s call this derivation 𝜆 1 .

Now, to show that T ⊬ R, we assume that T ⊢ R (with a
derivation 𝛿) and show that T then would be inconsistent. Let n
be the Gödel number of 𝛿. Since Prf represents the proof relation
in Q , there is a derivation 𝛿1 of Prf (n, ⌜R⌝). Furthermore, no k <
n is the Gödel number of a refutation of R since T is assumed to
be consistent, so for each k < n, Q ⊢ ¬Ref (k , ⌜R⌝); let 𝜌 k be the
corresponding derivation. We get a derivation of RProv (⌜R⌝):
199

[a < n] 1

𝜆1 𝜌k

a = 0∨... [a = k ] 2 ¬Ref (k , ⌜R⌝)

=Elim
∨a = n −1 ... ¬Ref (a, ⌜R⌝) ...
2 ∨Elim∗
𝛿1 ¬Ref (a, ⌜R⌝)
1 →Intro
a < n → ¬Ref (a, ⌜R⌝)
∀Intro
Prf (n, ⌜R⌝) ∀z (z < n → ¬Ref (z , ⌜R⌝))
∧Intro
Prf (n, ⌜R⌝) ∧ ∀z (z < n → ¬Ref (z , ⌜R⌝))
∃Intro
∃x ( Prf (x, ⌜R⌝) ∧ ∀z (z < x → ¬Ref (z , ⌜R⌝)))

(We abbreviate multiple applications of ∨Elim by ∨Elim∗ above.)

We’ve shown that if T ⊢ R there would be a derivation
of RProv (⌜R⌝). Then, since T ⊢ R ↔ ¬RProv (⌜R⌝), also
T ⊢ RProv (⌜R⌝) → ¬R, we’d have T ⊢ ¬R and T would be incon-
sistent.
Now let’s show that T ⊬ ¬R. Again, suppose it did. Then
there is a derivation 𝜌 of ¬R with Gödel number m—a refutation
of R—and so Q ⊢ Ref (m, ⌜R⌝) by a derivation 𝜌 1 . Since we
assume T is consistent, T ⊬ R. So for all k , k is not a Gödel
number of a derivation of R, and hence Q ⊢ ¬Prf (k , ⌜R⌝) by a
derivation 𝜋k . So we have:
200 APPENDIX A. DERIVATIONS IN ARITHMETIC THEORIES

[ Prf (a, ⌜R⌝)] 1

[a = k ] 2
𝜆2
𝜋k′
a = 0∨ ···∨ ⊥ ⊥I 𝜌1
a = m ∨m < a ... m<a ... [m < a] 2
2 ∨Elim∗
m<a Ref (m, ⌜R⌝)
∧Intro
m < a ∧ Ref (m, ⌜R⌝)
∃Intro
∃z (z < a ∧ Ref (z , ⌜R⌝))
1 →Intro
Prf (a, ⌜R⌝) → ∃z (z < a ∧ Ref (z , ⌜R⌝))
∀Intro
∀x ( Prf (x, ⌜R⌝) → ∃z (z < x ∧ Ref (z , ⌜R⌝)))

¬∃x ( Prf (x, ⌜R⌝) ∧ ∀z (z < x → ¬Ref (z , ⌜R⌝)))

where 𝜋k′ is the derivation

𝜋k
a =k Prf (a, ⌜R⌝)
=Elim
¬Prf (k , ⌜R⌝) Prf (k , ⌜R⌝)
⊥ ¬Elim

and 𝜆 2 is

[a < m] 3

𝜆1

𝜆3 a = 0∨···∨

(a < m ∨ a =m−1 [a = m] 3 [m < a] 3

∨Intro∗
a = m) ∨ a = 0∨ ···∨ a = 0∨ ···∨ a = 0∨ ···∨
m<a a =m ∨m < a a =m ∨m < a a =m ∨m < a
3 ∨Elim2
a = 0∨···∨a = m ∨m < a

(The derivation 𝜆 3 exists by Lemma 4.24. We abbreviate re-

peated use of ∨Intro by ∨Intro∗ and the double use of ∨Elim to
201

derive a = 0 ∨ · · · ∨ a = m ∨ m < a from (a < m ∨ a = m) ∨ m < a

as ∨Elim2 .) □
APPENDIX B

First-order
Logic
B.1 First-Order Languages
Expressions of first-order logic are built up from a basic vocab-
ulary containing variables, constant symbols, predicate symbols and
sometimes function symbols. From them, together with logical con-
nectives, quantifiers, and punctuation symbols such as parenthe-
ses and commas, terms and formulas are formed.
Informally, predicate symbols are names for properties and
relations, constant symbols are names for individual objects, and
function symbols are names for mappings. These, except for
the identity predicate =, are the non-logical symbols and together
make up a language. Any first-order language L is determined
by its non-logical symbols. In the most general case, L contains
infinitely many symbols of each kind.
In the general case, we make use of the following symbols in
first-order logic:

1. Logical symbols

a) Logical connectives: ¬ (negation), ∧ (conjunction),

∨ (disjunction), → (conditional), ∀ (universal quanti-

202
203 B.1. FIRST-ORDER LANGUAGES

fier), ∃ (existential quantifier).

b) The propositional constant for falsity ⊥.
c) The two-place identity predicate =.
d) A countably infinite set of variables: v0 , v1 , v2 , . . .

2. Non-logical symbols, making up the standard language of

first-order logic

a) A countably infinite set of n-place predicate symbols

for each n > 0: An0 , An1 , An2 , . . .
b) A countably infinite set of constant symbols: c0 , c1 ,
c2 , . . . .
c) A countably infinite set of n-place function symbols
for each n > 0: f0n , f1n , f2n , . . .

3. Punctuation marks: (, ), and the comma.

Most of our definitions and results will be formulated for the

full standard language of first-order logic. However, depending
on the application, we may also restrict the language to only a
few predicate symbols, constant symbols, and function symbols.

Example B.1. The language LA of arithmetic contains a single

two-place predicate symbol <, a single constant symbol 0, one
one-place function symbol ′, and two two-place function sym-
bols + and ×.

Example B.2. The language of set theory LZ contains only the

single two-place predicate symbol ∈.

Example B.3. The language of orders L≤ contains only the two-

place predicate symbol ≤.

Again, these are conventions: officially, these are just aliases,

e.g., <, ∈, and ≤ are aliases for A20 , 0 for c0 , ′ for f01 , + for f02 , ×
for f12 .
204 APPENDIX B. FIRST-ORDER LOGIC

In addition to the primitive connectives and quantifiers in-

troduced above, we also use the following defined symbols: ↔
(biconditional), truth ⊤
A defined symbol is not officially part of the language, but
is introduced as an informal abbreviation: it allows us to abbre-
viate formulas which would, if we only used primitive symbols,
get quite long. This is obviously an advantage. The bigger ad-
vantage, however, is that proofs become shorter. If a symbol is
primitive, it has to be treated separately in proofs. The more
primitive symbols, therefore, the longer our proofs.
We might treat all the propositional operators and both quan-
tifiers as primitive symbols of the language. We might instead
choose a smaller stock of primitive symbols and treat the other
logical operators as defined. “Truth functionally complete” sets
of Boolean operators include {¬, ∨}, {¬, ∧}, and {¬, →}—these
can be combined with either quantifier for an expressively com-
plete first-order language.
You may be familiar with two other logical operators: the
Sheffer stroke | (named after Henry Sheffer), and Peirce’s ar-
row ↓, also known as Quine’s dagger. When given their usual
readings of “nand” and “nor” (respectively), these operators are
truth functionally complete by themselves.

B.2 Terms and Formulas

Once a first-order language L is given, we can define expressions
built up from the basic vocabulary of L. These include in partic-
ular terms and formulas.

Definition B.4 (Terms). The set of terms Trm(L) of L is de-

fined inductively by:

1. Every variable is a term.

2. Every constant symbol of L is a term.

205 B.2. TERMS AND FORMULAS

3. If f is an n-place function symbol and t1 , . . . , tn are terms,

then f (t1 , . . . ,tn ) is a term.

4. Nothing else is a term.

A term containing no variables is a closed term.

The constant symbols appear in our specification of the lan-

guage and the terms as a separate category of symbols, but they
could instead have been included as zero-place function symbols.
We could then do without the second clause in the definition of
terms. We just have to understand f (t1 , . . . ,tn ) as just f by itself
if n = 0.

Definition B.5 (Formula). The set of formulas Frm(L) of the

language L is defined inductively as follows:

1. ⊥ is an atomic formula.

2. If R is an n-place predicate symbol of L and t1 , . . . , tn are

terms of L, then R (t1 , . . . ,tn ) is an atomic formula.

3. If t1 and t2 are terms of L, then =(t1 ,t2 ) is an atomic for-

mula.

4. If A is a formula, then ¬A is formula.

5. If A and B are formulas, then (A ∧ B) is a formula.

6. If A and B are formulas, then (A ∨ B) is a formula.

7. If A and B are formulas, then (A → B) is a formula.

8. If A is a formula and x is a variable, then ∀x A is a formula.

9. If A is a formula and x is a variable, then ∃x A is a formula.

10. Nothing else is a formula.

The definitions of the set of terms and that of formulas are

206 APPENDIX B. FIRST-ORDER LOGIC

inductive definitions. Essentially, we construct the set of formu-

las in infinitely many stages. In the initial stage, we pronounce
all atomic formulas to be formulas; this corresponds to the first
few cases of the definition, i.e., the cases for ⊥, R (t1 , . . . ,tn ) and
=(t1 ,t2 ). “Atomic formula” thus means any formula of this form.
The other cases of the definition give rules for constructing
new formulas out of formulas already constructed. At the second
stage, we can use them to construct formulas out of atomic for-
mulas. At the third stage, we construct new formulas from the
atomic formulas and those obtained in the second stage, and so
on. A formula is anything that is eventually constructed at such
a stage, and nothing else.
By convention, we write = between its arguments and leave
out the parentheses: t1 = t2 is an abbreviation for =(t1 ,t2 ). More-
over, ¬=(t1 ,t2 ) is abbreviated as t1 ≠ t2 . When writing a formula
(B ∗C ) constructed from B, C using a two-place connective ∗, we
will often leave out the outermost pair of parentheses and write
simply B ∗ C .

Definition B.6. Formulas constructed using the defined opera-

tors are to be understood as follows:

1. ⊤ abbreviates ¬⊥.

2. A ↔ B abbreviates (A → B) ∧ (B → A).

If we work in a language for a specific application, we will

often write two-place predicate symbols and function symbols
between the respective terms, e.g., t1 < t2 and (t1 + t2 ) in the
language of arithmetic and t1 ∈ t2 in the language of set the-
ory. The successor function in the language of arithmetic is even
written conventionally after its argument: t ′. Officially, however,
these are just conventional abbreviations for A20 (t1 ,t2 ), f02 (t1 ,t2 ),
A20 (t1 ,t2 ) and f 01 (t ), respectively.
207 B.3. FREE VARIABLES AND SENTENCES

Definition B.7 (Syntactic identity). The symbol ≡ expresses

syntactic identity between strings of symbols, i.e., A ≡ B iff A and
B are strings of symbols of the same length and which contain
the same symbol in each place.

The ≡ symbol may be flanked by strings obtained by con-

catenation, e.g., A ≡ (B ∨ C ) means: the string of symbols A is
the same string as the one obtained by concatenating an opening
parenthesis, the string B, the ∨ symbol, the string C , and a clos-
ing parenthesis, in this order. If this is the case, then we know
that the first symbol of A is an opening parenthesis, A contains
B as a substring (starting at the second symbol), that substring
is followed by ∨, etc.

B.3 Free Variables and Sentences

Definition B.8 (Free occurrences of a variable). The free oc-

currences of a variable in a formula are defined inductively as
follows:

1. A is atomic: all variable occurrences in A are free.

2. A ≡ ¬B: the free variable occurrences of A are exactly

those of B.

3. A ≡ (B ∗ C ): the free variable occurrences of A are those

in B together with those in C .

4. A ≡ ∀x B: the free variable occurrences in A are all of

those in B except for occurrences of x.

5. A ≡ ∃x B: the free variable occurrences in A are all of

those in B except for occurrences of x.
208 APPENDIX B. FIRST-ORDER LOGIC

Definition B.9 (Bound Variables). An occurrence of a vari-

able in a formula A is bound if it is not free.

Definition B.10 (Scope). If ∀x B is an occurrence of a subfor-

mula in a formula A, then the corresponding occurrence of B
in A is called the scope of the corresponding occurrence of ∀x.
Similarly for ∃x.
If B is the scope of a quantifier occurrence ∀x or ∃x in A, then
the free occurrences of x in B are bound in ∀x B and ∃x B. We
say that these occurrences are bound by the mentioned quantifier
occurrence.

Example B.11. Consider the following formula:

∃v0 A20 (v0 , v1 )

⏞ˉˉˉˉˉ⏟⏟ˉˉˉˉˉ⏞
B

B represents the scope of ∃v0 . The quantifier binds the occurence

of v0 in B, but does not bind the occurence of v1 . So v1 is a free
variable in this case.
We can now see how this might work in a more complicated
formula A:
D
⏟ˉˉˉ⏞⏞ˉˉˉ⏟
∀v0 (A10 (v0 ) → A20 (v0 , v1 )) →∃v1 (A21 (v0 , v1 ) ∨ ∀v0 ¬A11 (v0 ))
⏞ˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉ⏟⏟ˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉ⏞ ⏞ˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉ⏟⏟ˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉˉ⏞
B C

B is the scope of the first ∀v0 , C is the scope of ∃v1 , and D is the
scope of the second ∀v0 . The first ∀v0 binds the occurrences of v0
in B, ∃v1 the occurrence of v1 in C , and the second ∀v0 binds the
occurrence of v0 in D. The first occurrence of v1 and the fourth
occurrence of v0 are free in A. The last occurrence of v0 is free
in D, but bound in C and A.
209 B.4. SUBSTITUTION

Definition B.12 (Sentence). A formula A is a sentence iff it con-

tains no free occurrences of variables.

