Sider Theodore - Logic For Philosophy

Logic for Philosophy
Theodore Sider
June 2, 2008
Preface
This book is an elementary introduction to the logic that students of contempo-

rary philosophy ought to know. It covers i) basic approaches to logic, including
proof theory and especially model theory, ii) extensions of standard logic (such
as modal logic) that are important in philosophy, and iii) some elementary
philosophy of logic. It prepares students to read the logically sophisticated
articles in today’s philosophy journals, and helps them resist bullying by symbol-
mongerers. In short, it teaches the logic necessary for being a contemporary
philosopher.
For better or for worse (I think better), the last century-or-so’s developments
in logic are part of the shared knowledge base of philosophers, and inform, in
varying degrees of directness, nearly every area of philosophy. Logic is part
of our shared language and inheritance. The standard philosophy curriculum
therefore includes a healthy dose of logic. This is a good thing. But the
advanced logic that is part of this curriculum is usually a course in mathematical
logic, which usually means an intensive course in metalogic (for example, a
course based on the excellent Boolos and Jeffrey (1989).) I do believe in the
value of such a course. But advanced undergraduate philosophy majors and
beginning graduate students often take but a single advanced logic course; and
if there is to be only one, it should not, I think, be a course in metalogic. The
standard metalogic course is too mathematically demanding for the average
philosophy student, and omits material that the average student needs to know.
If there is to be only one advanced logic course, let it be a course designed to
instill logical literacy.
I begin with a sketch of standard propositional and predicate logic (de-
veloped more formally than in a typical intro course.) I briefly discuss a few
extensions and variations on each (e.g., three-valued logic, definite descrip-
tions). I then discuss modal logic and counterfactual conditionals in detail. I
presuppose familiarity with the contents of a typical intro logic course: the
i
PREFACE ii
meanings of the logical symbols of first-order predicate logic without identity

or function symbols; truth tables; translations from English into propositional
and predicate logic; some proof system (e.g., natural deduction) in propositional
and predicate logic.
I drew heavily from the following sources, which would be good for supple-
mental reading:
· Propositional logic: Mendelson (1987)
· Descriptions, multi-valued logic: Gamut (1991a)
· Sequents: Lemmon (1965)
· Further quantifiers: Glanzberg (2006); Sher (1991, chapter 2); Wester-

ståhl (1989); Boolos and Jeffrey (1989, chapter 18)
· Modal logic: Gamut (1991b); Cresswell and Hughes (1996)
· Semantics for intuitionism : Priest (2001)
· Counterfactuals: Lewis (1973)
· Two-dimensional modal logic: Davies and Humberstone (1980)
Another important source was Ed Gettier’s 1988 modal logic class at the Uni-
versity of Massachusetts. My notes from that course formed the basis of the
first incarnation of this work.
I am also deeply grateful for feedback from colleagues, and from students
in courses on this material. In particular, Marcello Antosh, Josh Armstrong,
Gabe Greenberg, Angela Harper, Sami Laine, Gregory Lavers, Alex Morgan,
Jeff Russell, Brock Sides, Jason Turner, Crystal Tychonievich, Jennifer Wang,
Brian Weatherson, and Evan Williams: thank you.
Contents
Preface i
1 Nature of Logic 1
1.1 Logical consequence and logical truth . . . . . . . . . . . . . . . . . 2
1.2 Form and abstraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Formal logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Correctness and application . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 The nature of logical consequence . . . . . . . . . . . . . . . . . . . 8
1.6 Extensions, deviations, variations . . . . . . . . . . . . . . . . . . . . 10
1.6.1 Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6.2 Deviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6.3 Variations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.7 Metalogic, metalanguages, and formalization . . . . . . . . . . . . 12
1.8 Set theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 Propositional Logic 18
2.1 Grammar of PL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2 The semantic approach to logic . . . . . . . . . . . . . . . . . . . . . 21
2.3 Semantics of PL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Natural deduction in PL . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.1 Sequents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.2 Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.4.3 Sequent proofs . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.4.4 Example sequent proofs . . . . . . . . . . . . . . . . . . . . . 36
2.5 Axiomatic proofs in PL . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.5.1 Example axiomatic proofs . . . . . . . . . . . . . . . . . . . . 43
2.6 Soundness of PL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