B.4 Substitution

Definition B.13 (Substitution in a term). We define s [t /x],

the result of substituting t for every occurrence of x in s , recur-
sively:

1. s ≡ c : s [t /x] is just s .

2. s ≡ y: s [t /x] is also just s , provided y is a variable and

y ̸≡ x.

3. s ≡ x: s [t /x] is t .

4. s ≡ f (t1 , . . . ,tn ): s [t /x] is f (t1 [t /x], . . . ,tn [t /x]).

Definition B.14. A term t is free for x in A if none of the free

occurrences of x in A occur in the scope of a quantifier that binds
a variable in t .

Example B.15.

1. v8 is free for v1 in ∃v3 A24 (v3 , v1 )

2. f12 (v1 , v2 ) is not free for v0 in ∀v2 A24 (v0 , v2 )

Definition B.16 (Substitution in a formula). If A is a for-

mula, x is a variable, and t is a term free for x in A, then A[t /x]
is the result of substituting t for all free occurrences of x in A.

1. A ≡ ⊥: A[t /x] is ⊥.

2. A ≡ P (t1 , . . . ,tn ): A[t /x] is P (t1 [t /x], . . . ,tn [t /x]).

210 APPENDIX B. FIRST-ORDER LOGIC

3. A ≡ t1 = t2 : A[t /x] is t1 [t /x] = t2 [t /x].

4. A ≡ ¬B: A[t /x] is ¬B [t /x].

5. A ≡ (B ∧ C ): A[t /x] is (B [t /x] ∧ C [t /x]).

6. A ≡ (B ∨ C ): A[t /x] is (B [t /x] ∨ C [t /x]).

7. A ≡ (B → C ): A[t /x] is (B [t /x] → C [t /x]).

8. A ≡ ∀y B: A[t /x] is ∀y B [t /x], provided y is a variable

other than x; otherwise A[t /x] is just A.

9. A ≡ ∃y B: A[t /x] is ∃y B [t /x], provided y is a variable

other than x; otherwise A[t /x] is just A.

Note that substitution may be vacuous: If x does not occur in

A at all, then A[t /x] is just A.
The restriction that t must be free for x in A is necessary
to exclude cases like the following. If A ≡ ∃y x < y and t ≡ y,
then A[t /x] would be ∃y y < y. In this case the free variable y
is “captured” by the quantifier ∃y upon substitution, and that is
undesirable. For instance, we would like it to be the case that
whenever ∀x B holds, so does B [t /x]. But consider ∀x ∃y x < y
(here B is ∃y x < y). It is sentence that is true about, e.g., the
natural numbers: for every number x there is a number y greater
than it. If we allowed y as a possible substitution for x, we would
end up with B [y/x] ≡ ∃y y < y, which is false. We prevent this by
requiring that none of the free variables in t would end up being
bound by a quantifier in A.
We often use the following convention to avoid cumbersume
notation: If A is a formula with a free variable x, we write A(x)
to indicate this. When it is clear which A and x we have in mind,
and t is a term (assumed to be free for x in A(x)), then we write
A(t ) as short for A(x) [t /x].
211 B.5. STRUCTURES FOR FIRST-ORDER LANGUAGES

B.5 Structures for First-order Languages

First-order languages are, by themselves, uninterpreted: the con-
stant symbols, function symbols, and predicate symbols have no
specific meaning attached to them. Meanings are given by spec-
ifying a structure. It specifies the domain, i.e., the objects which
the constant symbols pick out, the function symbols operate on,
and the quantifiers range over. In addition, it specifies which
constant symbols pick out which objects, how a function symbol
maps objects to objects, and which objects the predicate symbols
apply to. Structures are the basis for semantic notions in logic,
e.g., the notion of consequence, validity, satisfiablity. They are
variously called “structures,” “interpretations,” or “models” in
the literature.

Definition B.17 (Structures). A structure M, for a language L

of first-order logic consists of the following elements:

1. Domain: a non-empty set, |M|

2. Interpretation of constant symbols: for each constant symbol c

of L, an element c M ∈ |M|

3. Interpretation of predicate symbols: for each n-place predicate

symbol R of L (other than =), an n-place relation R M ⊆
|M| n

4. Interpretation of function symbols: for each n-place function

symbol f of L, an n-place function f M : |M| n → |M|

Example B.18. A structure M for the language of arithmetic

consists of a set, an element of |M|, 0M , as interpretation of the
constant symbol 0, a one-place function ′M : |M| → |M|, two two-
place functions +M and ×M , both |M| 2 → |M|, and a two-place
relation <M ⊆ |M| 2 .
An obvious example of such a structure is the following:

1. |N| = N
212 APPENDIX B. FIRST-ORDER LOGIC

2. 0N = 0

3. ′N (n) = n + 1 for all n ∈ N

4. +N (n,m) = n + m for all n,m ∈ N

5. ×N (n,m) = n · m for all n,m ∈ N

6. <N = {⟨n,m⟩ : n ∈ N,m ∈ N,n < m}

The structure N for LA so defined is called the standard model of

arithmetic, because it interprets the non-logical constants of LA
exactly how you would expect.
However, there are many other possible structures for LA . For
instance, we might take as the domain the set Z of integers instead
of N, and define the interpretations of 0, ′, +, ×, < accordingly.
But we can also define structures for LA which have nothing even
remotely to do with numbers.

Example B.19. A structure M for the language LZ of set theory

requires just a set and a single-two place relation. So technically,
e.g., the set of people plus the relation “x is older than y” could
be used as a structure for LZ , as well as N together with n ≥ m
for n,m ∈ N.
A particularly interesting structure for LZ in which the ele-
ments of the domain are actually sets, and the interpretation of
∈ actually is the relation “x is an element of y” is the structure
HF of hereditarily finite sets:

1. |HF| = ∅ ∪ ℘(∅) ∪ ℘(℘(∅)) ∪ ℘(℘(℘(∅))) ∪ . . . ;

2. ∈HF = {⟨x, y⟩ : x, y ∈ |HF| ,x ∈ y }.

The stipulations we make as to what counts as a structure

impact our logic. For example, the choice to prevent empty do-
mains ensures, given the usual account of satisfaction (or truth)
for quantified sentences, that ∃x (A(x) ∨ ¬A(x)) is valid—that
is, a logical truth. And the stipulation that all constant symbols
213 B.6. SATISFACTION OF A FORMULA IN A STRUCTURE

must refer to an object in the domain ensures that the existential

generalization is a sound pattern of inference: A(a), therefore
∃x A(x). If we allowed names to refer outside the domain, or to
not refer, then we would be on our way to a free logic, in which ex-
istential generalization requires an additional premise: A(a) and
∃x x = a, therefore ∃x A(x).

B.6 Satisfaction of a Formula in a Structure

The basic notion that relates expressions such as terms and for-
mulas, on the one hand, and structures on the other, are those
of value of a term and satisfaction of a formula. Informally, the
value of a term is an element of a structure—if the term is just a
constant, its value is the object assigned to the constant by the
structure, and if it is built up using function symbols, the value is
computed from the values of constants and the functions assigned
to the functions in the term. A formula is satisfied in a structure
if the interpretation given to the predicates makes the formula
true in the domain of the structure. This notion of satisfaction
is specified inductively: the specification of the structure directly
states when atomic formulas are satisfied, and we define when a
complex formula is satisfied depending on the main connective
or quantifier and whether or not the immediate subformulas are
satisfied. The case of the quantifiers here is a bit tricky, as the
immediate subformula of a quantified formula has a free variable,
and structures don’t specify the values of variables. In order to
deal with this difficulty, we also introduce variable assignments and
define satisfaction not with respect to a structure alone, but with
respect to a structure plus a variable assignment.
214 APPENDIX B. FIRST-ORDER LOGIC

Definition B.20 (Variable Assignment). A variable assign-

ment s for a structure M is a function which maps each variable
to an element of |M|, i.e., s : Var → |M|.

A structure assigns a value to each constant symbol, and a

variable assignment to each variable. But we want to use terms
built up from them to also name elements of the domain. For
this we define the value of terms inductively. For constant sym-
bols and variables the value is just as the structure or the variable
assignment specifies it; for more complex terms it is computed re-
cursively using the functions the structure assigns to the function
symbols.

Definition B.21 (Value of Terms). If t is a term of the lan-

guage L, M is a structure for L, and s is a variable assignment
for M, the value ValsM (t ) is defined as follows:

1. t ≡ c : ValsM (t ) = c M .

2. t ≡ x: ValsM (t ) = s (x).

3. t ≡ f (t1 , . . . ,tn ):

ValsM (t ) = f M
(ValsM (t1 ), . . . , ValsM (tn )).

Definition B.22 (x-Variant). If s is a variable assignment for a

structure M, then any variable assignment s ′ for M which differs
from s at most in what it assigns to x is called an x-variant of s .
If s ′ is an x-variant of s we write s ∼x s ′.

Note that an x-variant of an assignment s does not have to

assign something different to x. In fact, every assignment counts
as an x-variant of itself.
215 B.6. SATISFACTION OF A FORMULA IN A STRUCTURE

Definition B.23 (Satisfaction). Satisfaction of a formula A in

a structure M relative to a variable assignment s , in symbols:
M,s ⊨ A, is defined recursively as follows. (We write M,s ⊭ A to
mean “not M,s ⊨ A.”)

1. A ≡ ⊥: M,s ⊭ A.

2. A ≡ R (t1 , . . . ,tn ): M,s ⊨ A iff ⟨ValsM (t1 ), . . . , ValsM (tn )⟩ ∈

RM.

3. A ≡ t1 = t2 : M,s ⊨ A iff ValsM (t1 ) = ValsM (t2 ).

4. A ≡ ¬B: M,s ⊨ A iff M,s ⊭ B.

5. A ≡ (B ∧ C ): M,s ⊨ A iff M,s ⊨ B and M,s ⊨ C .

6. A ≡ (B ∨ C ): M,s ⊨ A iff M,s ⊨ A or M,s ⊨ B (or both).

7. A ≡ (B → C ): M,s ⊨ A iff M,s ⊭ B or M,s ⊨ C (or both).

8. A ≡ ∀x B: M,s ⊨ A iff for every x-variant s ′ of s , M,s ′ ⊨ B.

9. A ≡ ∃x B: M,s ⊨ A iff there is an x-variant s ′ of s so that

M,s ′ ⊨ B.

The variable assignments are important in the last two

clauses. We cannot define satisfaction of ∀x B (x) by “for all
a ∈ |M|, M ⊨ B (a).” We cannot define satisfaction of ∃x B (x) by
“for at least one a ∈ |M|, M ⊨ B (a).” The reason is that a is not
symbol of the language, and so B (a) is not a formula (that is,
B [a/x] is undefined). We also cannot assume that we have con-
stant symbols or terms available that name every element of M,
since there is nothing in the definition of structures that requires
it. Even in the standard language the set of constant symbols
is countably infinite, so if |M| is not countable there aren’t even
enough constant symbols to name every object.

Example B.24. Let L = {a,b, f ,R} where a and b are constant

symbols, f is a two-place function symbol, and R is a two-place
216 APPENDIX B. FIRST-ORDER LOGIC

predicate symbol. Consider the structure M defined by:

1. |M| = {1, 2, 3, 4}
2. a M = 1
3. b M = 2
4. f M (x, y) = x + y if x + y ≤ 3 and = 3 otherwise.
5. R M = {⟨1, 1⟩, ⟨1, 2⟩, ⟨2, 3⟩, ⟨2, 4⟩}
The function s (x) = 1 that assigns 1 ∈ |M| to every variable is a
variable assignment for M.
Then
ValsM ( f (a,b)) = f M
(ValsM (a), ValsM (b)).

Since a and b are constant symbols, ValsM (a) = a M = 1 and

ValsM (b) = b M = 2. So

ValsM ( f (a,b)) = f M
(1, 2) = 1 + 2 = 3.

To compute the value of f ( f (a,b),a) we have to consider

ValsM ( f ( f (a,b),a)) = f M
(ValsM ( f (a,b)), ValsM (a)) = f M
(3, 1) = 3,

since 3 + 1 > 3. Since s (x) = 1 and ValsM (x) = s (x), we also have

ValsM ( f ( f (a,b),x)) = f M
(ValsM ( f (a,b)), ValsM (x)) = f M
(3, 1) = 3,
An atomic formula R (t1 ,t2 ) is satisfied if the tuple of values of
its arguments, i.e., ⟨ValsM (t1 ), ValsM (t2 )⟩, is an element of R M . So,
e.g., we have M,s ⊨ R (b, f (a,b)) since ⟨ValM (b), ValM ( f (a,b))⟩ =
⟨2, 3⟩ ∈ R M , but M,s ⊭ R (x, f (a,b)) since ⟨1, 3⟩ ∉ R M [s ].
To determine if a non-atomic formula A is satisfied, you apply
the clauses in the inductive definition that applies to the main con-
nective. For instance, the main connective in R (a,a) → (R (b,x) ∨
R (x,b) is the →, and
M,s ⊨ R (a,a) → (R (b,x) ∨ R (x,b)) iff
217 B.6. SATISFACTION OF A FORMULA IN A STRUCTURE

M,s ⊭ R (a,a) or M,s ⊨ R (b,x) ∨ R (x,b)

Since M,s ⊨ R (a,a) (because ⟨1, 1⟩ ∈ R M ) we can’t yet determine

the answer and must first figure out if M,s ⊨ R (b,x) ∨ R (x,b):

M,s ⊨ R (b,x) ∨ R (x,b) iff

M,s ⊨ R (b,x) or M,s ⊨ R (x,b)

And this is the case, since M,s ⊨ R (x,b) (because ⟨1, 2⟩ ∈ R M ).

Recall that an x-variant of s is a variable assignment that

differs from s at most in what it assigns to x. For every element
of |M|, there is an x-variant of s : s 1 (x) = 1, s 2 (x) = 2, s 3 (x) = 3,
s 4 (x) = 4, and with si (y) = s (y) = 1 for all variables y other
than x. These are all the x-variants of s for the structure M,
since |M| = {1, 2, 3, 4}. Note, in particular, that s 1 = s is also an
x-variant of s , i.e., s is always an x-variant of itself.
To determine if an existentially quantified formula ∃x A(x) is
satisfied, we have to determine if M,s ′ ⊨ A(x) for at least one
x-variant s ′ of s . So,

M,s ⊨ ∃x (R (b,x) ∨ R (x,b)),

since M,s 1 ⊨ R (b,x) ∨ R (x,b) (s 3 would also fit the bill). But,

M,s ⊭ ∃x (R (b,x) ∧ R (x,b))

since for none of the si , M,si ⊨ R (b,x) ∧ R (x,b).

To determine if a universally quantified formula ∀x A(x) is
satisfied, we have to determine if M,s ′ ⊨ A(x) for all x-variants s ′
of s . So,
M,s ⊨ ∀x (R (x,a) → R (a,x)),
since M,si ⊨ R (x,a) → R (a,x) for all si (M,s 1 ⊨ R (a,x) and
M,s j ⊭ R (x,a) for j = 2, 3, and 4). But,

M,s ⊭ ∀x (R (a,x) → R (x,a))

218 APPENDIX B. FIRST-ORDER LOGIC

since M,s 2 ⊭ R (a,x) → R (x,a) (because M,s 2 ⊨ R (a,x) and

M,s 2 ⊭ R (x,a)).
For a more complicated case, consider

∀x (R (a,x) → ∃y R (x, y)).

Since M,s 3 ⊭ R (a,x) and M,s 4 ⊭ R (a,x), the interesting cases

where we have to worry about the consequent of the conditional
are only s 1 and s 2 . Does M,s 1 ⊨ ∃y R (x, y) hold? It does if there
is at least one y-variant s 1′ of s 1 so that M,s 1′ ⊨ R (x, y). In fact,
s 1 is such a y-variant (s 1 (x) = 1, s1 (y) = 1, and ⟨1, 1⟩ ∈ R M ),
so the answer is yes. To determine if M,s 2 ⊨ ∃y R (x, y) we have
to look at the y-variants of s 2 . Here, s 2 itself does not satisfy
R (x, y) (s 2 (x) = 2, s2 (y) = 1, and ⟨2, 1⟩ ∉ R M ). However, consider
s 2′ ∼y s 2 with s 2′ (y) = 3. M,s 2′ ⊨ R (x, y) since ⟨2, 3⟩ ∈ R M , and
so M,s 2 ⊨ ∃y R (x, y). In sum, for every x-variant si of s , either
M,si ⊭ R (a,x) (i = 3, 4) or M,si ⊨ ∃y R (x, y) (i = 1, 2), and so

M,s ⊨ ∀x (R (a,x) → ∃y R (x, y)).

On the other hand,

M,s ⊭ ∃x (R (a,x) ∧ ∀y R (x, y)).

The only x-variants si of s with M,si ⊨ R (a,x) are s 1 and s2 . But

for each, there is in turn a y-variant si′ ∼y si with si′ (y) = 4 so that
M,si′ ⊭ R (x, y) and so M,si ⊭ ∀y R (x, y) for i = 1, 2. In sum, none
of the x-variants si ∼x s are such that M,si ⊨ R (a,x) ∧ ∀y R (x, y).

B.7 Variable Assignments

A variable assignment s provides a value for every variable—and
there are infinitely many of them. This is of course not neces-
sary. We require variable assignments to assign values to all vari-
ables simply because it makes things a lot easier. The value of a
term t , and whether or not a formula A is satisfied in a structure
with respect to s , only depend on the assignments s makes to
219 B.7. VARIABLE ASSIGNMENTS

the variables in t and the free variables of A. This is the content

of the next two propositions. To make the idea of “depends on”
precise, we show that any two variable assignments that agree on
all the variables in t give the same value, and that A is satisfied
relative to one iff it is satisfied relative to the other if two variable
assignments agree on all free variables of A.

Proposition B.25. If the variables in a term t are among x 1 , . . . , x n ,

and s 1 (x i ) = s2 (x i ) for i = 1, . . . , n, then ValsM1 (t ) = ValsM2 (t ).

Proof. By induction on the complexity of t . For the base case, t

can be a constant symbol or one of the variables x 1 , . . . , x n . If
t = c , then ValsM1 (t ) = c M = ValsM2 (t ). If t = x i , s 1 (x i ) = s 2 (x i )
by the hypothesis of the proposition, and so ValsM1 (t ) = s 1 (x i ) =
s 2 (x i ) = ValsM2 (t ).
For the inductive step, assume that t = f (t1 , . . . ,tk ) and that
the claim holds for t1 , . . . , tk . Then

ValsM1 (t ) = ValsM1 ( f (t1 , . . . ,tk )) =

=f M
(ValsM1 (t1 ), . . . , ValsM1 (tk ))

For j = 1, . . . , k , the variables of t j are among x 1 , . . . , x n . So by

induction hypothesis, ValsM1 (t j ) = ValsM2 (t j ). So,

ValsM1 (t ) = ValsM2 ( f (t1 , . . . ,tk )) =

=f M
(ValsM1 (t1 ), . . . , ValsM1 (tk )) =
=f M
(ValsM2 (t1 ), . . . , ValsM2 (tk )) =
= ValsM2 ( f (t1 , . . . ,tk )) = ValsM2 (t ). □

Proposition B.26. If the free variables in A are among x 1 , . . . , x n ,

and s 1 (x i ) = s 2 (x i ) for i = 1, . . . , n, then M,s 1 ⊨ A iff M,s 2 ⊨ A.

Proof. We use induction on the complexity of A. For the base

case, where A is atomic, A can be: ⊥, R (t1 , . . . ,tk ) for a k -place
predicate R and terms t1 , . . . , tk , or t1 = t2 for terms t1 and t2 .
220 APPENDIX B. FIRST-ORDER LOGIC

1. A ≡ ⊥: both M,s 1 ⊭ A and M,s 2 ⊭ A.

2. A ≡ R (t1 , . . . ,tk ): let M,s1 ⊨ A. Then

⟨ValsM1 (t1 ), . . . , ValsM1 (tk )⟩ ∈ R M .

For i = 1, . . . , k , ValsM1 (ti ) = ValsM2 (ti ) by Proposition B.25.

So we also have ⟨ValsM2 (ti ), . . . , ValsM2 (tk )⟩ ∈ R M .

3. A ≡ t1 = t2 : suppose M,s 1 ⊨ A. Then ValsM1 (t1 ) = ValsM1 (t2 ).

So,

ValsM2 (t1 ) = ValsM1 (t1 ) (by Proposition B.25)

= ValsM1 (t2 ) (since M,s 1 ⊨ t1 = t2 )
= ValsM2 (t2 ) (by Proposition B.25),

so M,s 2 ⊨ t1 = t2 .

Now assume M,s 1 ⊨ B iff M,s 2 ⊨ B for all formulas B less com-
plex than A. The induction step proceeds by cases determined by
the main operator of A. In each case, we only demonstrate the
forward direction of the biconditional; the proof of the reverse
direction is symmetrical. In all cases except those for the quanti-
fiers, we apply the induction hypothesis to sub-formulas B of A.
The free variables of B are among those of A. Thus, if s 1 and s 2
agree on the free variables of A, they also agree on those of B,
and the induction hypothesis applies to B.

1. A ≡ ¬B: if M,s1 ⊨ A, then M,s 1 ⊭ B, so by the induction

hypothesis, M,s 2 ⊭ B, hence M,s2 ⊨ A.

2. A ≡ B ∧ C : exercise.

3. A ≡ B ∨ C : if M,s 1 ⊨ A, then M,s1 ⊨ B or M,s 1 ⊨ C . By

induction hypothesis, M,s 2 ⊨ B or M,s 2 ⊨ C , so M,s 2 ⊨ A.

4. A ≡ B → C : exercise.
221 B.7. VARIABLE ASSIGNMENTS

5. A ≡ ∃x B: if M,s 1 ⊨ A, there is an x-variant s 1′ of s 1 so

that M,s 1′ ⊨ B. Let s 2′ be the x-variant of s 2 that assigns
the same thing to x as does s 1′ . The free variables of B are
among x 1 , . . . , x n , and x. s 1′ (x i ) = s 2′ (x i ), since s 1′ and s 2′
are x-variants of s 1 and s 2 , respectively, and by hypothesis
s 1 (x i ) = s2 (x i ). s1′ (x) = s 2′ (x) by the way we have defined s 2′ .
Then the induction hypothesis applies to B and s 1′ , s 2′ , so
M,s 2′ ⊨ B. Hence, there is an x-variant of s 2 that satisfies B,
and so M,s 2 ⊨ A.

6. A ≡ ∀x B: exercise.

By induction, we get that M,s1 ⊨ A iff M,s 2 ⊨ A whenever the free

variables in A are among x 1 , . . . , x n and s1 (x i ) = s 2 (x i ) for i = 1,
. . . , n. □

Sentences have no free variables, so any two variable assign-

ments assign the same things to all the (zero) free variables of any
sentence. The proposition just proved then means that whether
or not a sentence is satisfied in a structure relative to a variable
assignment is completely independent of the assignment. We’ll
record this fact. It justifies the definition of satisfaction of a sen-
tence in a structure (without mentioning a variable assignment)
that follows.

Corollary B.27. If A is a sentence and s a variable assignment, then

M,s ⊨ A iff M,s ′ ⊨ A for every variable assignment s ′.

Proof. Let s ′ be any variable assignment. Since A is a sentence, it

has no free variables, and so every variable assignment s ′ trivially
assigns the same things to all free variables of A as does s . So the
condition of Proposition B.26 is satisfied, and we have M,s ⊨ A
iff M,s ′ ⊨ A. □
222 APPENDIX B. FIRST-ORDER LOGIC

Definition B.28. If A is a sentence, we say that a structure M

satisfies A, M ⊨ A, iff M,s ⊨ A for all variable assignments s .

If M ⊨ A, we also simply say that A is true in M.

Proposition B.29. Let M be a structure, A be a sentence, and s a

variable assignment. M ⊨ A iff M,s ⊨ A.

Proof. Exercise. □

Proposition B.30. Suppose A(x) only contains x free, and M is

a structure. Then:

1. M ⊨ ∃x A(x) iff M,s ⊨ A(x) for at least one variable assign-

ment s .

2. M ⊨ ∀x A(x) iff M,s ⊨ A(x) for all variable assignments s .

Proof. Exercise. □

B.8 Extensionality
Extensionality, sometimes called relevance, can be expressed in-
formally as follows: the only factors that bears upon the satisfac-
tion of formula A in a structure M relative to a variable assign-
ment s , are the size of the domain and the assignments made
by M and s to the elements of the language that actually appear
in A.
One immediate consequence of extensionality is that where
two structures M and M ′ agree on all the elements of the lan-
guage appearing in a sentence A and have the same domain, M
and M ′ must also agree on whether or not A itself is true.

Proposition B.31 (Extensionality). Let A be a formula, and M1

and M2 be structures with |M1 | = |M2 |, and s a variable assignment
on |M1 | = |M2 |. If c M1 = c M2 , R M1 = R M2 , and f M1 = f M2 for every
223 B.8. EXTENSIONALITY

constant symbol c , relation symbol R, and function symbol f occurring

in A, then M1 ,s ⊨ A iff M2 ,s ⊨ A.

Proof. First prove (by induction on t ) that for every term,

ValsM1 (t ) = ValsM2 (t ). Then prove the proposition by induction
on A, making use of the claim just proved for the induction basis
(where A is atomic). □

Corollary B.32 (Extensionality for Sentences). Let A be a sen-

tence and M1 , M2 as in Proposition B.31. Then M1 ⊨ A iff M2 ⊨ A.

Proof. Follows from Proposition B.31 by Corollary B.27. □

Moreover, the value of a term, and whether or not a structure

satisfies a formula, only depends on the values of its subterms.

Proposition B.33. Let M be a structure, t and t ′ terms, and s a

variable assignment. Let s ′ ∼x s be the x-variant of s given by s ′ (x) =
ValsM (t ′). Then ValsM (t [t ′/x]) = ValsM′ (t ).

Proof. By induction on t .

1. If t is a constant, say, t ≡ c , then t [t ′/x] = c , and ValsM (c ) =

c M = ValsM′ (c ).

2. If t is a variable other than x, say, t ≡ y, then t [t ′/x] = y,

and ValsM (y) = ValsM′ (y) since s ′ ∼x s .

3. If t ≡ x, then t [t ′/x] = t ′. But ValsM′ (x) = ValsM (t ′) by

definition of s ′.

4. If t ≡ f (t1 , . . . ,tn ) then we have:

ValsM (t [t ′/x]) =
224 APPENDIX B. FIRST-ORDER LOGIC

= ValsM ( f (t1 [t ′/x], . . . ,tn [t ′/x]))

by definition of t [t ′/x]
=f M
(ValsM (t1 [t ′/x]), . . . , ValsM (tn [t ′/x]))
by definition of ValsM ( f (. . . ))
=f M
(ValsM′ (t1 ), . . . , ValsM′ (tn ))
by induction hypothesis
= ValsM′ (t ) by definition of ValsM′ ( f (. . . )) □

Proposition B.34. Let M be a structure, A a formula, t a term, and s

a variable assignment. Let s ′ ∼x s be the x-variant of s given by s ′ (x) =
ValsM (t ). Then M,s ⊨ A[t /x] iff M,s ′ ⊨ A.

Proof. Exercise. □

B.9 Semantic Notions

Give the definition of structures for first-order languages, we can
define some basic semantic properties of and relationships be-
tween sentences. The simplest of these is the notion of validity
of a sentence. A sentence is valid if it is satisfied in every struc-
ture. Valid sentences are those that are satisfied regardless of how
the non-logical symbols in it are interpreted. Valid sentences are
therefore also called logical truths—they are true, i.e., satisfied, in
any structure and hence their truth depends only on the logical
symbols occurring in them and their syntactic structure, but not
on the non-logical symbols or their interpretation.

Definition B.35 (Validity). A sentence A is valid, ⊨ A, iff M ⊨ A

for every structure M.
225 B.9. SEMANTIC NOTIONS

Definition B.36 (Entailment). A set of sentences 𝛤 entails a

sentence A, 𝛤 ⊨ A, iff for every structure M with M ⊨ 𝛤, M ⊨ A.

Definition B.37 (Satisfiability). A set of sentences 𝛤 is satisfi-

able if M ⊨ 𝛤 for some structure M. If 𝛤 is not satisfiable it is
called unsatisfiable.

Proposition B.38. A sentence A is valid iff 𝛤 ⊨ A for every set of

sentences 𝛤.

Proof. For the forward direction, let A be valid, and let 𝛤 be a

set of sentences. Let M be a structure so that M ⊨ 𝛤. Since A is
valid, M ⊨ A, hence 𝛤 ⊨ A.
For the contrapositive of the reverse direction, let A be in-
valid, so there is a structure M with M ⊭ A. When 𝛤 = {⊤}, since
⊤ is valid, M ⊨ 𝛤. Hence, there is a structure M so that M ⊨ 𝛤
but M ⊭ A, hence 𝛤 does not entail A. □

Proposition B.39. 𝛤 ⊨ A iff 𝛤 ∪ {¬A} is unsatisfiable.

Proof. For the forward direction, suppose 𝛤 ⊨ A and suppose to

the contrary that there is a structure M so that M ⊨ 𝛤 ∪ {¬A}.
Since M ⊨ 𝛤 and 𝛤 ⊨ A, M ⊨ A. Also, since M ⊨ 𝛤 ∪ {¬A}, M ⊨
¬A, so we have both M ⊨ A and M ⊭ A, a contradiction. Hence,
there can be no such structure M, so 𝛤 ∪ {A} is unsatisfiable.
For the reverse direction, suppose 𝛤 ∪ {¬A} is unsatisfiable.
So for every structure M, either M ⊭ 𝛤 or M ⊨ A. Hence, for
every structure M with M ⊨ 𝛤, M ⊨ A, so 𝛤 ⊨ A. □
226 APPENDIX B. FIRST-ORDER LOGIC

Proposition B.40. If 𝛤 ⊆ 𝛤 ′ and 𝛤 ⊨ A, then 𝛤 ′ ⊨ A.

Proof. Suppose that 𝛤 ⊆ 𝛤 ′ and 𝛤 ⊨ A. Let M be such that

M ⊨ 𝛤 ′; then M ⊨ 𝛤, and since 𝛤 ⊨ A, we get that M ⊨ A. Hence,
whenever M ⊨ 𝛤 ′, M ⊨ A, so 𝛤 ′ ⊨ A. □

Theorem B.41 (Semantic Deduction Theorem). 𝛤 ∪ {A} ⊨ B

iff 𝛤 ⊨ A → B.

Proof. For the forward direction, let 𝛤 ∪ {A} ⊨ B and let M be a

structure so that M ⊨ 𝛤. If M ⊨ A, then M ⊨ 𝛤 ∪ {A}, so since
𝛤 ∪ {A} entails B, we get M ⊨ B. Therefore, M ⊨ A → B, so
𝛤 ⊨ A → B.
For the reverse direction, let 𝛤 ⊨ A → B and M be a structure
so that M ⊨ 𝛤 ∪ {A}. Then M ⊨ 𝛤, so M ⊨ A → B, and since
M ⊨ A, M ⊨ B. Hence, whenever M ⊨ 𝛤 ∪ {A}, M ⊨ B, so
𝛤 ∪ {A} ⊨ B. □

Proposition B.42. Let M be a structure, and A(x) a formula with

one free variable x, and t a closed term. Then:

1. A(t ) ⊨ ∃x A(x)

2. ∀x A(x) ⊨ A(t )

Proof. 1. Suppose M ⊨ A(t ). Let s be a variable assignment

with s (x) = ValM (t ). Then M,s ⊨ A(t ) since A(t ) is
a sentence. By Proposition B.34, M,s ⊨ A(x). By Propo-
sition B.30, M ⊨ ∃x A(x).

2. Exercise. □
227 B.10. THEORIES

B.10 Theories
Definition B.43. A set of sentences 𝛤 is closed iff, whenever 𝛤 ⊨
A then A ∈ 𝛤. The closure of a set of sentences 𝛤 is {A : 𝛤 ⊨ A}.
We say that 𝛤 is axiomatized by a set of sentences 𝛥 if 𝛤 is the
closure of 𝛥.

Example B.44. The theory of strict linear orders in the lan-

guage L< is axiomatized by the set

∀x ¬x < x,
∀x ∀y ((x < y ∨ y < x) ∨ x = y),
∀x ∀y ∀z ((x < y ∧ y < z ) → x < z )

It completely captures the intended structures: every strict linear

order is a model of this axiom system, and vice versa, if R is a
linear order on a set X , then the structure M with |M| = X and
<M = R is a model of this theory.

Example B.45. The theory of groups in the language 1 (con-

stant symbol), · (two-place function symbol) is axiomatized by

∀x (x · 1) = x
∀x ∀y ∀z (x · (y · z )) = ((x · y) · z )
∀x ∃y (x · y) = 1

Example B.46. The theory of Peano arithmetic is axiomatized

by the following sentences in the language of arithmetic LA .

¬∃x x ′ = 0
∀x ∀y (x ′ = y ′ → x = y)
∀x ∀y (x < y ↔ ∃z (z ′ + x) = y)
∀x (x + 0) = x
∀x ∀y (x + y ′) = (x + y) ′
∀x (x × 0) = 0
228 APPENDIX B. FIRST-ORDER LOGIC

∀x ∀y (x × y ′) = ((x × y) + x)

plus all sentences of the form

(A(0) ∧ ∀x (A(x) → A(x ′))) → ∀x A(x)

Since there are infinitely many sentences of the latter form, this
axiom system is infinite. The latter form is called the induction
schema. (Actually, the induction schema is a bit more complicated
than we let on here.)
The third axiom is an explicit definition of <.

Summary
A first-order language consists of constant, function, and
predicate symbols. Function and constant symbols take a speci-
fied number of arguments. In the language of arithmetic, e.g.,
we have a single constant symbol 0, one 1-place function symbol ′,
two 2-place function symbols + and ×, and one 2-place predicate
symbol <. From variables and constant and function symbols
we form the terms of a language. From the terms of a language
together with its predicate symbol, as well as the identity sym-
bol =, we form the atomic formulas. And in turn from them,
using the logical connectives ¬, ∨, ∧, →, ↔ and the quantifiers ∀
and ∃ we form its formulas. Since we are careful to always include
necessary parentheses in the process of forming terms and formu-
las, there is always exactly one way of reading a formula. This
makes it possible to define things by induction on the structure
of formulas.
Occurrences of variables in formulas are sometimes governed
by a corresponding quantifier: if a variable occurs in the scope
of a quantifier it is considered bound, otherwise free. These
concepts all have inductive definitions, and we also inductively
define the operation of substitution of a term for a variable in
a formula. Formulas without free variable occurrences are called
sentences.
229 B.10. THEORIES

The semantics for a first-order language is given by a struc-

ture for that language. It consists of a domain and elements
of that domain are assigned to each constant symbol. Function
symbols are interpreted by functions and relation symbols by re-
lation on the domain. A function from the set of variables to the
domain is a variable assignment. The relation of satisfaction
relates structures, variable assignments and formulas; M,s ⊨ A is
defined by induction on the structure of A. M,s ⊨ A only depends
on the interpretation of the symbols actually occurring in A, and
in particular does not depend on s if A contains no free variables.
So if A is a sentence, M ⊨ A if M,s ⊨ A for any (or all) s .
The satisfaction relation is the basis for all semantic notions.
A sentence is valid, ⊨ A, if it is satisfied in every structure. A
sentence A is entailed by set of sentences 𝛤, 𝛤 ⊨ A, iff M ⊨ A for
all M which satisfy every sentence in 𝛤. A set 𝛤 is satisfiable iff
there is some structure that satisfies every sentence in 𝛤, other-
wise unsatisfiable. These notions are interrelated, e.g., 𝛤 ⊨ A iff
𝛤 ∪ {¬A} is unsatisfiable.

Problems
Problem B.1. Give an inductive definition of the bound variable
occurrences along the lines of Definition B.8.

Problem B.2. Let L = {c , f ,A} with one constant symbol, one

one-place function symbol and one two-place predicate symbol,
and let the structure M be given by

1. |M| = {1, 2, 3}

2. c M = 3

3. f M (1) = 2, f M (2) = 3, f M (3) =2

4. AM = {⟨1, 2⟩, ⟨2, 3⟩, ⟨3, 3⟩}

230 APPENDIX B. FIRST-ORDER LOGIC

(a) Let s (v ) = 1 for all variables v . Find out whether

M,s ⊨ ∃x (A( f (z ),c ) → ∀y (A(y,x) ∨ A( f (y),x)))

Explain why or why not.

(b) Give a different structure and variable assignment in
which the formula is not satisfied.

Problem B.3. Complete the proof of Proposition B.26.

Problem B.4. Prove Proposition B.29

Problem B.5. Prove Proposition B.30.

Problem B.6. Suppose L is a language without function sym-

bols. Given a structure M, c a constant symbol and a ∈ |M|,
define M [a/c ] to be the structure that is just like M, except that
c M [a/c ] = a. Define M ||= A for sentences A by:

1. A ≡ ⊥: not M ||= A.

2. A ≡ R (d1 , . . . ,dn ): M ||= A iff ⟨d1M , . . . ,dnM ⟩ ∈ R M .

3. A ≡ d1 = d2 : M ||= A iff d1M = d2M .

4. A ≡ ¬B: M ||= A iff not M ||= B.

5. A ≡ (B ∧ C ): M ||= A iff M ||= B and M ||= C .

6. A ≡ (B ∨ C ): M ||= A iff M ||= B or M ||= C (or both).

7. A ≡ (B →C ): M ||= A iff not M ||= B or M ||= C (or both).

8. A ≡ ∀x B: M ||= A iff for all a ∈ |M|, M [a/c ] ||= B [c /x], if

c does not occur in B.

9. A ≡ ∃x B: M ||= A iff there is an a ∈ |M| such that

M [a/c ] ||= B [c /x], if c does not occur in B.
231 B.10. THEORIES

Let x 1 , . . . , x n be all free variables in A, c 1 , . . . , c n constant sym-

bols not in A, a1 , . . . , an ∈ |M|, and s (x i ) = ai .
Show that M,s ⊨ A iff M [a 1 /c 1 , . . . ,an /c n ] ||=
A[c 1 /x 1 ] . . . [c n /x n ].
(This problem shows that it is possible to give a semantics for
first-order logic that makes do without variable assignments.)

Problem B.7. Suppose that f is a function symbol not in A(x, y).

Show that there is a structure M such that M ⊨ ∀x ∃y A(x, y) iff
there is an M ′ such that M ′ ⊨ ∀x A(x, f (x)).
(This problem is a special case of what’s known as Skolem’s
Theorem; ∀x A(x, f (x)) is called a Skolem normal form of
∀x ∃y A(x, y).)

Problem B.8. Carry out the proof of Proposition B.31 in detail.

Problem B.9. Prove Proposition B.34

Problem B.10. 1. Show that 𝛤 ⊨ ⊥ iff 𝛤 is unsatisfiable.

2. Show that 𝛤 ∪ {A} ⊨ ⊥ iff 𝛤 ⊨ ¬A.

3. Suppose c does not occur in A or 𝛤. Show that 𝛤 ⊨ ∀x A iff

𝛤 ⊨ A[c /x].

Problem B.11. Complete the proof of Proposition B.42.

APPENDIX C

Natural
Deduction
C.1 Natural Deduction
Natural deduction is a derivation system intended to mirror ac-
tual reasoning (especially the kind of regimented reasoning em-
ployed by mathematicians). Actual reasoning proceeds by a num-
ber of “natural” patterns. For instance, proof by cases allows us
to establish a conclusion on the basis of a disjunctive premise,
by establishing that the conclusion follows from either of the dis-
juncts. Indirect proof allows us to establish a conclusion by show-
ing that its negation leads to a contradiction. Conditional proof
establishes a conditional claim “if . . . then . . . ” by showing that
the consequent follows from the antecedent. Natural deduction
is a formalization of some of these natural inferences. Each of
the logical connectives and quantifiers comes with two rules, an
introduction and an elimination rule, and they each correspond
to one such natural inference pattern. For instance, →Intro cor-
responds to conditional proof, and ∨Elim to proof by cases. A
particularly simple rule is ∧Elim which allows the inference from
A ∧ B to A (or B).
One feature that distinguishes natural deduction from other
derivation systems is its use of assumptions. A derivation in nat-

232
233 C.1. NATURAL DEDUCTION

ural deduction is a tree of formulas. A single formula stands

at the root of the tree of formulas, and the “leaves” of the tree
are formulas from which the conclusion is derived. In natural
deduction, some leaf formulas play a role inside the derivation
but are “used up” by the time the derivation reaches the conclu-
sion. This corresponds to the practice, in actual reasoning, of
introducing hypotheses which only remain in effect for a short
while. For instance, in a proof by cases, we assume the truth of
each of the disjuncts; in conditional proof, we assume the truth
of the antecedent; in indirect proof, we assume the truth of the
negation of the conclusion. This way of introducing hypotheti-
cal assumptions and then doing away with them in the service of
establishing an intermediate step is a hallmark of natural deduc-
tion. The formulas at the leaves of a natural deduction derivation
are called assumptions, and some of the rules of inference may
“discharge” them. For instance, if we have a derivation of B from
some assumptions which include A, then the →Intro rule allows
us to infer A → B and discharge any assumption of the form A.
(To keep track of which assumptions are discharged at which in-
ferences, we label the inference and the assumptions it discharges
with a number.) The assumptions that remain undischarged at
the end of the derivation are together sufficient for the truth of the
conclusion, and so a derivation establishes that its undischarged
assumptions entail its conclusion.
The relation 𝛤 ⊢ A based on natural deduction holds iff there
is a derivation in which A is the last sentence in the tree, and every
leaf which is undischarged is in 𝛤. A is a theorem in natural de-
duction iff there is a derivation in which A is the last sentence and
all assumptions are discharged. For instance, here is a derivation
that shows that ⊢ (A ∧ B) → A:
[A ∧ B] 1
∧Elim
A
1 →Intro
(A ∧ B) → A
The label 1 indicates that the assumption A ∧ B is discharged at
the →Intro inference.
234 APPENDIX C. NATURAL DEDUCTION

A set 𝛤 is inconsistent iff 𝛤 ⊢ ⊥ in natural deduction. The

rule ⊥I makes it so that from an inconsistent set, any sentence
can be derived.
Natural deduction systems were developed by Gerhard
Gentzen and Stanisław Jaśkowski in the 1930s, and later devel-
oped by Dag Prawitz and Frederic Fitch. Because its inferences
mirror natural methods of proof, it is favored by philosophers.
The versions developed by Fitch are often used in introductory
logic textbooks. In the philosophy of logic, the rules of natural
deduction have sometimes been taken to give the meanings of
the logical operators (“proof-theoretic semantics”).

C.2 Rules and Derivations

Natural deduction systems are meant to closely parallel the infor-
mal reasoning used in mathematical proof (hence it is somewhat
“natural”). Natural deduction proofs begin with assumptions. In-
ference rules are then applied. Assumptions are “discharged” by
the ¬Intro, →Intro, ∨Elim and ∃Elim inference rules, and the
label of the discharged assumption is placed beside the inference
for clarity.

Definition C.1 (Assumption). An assumption is any sentence

in the topmost position of any branch.

Derivations in natural deduction are certain trees of sen-

tences, where the topmost sentences are assumptions, and if
a sentence stands below one, two, or three other sequents, it
must follow correctly by a rule of inference. The sentences at
the top of the inference are called the premises and the sentence
below the conclusion of the inference. The rules come in pairs, an
introduction and an elimination rule for each logical operator.
They introduce a logical operator in the conclusion or remove
a logical operator from a premise of the rule. Some of the rules
allow an assumption of a certain type to be discharged. To indi-
cate which assumption is discharged by which inference, we also
235 C.3. PROPOSITIONAL RULES

assign labels to both the assumption and the inference. This is

indicated by writing the assumption as “[A] n .”
It is customary to consider rules for all the logical operators
∧, ∨, →, ¬, and ⊥, even if some of those are consider as defined.

C.3 Propositional Rules

Rules for ∧
A∧B
∧Elim
A B A
∧Intro
A∧B A∧B
∧Elim
B

Rules for ∨
[A] n [B] n
A
∨Intro
A∨B
B
∨Intro A∨B C C
A∨B n ∨Elim
C

Rules for →
[A] n
A→B A
→Elim
B
B
n →Intro
A→B

Rules for ¬
236 APPENDIX C. NATURAL DEDUCTION

[A] n
¬A A
⊥ ¬Elim
⊥
n ¬Intro
¬A

Rules for ⊥
[¬A] n
⊥ ⊥
I
A
n ⊥ ⊥
C
A

Note that ¬Intro and ⊥C are very similar: The difference is

that ¬Intro derives a negated sentence ¬A but ⊥C a positive sen-
tence A.
Whenever a rule indicates that some assumption may be dis-
charged, we take this to be a permission, but not a requirement.
E.g., in the →Intro rule, we may discharge any number of assump-
tions of the form A in the derivation of the premise B, including
zero.

C.4 Quantifier Rules

Rules for ∀
A(a) ∀x A(x)
∀Intro ∀Elim
∀x A(x) A(t )

In the rules for ∀, t is a ground term (a term that does not

contain any variables), and a is a constant symbol which does
not occur in the conclusion ∀x A(x), or in any assumption which
is undischarged in the derivation ending with the premise A(a).
We call a the eigenvariable of the ∀Intro inference.
237 C.4. QUANTIFIER RULES

Rules for ∃
[A(a)]n
A(t )
∃Intro
∃x A(x)
∃x A(x) C
n ∃Elim
C

Again, t is a ground term, and a is a constant which does

not occur in the premise ∃x A(x), in the conclusion C , or any
assumption which is undischarged in the derivations ending with
the two premises (other than the assumptions A(a)). We call a
the eigenvariable of the ∃Elim inference.
The condition that an eigenvariable neither occur in the
premises nor in any assumption that is undischarged in the
derivations leading to the premises for the ∀Intro or ∃Elim in-
ference is called the eigenvariable condition.
We use the term “eigenvariable” even though a in the above
rules is a constant. This has historical reasons.
In ∃Intro and ∀Elim there are no restrictions, and the term t
can be anything, so we do not have to worry about any conditions.
On the other hand, in the ∃Elim and ∀Intro rules, the eigenvari-
able condition requires that the constant symbol a does not occur
anywhere in the conclusion or in an undischarged assumption.
The condition is necessary to ensure that the system is sound,
i.e., only derives sentences from undischarged assumptions from
which they follow. Without this condition, the following would
be allowed:
[A(a)] 1
*∀Intro
∃x A(x) ∀x A(x)
∃Elim
∀x A(x)

However, ∃x A(x) ⊭ ∀x A(x).

238 APPENDIX C. NATURAL DEDUCTION

C.5 Derivations
We’ve said what an assumption is, and we’ve given the rules of
inference. Derivations in natural deduction are inductively gen-
erated from these: each derivation either is an assumption on its
own, or consists of one, two, or three derivations followed by a
correct inference.

Definition C.2 (Derivation). A derivation of a sentence A from

assumptions 𝛤 is a tree of sentences satisfying the following con-
ditions:

1. The topmost sentences of the tree are either in 𝛤 or are

discharged by an inference in the tree.

2. The bottommost sentence of the tree is A.

3. Every sentence in the tree except the sentence A at the bot-

tom is a premise of a correct application of an inference
rule whose conclusion stands directly below that sentence
in the tree.

We then say that A is the conclusion of the derivation and that A

is derivable from 𝛤.

Example C.3. Every assumption on its own is a derivation. So,

e.g., C by itself is a derivation, and so is D by itself. We can
obtain a new derivation from these by applying, say, the ∧Intro
rule,
A B
∧Intro
A∧B

These rules are meant to be general: we can replace the A and B

in it with any sentences, e.g., by C and D. Then the conclusion
would be C ∧ D, and so
C D
∧Intro
C ∧D
239 C.6. EXAMPLES OF DERIVATIONS

is a correct derivation. Of course, we can also switch the assump-

tions, so that D plays the role of A and C that of B. Thus,
D C
∧Intro
D ∧C

is also a correct derivation.

We can now apply another rule, say, →Intro, which allows
us to conclude a conditional and allows us to discharge any as-
sumption that is identical to the antecedent of that conditional.
So both of the following would be correct derivations:
[C ] 1 D C [D] 1
∧Intro ∧Intro
C ∧D C ∧D
1 →Intro 1 →Intro
C → (C ∧ D) D → (C ∧ D)

Remember that discharging of assumptions is a permission,

not a requirement: we don’t have to discharge the assumptions.
In particular, we can apply a rule even if the assumptions are
not present in the derivation. For instance, the following is legal,
even though there is no assumption A to be discharged:
B
1 →Intro
A→B

C.6 Examples of Derivations

Example C.4. Let’s give a derivation of the sentence (A∧B)→A.
We begin by writing the desired conclusion at the bottom of
the derivation.

(A ∧ B) → A

Next, we need to figure out what kind of inference could result

in a sentence of this form. The main operator of the conclusion
is →, so we’ll try to arrive at the conclusion using the →Intro
rule. It is best to write down the assumptions involved and label
the inference rules as you progress, so it is easy to see whether
all assumptions have been discharged at the end of the proof.
240 APPENDIX C. NATURAL DEDUCTION

[A ∧ B] 1

A
1 →Intro
(A ∧ B) → A

We now need to fill in the steps from the assumption A ∧ B

to A. Since we only have one connective to deal with, ∧, we must
use the ∧ elim rule. This gives us the following proof:
[A ∧ B] 1
∧Elim
A
1 →Intro
(A ∧ B) → A

We now have a correct derivation of (A ∧ B) → A.

Example C.5. Now let’s give a derivation of (¬A∨B) → (A→B).

We begin by writing the desired conclusion at the bottom of
the derivation.

(¬A ∨ B) → (A → B)

To find a logical rule that could give us this conclusion, we look at

the logical connectives in the conclusion: ¬, ∨, and →. We only
care at the moment about the first occurence of → because it is
the main operator of the sentence in the end-sequent, while ¬, ∨
and the second occurence of → are inside the scope of another
connective, so we will take care of those later. We therefore start
with the →Intro rule. A correct application must look like this:
[¬A ∨ B] 1

A→B
1 →Intro
(¬A ∨ B) → (A → B)

This leaves us with two possibilities to continue. Either we can

keep working from the bottom up and look for another applica-
tion of the →Intro rule, or we can work from the top down and
241 C.6. EXAMPLES OF DERIVATIONS

apply a ∨Elim rule. Let us apply the latter. We will use the as-
sumption ¬A ∨ B as the leftmost premise of ∨Elim. For a valid
application of ∨Elim, the other two premises must be identical
to the conclusion A → B, but each may be derived in turn from
another assumption, namely the two disjuncts of ¬A ∨ B. So our
derivation will look like this:
[¬A] 2 [B] 2

[¬A ∨ B] 1 A→B A→B

2 ∨Elim
A→B
1 →Intro
(¬A ∨ B) → (A → B)

In each of the two branches on the right, we want to derive

A → B, which is best done using →Intro.

[¬A] 2 , [A] 3 [B] 2 , [A] 4

B B
3 →Intro 4 →Intro
[¬A ∨ B] 1 A→B A→B
2 ∨Elim
A→B
1 →Intro
(¬A ∨ B) → (A → B)

For the two missing parts of the derivation, we need deriva-

tions of B from ¬A and A in the middle, and from A and B on the
left. Let’s take the former first. ¬A and A are the two premises of
¬Elim:
[¬A] 2 [A] 3
⊥ ¬Elim

By using ⊥I , we can obtain B as a conclusion and complete the

branch.
242 APPENDIX C. NATURAL DEDUCTION

[B] 2 , [A] 4
[¬A] 2 [A] 3
⊥ ⊥ ⊥Intro
I
B B
3 →Intro 4 →Intro
[¬A ∨ B] 1 A→B A→B
2 ∨Elim
A→B
1 →Intro
(¬A ∨ B) → (A → B)

Let’s now look at the rightmost branch. Here it’s important

to realize that the definition of derivation allows assumptions to be
discharged but does not require them to be. In other words, if we
can derive B from one of the assumptions A and B without using
the other, that’s ok. And to derive B from B is trivial: B by itself
is such a derivation, and no inferences are needed. So we can
simply delete the assumption A.

[¬A] 2 [A] 3
⊥ ⊥ ¬Elim
B
I [B] 2
3 →Intro →Intro
[¬A ∨ B] 1 A→B A→B
2 ∨Elim
A→B
1 →Intro
(¬A ∨ B) → (A → B)

Note that in the finished derivation, the rightmost →Intro infer-

ence does not actually discharge any assumptions.

Example C.6. So far we have not needed the ⊥C rule. It is spe-

cial in that it allows us to discharge an assumption that isn’t a
sub-formula of the conclusion of the rule. It is closely related to
the ⊥I rule. In fact, the ⊥I rule is a special case of the ⊥C rule—
there is a logic called “intuitionistic logic” in which only ⊥I is
allowed. The ⊥C rule is a last resort when nothing else works.
For instance, suppose we want to derive A ∨ ¬A. Our usual strat-
egy would be to attempt to derive A ∨ ¬A using ∨Intro. But this
would require us to derive either A or ¬A from no assumptions,
and this can’t be done. ⊥C to the rescue!
243 C.6. EXAMPLES OF DERIVATIONS

[¬(A ∨ ¬A)] 1

1
⊥ ⊥C
A ∨ ¬A

Now we’re looking for a derivation of ⊥ from ¬(A ∨ ¬A). Since

⊥ is the conclusion of ¬Elim we might try that:

[¬(A ∨ ¬A)] 1 [¬(A ∨ ¬A)] 1

¬A A
⊥ ¬Elim
1 ⊥C
A ∨ ¬A

Our strategy for finding a derivation of ¬A calls for an application

of ¬Intro:
[¬(A ∨ ¬A)] 1 , [A] 2
[¬(A ∨ ¬A)] 1

⊥
2 ¬Intro
¬A A
⊥ ¬Elim
1 ⊥C
A ∨ ¬A

Here, we can get ⊥ easily by applying ¬Elim to the assumption

¬(A ∨ ¬A) and A ∨ ¬A which follows from our new assumption
A by ∨Intro:

[A] 2 [¬(A ∨ ¬A)] 1

∨Intro
[¬(A ∨ ¬A)] 1 A ∨ ¬A
⊥ ¬Elim
2 ¬Intro
¬A A
⊥ ¬Elim
1 ⊥C
A ∨ ¬A

On the right side we use the same strategy, except we get A by ⊥C :

244 APPENDIX C. NATURAL DEDUCTION

[A] 2 [¬A] 3
∨Intro ∨Intro
[¬(A ∨ ¬A)] 1 A ∨ ¬A [¬(A ∨ ¬A)] 1 A ∨ ¬A
⊥ ¬Elim ⊥ ⊥ ¬Elim
2 ¬Intro 3 C
¬A A
⊥ ¬Elim
1 ⊥C
A ∨ ¬A

C.7 Derivations with Quantifiers

Example C.7. When dealing with quantifiers, we have to make
sure not to violate the eigenvariable condition, and sometimes
this requires us to play around with the order of carrying out
certain inferences. In general, it helps to try and take care of rules
subject to the eigenvariable condition first (they will be lower
down in the finished proof).
Let’s see how we’d give a derivation of the formula ∃x ¬A(x)→
¬∀x A(x). Starting as usual, we write

∃x ¬A(x) → ¬∀x A(x)

We start by writing down what it would take to justify that last

step using the →Intro rule.

[∃x ¬A(x)] 1

¬∀x A(x)
1 →Intro
∃x ¬A(x) → ¬∀x A(x)

Since there is no obvious rule to apply to ¬∀x A(x), we will pro-

ceed by setting up the derivation so we can use the ∃Elim rule.
Here we must pay attention to the eigenvariable condition, and
choose a constant that does not appear in ∃x A(x) or any assump-
tions that it depends on. (Since no constant symbols appear,
however, any choice will do fine.)
245 C.7. DERIVATIONS WITH QUANTIFIERS

[¬A(a)] 2

[∃x ¬A(x)] 1 ¬∀x A(x)

2 ∃Elim
¬∀x A(x)
1 →Intro
∃x ¬A(x) → ¬∀x A(x)
In order to derive ¬∀x A(x), we will attempt to use the ¬Intro
rule: this requires that we derive a contradiction, possibly using
∀x A(x) as an additional assumption. Of course, this contradic-
tion may involve the assumption ¬A(a) which will be discharged
by the →Intro inference. We can set it up as follows:
[¬A(a)] 2 , [∀x A(x)] 3

⊥
3 ¬Intro
[∃x ¬A(x)] 1 ¬∀x A(x)
2 ∃Elim
¬∀x A(x)
1 →Intro
∃x ¬A(x) → ¬∀x A(x)
It looks like we are close to getting a contradiction. The easiest
rule to apply is the ∀Elim, which has no eigenvariable conditions.
Since we can use any term we want to replace the universally
quantified x, it makes the most sense to continue using a so we
can reach a contradiction.
[∀x A(x)] 3
∀Elim
[¬A(a)] 2 A(a)
⊥ ¬Elim
1 3 ¬Intro
[∃x ¬A(x)] ¬∀x A(x)
2 ∃Elim
¬∀x A(x)
1 →Intro
∃x ¬A(x) → ¬∀x A(x)
It is important, especially when dealing with quantifiers, to
double check at this point that the eigenvariable condition has
not been violated. Since the only rule we applied that is subject
to the eigenvariable condition was ∃Elim, and the eigenvariable a
246 APPENDIX C. NATURAL DEDUCTION

does not occur in any assumptions it depends on, this is a correct

derivation.

Example C.8. Sometimes we may derive a formula from other

formulas. In these cases, we may have undischarged assumptions.
It is important to keep track of our assumptions as well as the end
goal.
Let’s see how we’d give a derivation of the formula ∃x C (x,b)
from the assumptions ∃x (A(x) ∧ B (x)) and ∀x (B (x) → C (x,b)).
Starting as usual, we write the conclusion at the bottom.

∃x C (x,b)

We have two premises to work with. To use the first, i.e., try
to find a derivation of ∃x C (x,b) from ∃x (A(x) ∧ B (x)) we would
use the ∃Elim rule. Since it has an eigenvariable condition, we
will apply that rule first. We get the following:

[A(a) ∧ B (a)] 1

∃x (A(x) ∧ B (x)) ∃x C (x,b)

1 ∃Elim
∃x C (x,b)

The two assumptions we are working with share B. It may be

useful at this point to apply ∧Elim to separate out B (a).

[A(a) ∧ B (a)] 1
∧Elim
B (a)

∃x (A(x) ∧ B (x)) ∃x C (x,b)

1 ∃Elim
∃x C (x,b)

The second assumption we have to work with is ∀x (B (x) →

C (x,b)). Since there is no eigenvariable condition we can instan-
tiate x with the constant symbol a using ∀Elim to get B (a) →
247 C.7. DERIVATIONS WITH QUANTIFIERS

C (a,b). We now have both B (a) → C (a,b) and B (a). Our next
move should be a straightforward application of the →Elim rule.
∀x (B (x) → C (x,b)) [A(a) ∧ B (a)] 1
∀Elim ∧Elim
B (a) → C (a,b) B (a)
→Elim
C (a,b)

∃x (A(x) ∧ B (x)) ∃x C (x,b)

1 ∃Elim
∃x C (x,b)
We are so close! One application of ∃Intro and we have reached
our goal.
∀x (B (x) → C (x,b)) [A(a) ∧ B (a)] 1
∀Elim ∧Elim
B (a) → C (a,b) B (a)
→Elim
C (a,b)
∃Intro
∃x (A(x) ∧ B (x)) ∃x C (x,b)
1 ∃Elim
∃x C (x,b)
Since we ensured at each step that the eigenvariable conditions
were not violated, we can be confident that this is a correct deriva-
tion.

Example C.9. Give a derivation of the formula ¬∀x A(x) from

the assumptions ∀x A(x) → ∃y B (y) and ¬∃y B (y). Starting as
usual, we write the target formula at the bottom.

¬∀x A(x)
The last line of the derivation is a negation, so let’s try using
¬Intro. This will require that we figure out how to derive a con-
tradiction.
[∀x A(x)] 1

⊥
1 ¬Intro
¬∀x A(x)
248 APPENDIX C. NATURAL DEDUCTION

So far so good. We can use ∀Elim but it’s not obvious if that will
help us get to our goal. Instead, let’s use one of our assumptions.
∀x A(x) → ∃y B (y) together with ∀x A(x) will allow us to use the
→Elim rule.
∀x A(x) → ∃y B (y) [∀x A(x)] 1
→Elim
∃y B (y)

⊥
1 ¬Intro
¬∀x A(x)

We now have one final assumption to work with, and it looks like
this will help us reach a contradiction by using ¬Elim.

∀x A(x) → ∃y B (y) [∀x A(x)] 1

→Elim
¬∃y B (y) ∃y B (y)
⊥ ¬Elim
1 ¬Intro
¬∀x A(x)

C.8 Derivations with Identity predicate

Derivations with identity predicate require additional inference
rules.

t1 = t2 A(t1 )
=Elim
A(t2 )
t = t =Intro
t1 = t2 A(t2 )
=Elim
A(t1 )

In the above rules, t , t1 , and t2 are closed terms. The =Intro

rule allows us to derive any identity statement of the form t = t
outright, from no assumptions.

Example C.10. If s and t are closed terms, then A(s ),s = t ⊢

A(t ):
249 C.8. DERIVATIONS WITH IDENTITY PREDICATE

s =t A(s )
=Elim
A(t )

This may be familiar as the “principle of substitutability of iden-

ticals,” or Leibniz’ Law.

Example C.11. We derive the sentence

∀x ∀y ((A(x) ∧ A(y)) → x = y)

from the sentence

∃x ∀y (A(y) → y = x)

We develop the derivation backwards:

∃x ∀y (A(y) → y = x) [A(a) ∧ A(b)] 1

a =b
1 →Intro
((A(a) ∧ A(b)) → a = b)
∀Intro
∀y ((A(a) ∧ A(y)) → a = y)
∀Intro
∀x ∀y ((A(x) ∧ A(y)) → x = y)

We’ll now have to use the main assumption: since it is an existen-

tial formula, we use ∃Elim to derive the intermediary conclusion
a = b.
[∀y (A(y) → y = c )] 2
[A(a) ∧ A(b)] 1

∃x ∀y (A(y) → y = x) a =b
2 ∃Elim
a = b
1 →Intro
((A(a) ∧ A(b)) → a = b)
∀Intro
∀y ((A(a) ∧ A(y)) → a = y)
∀Intro
∀x ∀y ((A(x) ∧ A(y)) → x = y)
250 APPENDIX C. NATURAL DEDUCTION

The sub-derivation on the top right is completed by using its

assumptions to show that a = c and b = c . This requires two
separate derivations. The derivation for a = c is as follows:

[∀y (A(y) → y = c )] 2 [A(a) ∧ A(b)] 1

∀Elim ∧Elim
A(a) → a = c A(a)
a =c →Elim

From a = c and b = c we derive a = b by =Elim.

C.9 Proof-Theoretic Notions

Just as we’ve defined a number of important semantic notions
(validity, entailment, satisfiabilty), we now define corresponding
proof-theoretic notions. These are not defined by appeal to satisfac-
tion of sentences in structures, but by appeal to the derivability
or non-derivability of certain sentences from others. It was an
important discovery that these notions coincide. That they do is
the content of the soundness and completeness theorems.

Definition C.12 (Theorems). A sentence A is a theorem if there

is a derivation of A in natural deduction in which all assumptions
are discharged. We write ⊢ A if A is a theorem and ⊬ A if it is not.

Definition C.13 (Derivability). A sentence A is derivable from

a set of sentences 𝛤, 𝛤 ⊢ A, if there is a derivation with conclu-
sion A and in which every assumption is either discharged or is
in 𝛤. If A is not derivable from 𝛤 we write 𝛤 ⊬ A.

Definition C.14 (Consistency). A set of sentences 𝛤 is incon-

sistent iff 𝛤 ⊢ ⊥. If 𝛤 is not inconsistent, i.e., if 𝛤 ⊬ ⊥, we say it is
consistent.
251 C.9. PROOF-THEORETIC NOTIONS

Proposition C.15 (Reflexivity). If A ∈ 𝛤, then 𝛤 ⊢ A.

Proof. The assumption A by itself is a derivation of A where every

undischarged assumption (i.e., A) is in 𝛤. □

Proposition C.16 (Monotony). If 𝛤 ⊆ 𝛥 and 𝛤 ⊢ A, then 𝛥 ⊢ A.

Proof. Any derivation of A from 𝛤 is also a derivation of A

from 𝛥. □

Proposition C.17 (Transitivity). If 𝛤 ⊢ A and {A} ∪ 𝛥 ⊢ B, then

𝛤 ∪ 𝛥 ⊢ B.

Proof. If 𝛤 ⊢ A, there is a derivation 𝛿0 of A with all undischarged

assumptions in 𝛤. If {A} ∪ 𝛥 ⊢ B, then there is a derivation 𝛿1
of B with all undischarged assumptions in {A}∪ 𝛥. Now consider:
𝛥, [A] 1
𝛤
𝛿1
𝛿0
B
1 →Intro
A→B A
→Elim
B

The undischarged assumptions are now all among 𝛤 ∪ 𝛥, so this

shows 𝛤 ∪ 𝛥 ⊢ B. □

When 𝛤 = {A1 ,A2 , . . . ,Ak } is a finite set we may use the sim-
plified notation A1 ,A2 , . . . ,Ak ⊢ B for 𝛤 ⊢ B, in particular A ⊢ B
means that {A} ⊢ B.
Note that if 𝛤 ⊢ A and A ⊢ B, then 𝛤 ⊢ B. It follows also that
if A1 , . . . ,An ⊢ B and 𝛤 ⊢ Ai for each i , then 𝛤 ⊢ B.

Proposition C.18. 𝛤 is inconsistent iff 𝛤 ⊢ A for every sentence A.

Proof. Exercise. □
252 APPENDIX C. NATURAL DEDUCTION

Proposition C.19 (Compactness). 1. If 𝛤 ⊢ A then there is a

finite subset 𝛤0 ⊆ 𝛤 such that 𝛤0 ⊢ A.

2. If every finite subset of 𝛤 is consistent, then 𝛤 is consistent.

Proof. 1. If 𝛤 ⊢ A, then there is a derivation 𝛿 of A from 𝛤.

Let 𝛤0 be the set of undischarged assumptions of 𝛿. Since
any derivation is finite, 𝛤0 can only contain finitely many
sentences. So, 𝛿 is a derivation of A from a finite 𝛤0 ⊆ 𝛤.

2. This is the contrapositive of (1) for the special case A ≡ ⊥.

□

Summary
Proof systems provide purely syntactic methods for characteriz-
ing consequence and compatibility between sentences. Natural
deduction is one such proof system. A derivation in it con-
sists of a tree of formulas. The topmost formula a derivation are
assumptions. All other formulas, for the derivation to be cor-
rect, must be correctly justified by one of a number of inference
rules. These come in pairs; an introduction and an elimination
rule for each connective and quantifier. For instance, if a for-
mula A is justified by a →Elim rule, the preceding formulas (the
premises) must be B → A and B (for some B). Some inference
rules also allow assumptions to be discharged. For instance, if
A → B is inferred from B using →Intro, any occurrences of A as
assumptions in the derivation leading to the premise B may be
discharged, given a label that is also recorded at the inference.
If there is a derivation with end formula A and all assump-
tions are discharged, we say A is a theorem and write ⊢ A. If all
undischarged assumptions are in some set 𝛤, we say A is deriv-
able from 𝛤 and write 𝛤 ⊢ A. If 𝛤 ⊢ ⊥ we say 𝛤 is inconsistent,
otherwise consistent. These notions are interrelated, e.g., 𝛤 ⊢ A
iff 𝛤 ∪ {¬A} ⊢ ⊥. They are also related to the corresponding
253 C.9. PROOF-THEORETIC NOTIONS

semantic notions, e.g., if 𝛤 ⊢ A then 𝛤 ⊨ A. This property of

natural deduction—what can be derived from 𝛤 is guaranteed to
be entailed by 𝛤—is called soundness. The soundness theo-
rem is proved by induction on the length of derivations, showing
that each individual inference preserves entailment of its conclu-
sion from open assumptions provided its premises are entailed
by their open assumptions.

Problems
Problem C.1. Give derivations of the following:

1. ¬(A → B) → (A ∧ ¬B)

2. (A → C ) ∨ (B → C ) from the assumption (A ∧ B) → C

Problem C.2. Give derivations of the following:

1. ∃y A(y) → B from the assumption ∀x (A(x) → B)

2. ∃x (A(x) → ∀y A(y))

Problem C.3. Prove that = is both symmetric and transitive, i.e.,

give derivations of ∀x ∀y (x = y →y = x) and ∀x ∀y ∀z ((x = y ∧y =
z) → x = z)

Problem C.4. Give derivations of the following formulas:

1. ∀x ∀y ((x = y ∧ A(x)) → A(y))

2. ∃x A(x) ∧ ∀y ∀z ((A(y) ∧ A(z )) → y = z ) → ∃x (A(x) ∧

∀y (A(y) → y = x))

Problem C.5. Prove Proposition C.18

APPENDIX D

Biographies
D.1 Alonzo Church
Alonzo Church was born in
Washington, DC on June 14,
1903. In early childhood, an
air gun incident left Church
blind in one eye. He fin-
ished preparatory school in
Connecticut in 1920 and be-
gan his university education
at Princeton that same year.
He completed his doctoral
studies in 1927. After a cou-
ple years abroad, Church re-
turned to Princeton. Church
was known exceedingly polite Fig. D.1: Alonzo Church
and careful. His blackboard
writing was immaculate, and he would preserve important pa-
pers by carefully covering them in Duco cement (a clear glue).
Outside of his academic pursuits, he enjoyed reading science fic-
tion magazines and was not afraid to write to the editors if he
spotted any inaccuracies in the writing.
Church’s academic achievements were great. Together with
his students Stephen Kleene and Barkley Rosser, he developed

254
255 D.2. KURT GÖDEL

a theory of effective calculability, the lambda calculus, indepen-

dently of Alan Turing’s development of the Turing machine. The
two definitions of computability are equivalent, and give rise to
what is now known as the Church-Turing Thesis, that a function
of the natural numbers is effectively computable if and only if
it is computable via Turing machine (or lambda calculus). He
also proved what is now known as Church’s Theorem: The deci-
sion problem for the validity of first-order formulas is unsolvable.
Church continued his work into old age. In 1967 he left
Princeton for UCLA, where he was professor until his retirement
in 1990. Church passed away on August 1, 1995 at the age of 92.

Further Reading For a brief biography of Church, see En-

derton (2019). Church’s original writings on the lambda calcu-
lus and the Entscheidungsproblem (Church’s Thesis) are Church
(1936a,b). Aspray (1984) records an interview with Church about
the Princeton mathematics community in the 1930s. Church
wrote a series of book reviews of the Journal of Symbolic Logic from
1936 until 1979. They are all archived on John MacFarlane’s web-
site (MacFarlane, 2015).

D.2 Kurt Gödel

Kurt Gödel (ger-dle) was born on April 28, 1906 in Brünn in
the Austro-Hungarian empire (now Brno in the Czech Republic).
Due to his inquisitive and bright nature, young Kurtele was often
called “Der kleine Herr Warum” (Little Mr. Why) by his family.
He excelled in academics from primary school onward, where
he got less than the highest grade only in mathematics. Gödel
was often absent from school due to poor health and was exempt
from physical education. He was diagnosed with rheumatic fever
during his childhood. Throughout his life, he believed this per-
manently affected his heart despite medical assessment saying
otherwise.
256 APPENDIX D. BIOGRAPHIES

Gödel began studying at the University of Vienna in 1924

and completed his doctoral studies in 1929. He first intended to
study physics, but his interests soon moved to mathematics and
especially logic, in part due to the influence of the philosopher
Rudolf Carnap. His dissertation, written under the supervision
of Hans Hahn, proved the completeness theorem of first-order
predicate logic with identity (Gödel, 1929). Only a year later, he
obtained his most famous results—the first and second incom-
pleteness theorems (published in Gödel 1931). During his time
in Vienna, Gödel was heavily involved with the Vienna Circle,
a group of scientifically-minded philosophers that included Car-
nap, whose work was especially influenced by Gödel’s results.
In 1938, Gödel married
Adele Nimbursky. His par-
ents were not pleased: not
only was she six years older
than him and already di-
vorced, but she worked as a
dancer in a nightclub. So-
cial pressures did not affect
Gödel, however, and they re-
mained happily married until
his death.
After Nazi Germany an-
nexed Austria in 1938, Gödel
and Adele emigrated to the
United States, where he took
up a position at the Institute
for Advanced Study in Prince- Fig. D.2: Kurt Gödel
ton, New Jersey. Despite his
introversion and eccentric nature, Gödel’s time at Princeton was
collaborative and fruitful. He published essays in set theory, phi-
losophy and physics. Notably, he struck up a particularly strong
friendship with his colleague at the IAS, Albert Einstein.
In his later years, Gödel’s mental health deteriorated. His
wife’s hospitalization in 1977 meant she was no longer able to
257 D.3. RÓZSA PÉTER

cook his meals for him. Having suffered from mental health issues
throughout his life, he succumbed to paranoia. Deathly afraid of
being poisoned, Gödel refused to eat. He died of starvation on
January 14, 1978, in Princeton.

Further Reading For a complete biography of Gödel’s life is

available, see John Dawson (1997). For further biographical
pieces, as well as essays about Gödel’s contributions to logic and
philosophy, see Wang (1990), Baaz et al. (2011), Takeuti et al.
(2003), and Sigmund et al. (2007).
Gödel’s PhD thesis is available in the original German (Gödel,
1929). The original text of the incompleteness theorems is
(Gödel, 1931). All of Gödel’s published and unpublished writ-
ings, as well as a selection of correspondence, are available in
English in his Collected Papers Feferman et al. (1986, 1990).
For a detailed treatment of Gödel’s incompleteness theorems,
see Smith (2013). For an informal, philosophical discussion
of Gödel’s theorems, see Mark Linsenmayer’s podcast (Linsen-
mayer, 2014).

D.3 Rózsa Péter

Rózsa Péter was born Rósza Politzer, in Budapest, Hungary, on
February 17, 1905. She is best known for her work on recursive
functions, which was essential for the creation of the field of re-
cursion theory.
Péter was raised during harsh political times—WWI raged
when she was a teenager—but was able to attend the affluent
Maria Terezia Girls’ School in Budapest, from where she grad-
uated in 1922. She then studied at Pázmány Péter University
(later renamed Loránd Eötvös University) in Budapest. She be-
gan studying chemistry at the insistence of her father, but later
switched to mathematics, and graduated in 1927. Although she
had the credentials to teach high school mathematics, the eco-
nomic situation at the time was dire as the Great Depression af-
258 APPENDIX D. BIOGRAPHIES

fected the world economy. During this time, Péter took odd jobs
as a tutor and private teacher of mathematics. She eventually re-
turned to university to take up graduate studies in mathematics.
She had originally planned to work in number theory, but after
finding out that her results had already been proven, she almost
gave up on mathematics altogether. She was encouraged to work
on Gödel’s incompleteness theorems, and unknowingly proved
several of his results in different ways. This restored her confi-
dence, and Péter went on to write her first papers on recursion
theory, inspired by David Hilbert’s foundational program. She
received her PhD in 1935, and in 1937 she became an editor for
the Journal of Symbolic Logic.
Péter’s early papers are
widely credited as founding
contributions to the field of
recursive function theory. In
Péter (1935a), she investi-
gated the relationship be-
tween different kinds of recur-
sion. In Péter (1935b), she
showed that a certain recur-
sively defined function is not
primitive recursive. This sim-
plified an earlier result due to
Wilhelm Ackermann. Péter’s
simplified function is what’s
now often called the Ack- Fig. D.3: Rózsa Péter
ermann function—and some-
times, more properly, the Ackermann-Péter function. She wrote
the first book on recursive function theory (Péter, 1951).
Despite the importance and influence of her work, Péter did
not obtain a full-time teaching position until 1945. During the
Nazi occupation of Hungary during World War II, Péter was not
allowed to teach due to anti-Semitic laws. In 1944 the government
created a Jewish ghetto in Budapest; the ghetto was cut off from
the rest of the city and attended by armed guards. Péter was
259 D.4. JULIA ROBINSON

forced to live in the ghetto until 1945 when it was liberated. She
then went on to teach at the Budapest Teachers Training College,
and from 1955 onward at Eötvös Loránd University. She was the
first female Hungarian mathematician to become an Academic
Doctor of Mathematics, and the first woman to be elected to the
Hungarian Academy of Sciences.
Péter was known as a passionate teacher of mathematics, who
preferred to explore the nature and beauty of mathematical prob-
lems with her students rather than to merely lecture. As a result,
she was affectionately called “Aunt Rosa” by her students. Péter
died in 1977 at the age of 71.

Further Reading For more biographical reading, see

(O’Connor and Robertson, 2014) and (Andrásfai, 1986).
Tamassy (1994) conducted a brief interview with Péter. For
a fun read about mathematics, see Péter’s book Playing With
Infinity (Péter, 2010).

D.4 Julia Robinson

Julia Bowman Robinson was an American mathematician. She
is known mainly for her work on decision problems, and most
famously for her contributions to the solution of Hilbert’s tenth
problem. Robinson was born in St. Louis, Missouri, on December
8, 1919. Robinson recalls being intrigued by numbers already as
a child (Reid, 1986, 4). At age nine she contracted scarlet fever
and suffered from several recurrent bouts of rheumatic fever. This
forced her to spend much of her time in bed, putting her behind
in her education. Although she was able to catch up with the help
of private tutors, the physical effects of her illness had a lasting
impact on her life.
Despite her childhood struggles, Robinson graduated high
school with several awards in mathematics and the sciences. She
started her university career at San Diego State College, and
transferred to the University of California, Berkeley, as a senior.
260 APPENDIX D. BIOGRAPHIES

There she was influenced by the mathematician Raphael Robin-

son. They became good friends, and married in 1941. As a
spouse of a faculty member, Robinson was barred from teaching
in the mathematics department at Berkeley. Although she contin-
ued to audit mathematics classes, she hoped to leave university
and start a family. Not long after her wedding, however, Robin-
son contracted pneumonia. She was told that there was substan-
tial scar tissue build up on her heart due to the rheumatic fever
she suffered as a child. Due to the severity of the scar tissue, the
doctor predicted that she would not live past forty and she was
advised not to have children (Reid, 1986, 13).
Robinson was depressed
for a long time, but eventually
decided to continue studying
mathematics. She returned
to Berkeley and completed
her PhD in 1948 under the
supervision of Alfred Tarski.
The first-order theory of the
real numbers had been shown
to be decidable by Tarski,
and from Gödel’s work it fol-
lowed that the first-order the-
ory of the natural numbers
is undecidable. It was a ma-
jor open problem whether the
first-order theory of the ratio- Fig. D.4: Julia Robinson
nals is decidable or not. In
her thesis (1949), Robinson proved that it was not.
Interested in decision problems, Robinson next attempted to
find a solution to Hilbert’s tenth problem. This problem was one
of a famous list of 23 mathematical problems posed by David
Hilbert in 1900. The tenth problem asks whether there is an
algorithm that will answer, in a finite amount of time, whether
or not a polynomial equation with integer coefficients, such as
3x 2 − 2y + 3 = 0, has a solution in the integers. Such questions
261 D.5. ALFRED TARSKI

are known as Diophantine problems. After some initial successes,

Robinson joined forces with Martin Davis and Hilary Putnam,
who were also working on the problem. They succeeded in show-
ing that exponential Diophantine problems (where the unknowns
may also appear as exponents) are undecidable, and showed that
a certain conjecture (later called “J.R.”) implies that Hilbert’s
tenth problem is undecidable (Davis et al., 1961). Robinson
continued to work on the problem throughout the 1960s. In
1970, the young Russian mathematician Yuri Matijasevich finally
proved the J.R. hypothesis. The combined result is now called
the Matijasevich–Robinson–Davis–Putnam theorem, or MDRP
theorem for short. Matijasevich and Robinson became friends
and collaborated on several papers. In a letter to Matijasevich,
Robinson once wrote that “actually I am very pleased that work-
ing together (thousands of miles apart) we are obviously making
more progress than either one of us could alone” (Matijasevich,
1992, 45).
Robinson was the first female president of the American
Mathematical Society, and the first woman to be elected to the
National Academy of Science. She died on July 30, 1985 at the
age of 65 after being diagnosed with leukemia.

Further Reading Robinson’s mathematical papers are avail-

able in her Collected Works (Robinson, 1996), which also in-
cludes a reprint of her National Academy of Sciences biographi-
cal memoir (Feferman, 1994). Robinson’s older sister Constance
Reid published an “Autobiography of Julia,” based on interviews
(Reid, 1986), as well as a full memoir (Reid, 1996). A short
documentary about Robinson and Hilbert’s tenth problem was
directed by George Csicsery (Csicsery, 2016). For a brief mem-
oir about Yuri Matijasevich’s collaborations with Robinson, and
her influence on his work, see (Matijasevich, 1992).

D.5 Alfred Tarski

262 APPENDIX D. BIOGRAPHIES

Alfred Tarski was born on

January 14, 1901 in War-
saw, Poland (then part of the
Russian Empire). Described
as “Napoleonic,” Tarski was
boisterous, talkative, and in-
tense. His energy was often
reflected in his lectures—he
once set fire to a wastebasket
while disposing of a cigarette
during a lecture, and was for-
bidden from lecturing in that
building again.
Tarski had a thirst for
knowledge from a young age.
Although later in life he would Fig. D.5: Alfred Tarski
tell students that he studied
logic because it was the only class in which he got a B, his high
school records show that he got A’s across the board—even in
logic. He studied at the University of Warsaw from 1918 to 1924.
Tarski first intended to study biology, but became interested in
mathematics, philosophy, and logic, as the university was the
center of the Warsaw School of Logic and Philosophy. Tarski
earned his doctorate in 1924 under the supervision of Stanisław
Leśniewski.
Before emigrating to the United States in 1939, Tarski com-
pleted some of his most important work while working as a sec-
ondary school teacher in Warsaw. His work on logical conse-
quence and logical truth were written during this time. In 1939,
Tarski was visiting the United States for a lecture tour. During
his visit, Germany invaded Poland, and because of his Jewish her-
itage, Tarski could not return. His wife and children remained in
Poland until the end of the war, but were then able to emigrate to
the United States as well. Tarski taught at Harvard, the College
of the City of New York, and the Institute for Advanced Study
at Princeton, and finally the University of California, Berkeley.
263 D.5. ALFRED TARSKI

There he founded the multidisciplinary program in Logic and

the Methodology of Science. Tarski died on October 26, 1983 at
the age of 82.

Further Reading For more on Tarski’s life, see the biogra-

phy Alfred Tarski: Life and Logic (Feferman and Feferman, 2004).
Tarski’s seminal works on logical consequence and truth are avail-
able in English in (Corcoran, 1983). All of Tarski’s original works
have been collected into a four volume series, (Tarski, 1981).
Photo Credits
Alonzo Church, p. 254: Portrait of Alonzo Church, undated, pho-
tographer unknown. Alonzo Church Papers; 1924–1995, (C0948)
Box 60, Folder 3. Manuscripts Division, Department of Rare
Books and Special Collections, Princeton University Library. ⃝ C
Princeton University. The Open Logic Project has obtained per-
mission to use this image for inclusion in non-commercial OLP-
derived materials. Permission from Princeton University is re-
quired for any other use.
Kurt Gödel, p. 256: Portrait of Kurt Gödel, ca. 1925, photog-
rapher unknown. From the Shelby White and Leon Levy Archives
Center, Institute for Advanced Study, Princeton, NJ, USA, on de-
posit at Princeton University Library, Manuscript Division, De-
partment of Rare Books and Special Collections, Kurt Gödel Pa-
pers, (C0282), Box 14b, #110000. The Open Logic Project has
obtained permission from the Institute’s Archives Center to use
this image for inclusion in non-commercial OLP-derived materi-
als. Permission from the Archives Center is required for any other
use.
Rózsa Péter, p. 258: Portrait of Rózsa Péter, undated, photog-
rapher unknown. Courtesy of Béla Andrásfai.
Julia Robinson, p. 260: Portrait of Julia Robinson, unknown
photographer, courtesy of Neil D. Reid. The Open Logic Project
has obtained permission to use this image for inclusion in non-
commercial OLP-derived materials. Permission is required for
any other use.

264
265 Photo Credits

Alfred Tarski, p. 262: Passport photo of Alfred Tarski, 1939.

Cropped and restored from a scan of Tarski’s passport by Joel
Fuller. Original courtesy of Bancroft Library, University of Cal-
ifornia, Berkeley. Alfred Tarski Papers, Banc MSS 84/49. The
Open Logic Project has obtained permission to use this image
for inclusion in non-commercial OLP-derived materials. Permis-
sion from Bancroft Library is required for any other use.
Bibliography
Andrásfai, Béla. 1986. Rózsa (Rosa) Péter. Periodica Polytechnica
Electrical Engineering 30(2-3): 139–145. URL http://www.pp.
bme.hu/ee/article/view/4651.

Aspray, William. 1984. The Princeton mathematics community

in the 1930s: Alonzo Church. URL http://www.princeton.
edu/mudd/finding_aids/mathoral/pmc05.htm. Interview.

Baaz, Matthias, Christos H. Papadimitriou, Hilary W. Putnam,

Dana S. Scott, and Charles L. Harper Jr. 2011. Kurt Gödel and
the Foundations of Mathematics: Horizons of Truth. Cambridge:
Cambridge University Press.

Church, Alonzo. 1936a. A note on the Entscheidungsproblem.

Journal of Symbolic Logic 1: 40–41.

Church, Alonzo. 1936b. An unsolvable problem of elementary

number theory. American Journal of Mathematics 58: 345–363.

Corcoran, John. 1983. Logic, Semantics, Metamathematics. Indi-

anapolis: Hackett, 2nd ed.

Csicsery, George. 2016. Zala films: Julia Robinson and Hilbert’s

tenth problem. URL http://www.zalafilms.com/films/
juliarobinson.html.

266
267 BIBLIOGRAPHY

Davis, Martin, Hilary Putnam, and Julia Robinson. 1961. The

decision problem for exponential Diophantine equations. An-
nals of Mathematics 74(3): 425–436. URL http://www.jstor.
org/stable/1970289.
Enderton, Herbert B. 2019. Alonzo Church: Life and Work. In
The Collected Works of Alonzo Church, eds. Tyler Burge and Her-
bert B. Enderton. Cambridge, MA: MIT Press.
Feferman, Anita and Solomon Feferman. 2004. Alfred Tarski: Life
and Logic. Cambridge: Cambridge University Press.
Feferman, Solomon. 1994. Julia Bowman Robinson 1919–1985.
Biographical Memoirs of the National Academy of Sciences 63:
1–28. URL http://www.nasonline.org/publications/
biographical-memoirs/memoir-pdfs/robinson-julia.
pdf.
Feferman, Solomon, John W. Dawson Jr., Stephen C. Kleene, Gre-
gory H. Moore, Robert M. Solovay, and Jean van Heijenoort.
1986. Kurt Gödel: Collected Works. Vol. 1: Publications 1929–1936.
Oxford: Oxford University Press.
Feferman, Solomon, John W. Dawson Jr., Stephen C. Kleene, Gre-
gory H. Moore, Robert M. Solovay, and Jean van Heijenoort.
1990. Kurt Gödel: Collected Works. Vol. 2: Publications 1938–1974.
Oxford: Oxford University Press.
Gödel, Kurt. 1929. Über die Vollständigkeit des Logikkalküls
[On the completeness of the calculus of logic]. Dissertation,
Universität Wien. Reprinted and translated in Feferman et al.
(1986), pp. 60–101.
Gödel, Kurt. 1931. über formal unentscheidbare Sätze der Prin-
cipia Mathematica und verwandter Systeme I [On formally unde-
cidable propositions of Principia Mathematica and related sys-
tems I]. Monatshefte für Mathematik und Physik 38: 173–198.
Reprinted and translated in Feferman et al. (1986), pp. 144–
195.
268 BIBLIOGRAPHY

John Dawson, Jr. 1997. Logical Dilemmas: The Life and Work of
Kurt Gödel. Boca Raton: CRC Press.

Linsenmayer, Mark. 2014. The partially examined life: Gödel

on math. URL http://www.partiallyexaminedlife.com/
2014/06/16/ep95-godel/. Podcast audio.

MacFarlane, John. 2015. Alonzo Church’s JSL reviews. URL

http://johnmacfarlane.net/church.html.

Matijasevich, Yuri. 1992. My collaboration with Julia Robinson.

The Mathematical Intelligencer 14(4): 38–45.

O’Connor, John J. and Edmund F. Robertson. 2014. Rózsa Péter.

URL http://www-groups.dcs.st-and.ac.uk/~history/
Biographies/Peter.html.

Péter, Rózsa. 1935a. Über den Zusammenhang der verschiede-

nen Begriffe der rekursiven Funktion. Mathematische Annalen
110: 612–632.

Péter, Rózsa. 1935b. Konstruktion nichtrekursiver Funktionen.

Mathematische Annalen 111: 42–60.

Péter, Rózsa. 1951. Rekursive Funktionen. Budapest: Akademiai

Kiado. English translation in (Péter, 1967).

Péter, Rózsa. 1967. Recursive Functions. New York: Academic

Press.

Péter, Rózsa. 2010. Playing with Infinity. New York: Dover. URL
https://books.google.ca/books?id=6V3wNs4uv_4C&lpg=
PP1&ots=BkQZaHcR99&lr&pg=PP1#v=onepage&q&f=false.

Reid, Constance. 1986. The autobiography of Julia Robinson.

The College Mathematics Journal 17: 3–21.

Reid, Constance. 1996. Julia: A Life in Mathemat-

ics. Cambridge: Cambridge University Press. URL
269 BIBLIOGRAPHY

https://books.google.ca/books?id=lRtSzQyHf9UC&
lpg=PP1&pg=PP1#v=onepage&q&f=false.

Robinson, Julia. 1949. Definability and decision problems in

arithmetic. Journal of Symbolic Logic 14(2): 98–114. URL
http://www.jstor.org/stable/2266510.

Robinson, Julia. 1996. The Collected Works of Julia Robinson. Prov-

idence: American Mathematical Society.

Sigmund, Karl, John Dawson, Kurt Mühlberger, Hans Magnus

Enzensberger, and Juliette Kennedy. 2007. Kurt Gödel: Das
Album–The Album. The Mathematical Intelligencer 29(3): 73–
76.

Smith, Peter. 2013. An Introduction to Gödel’s Theorems. Cambridge:

Cambridge University Press.

Takeuti, Gaisi, Nicholas Passell, and Mariko Yasugi. 2003. Mem-

oirs of a Proof Theorist: Gödel and Other Logicians. Singapore:
World Scientific.

Tamassy, Istvan. 1994. Interview with Róza Péter. Modern Logic

4(3): 277–280.

Tarski, Alfred. 1981. The Collected Works of Alfred Tarski, vol. I–IV.
Basel: Birkhäuser.

Wang, Hao. 1990. Reflections on Kurt Gödel. Cambridge: MIT

Press.
About the Open
Logic Project
The Open Logic Text is an open-source, collaborative textbook of
formal meta-logic and formal methods, starting at an intermedi-
ate level (i.e., after an introductory formal logic course). Though
aimed at a non-mathematical audience (in particular, students of
philosophy and computer science), it is rigorous.
Coverage of some topics currently included may not yet be
complete, and many sections still require substantial revision.
We plan to expand the text to cover more topics in the future.
We also plan to add features to the text, such as a glossary, a
list of further reading, historical notes, pictures, better explana-
tions, sections explaining the relevance of results to philosophy,
computer science, and mathematics, and more problems and ex-
amples. If you find an error, or have a suggestion, please let the
project team know.
The project operates in the spirit of open source. Not only
is the text freely available, we provide the LaTeX source un-
der the Creative Commons Attribution license, which gives any-
one the right to download, use, modify, re-arrange, convert, and
re-distribute our work, as long as they give appropriate credit.
Please see the Open Logic Project website at openlogicproject.org
for additional information.

270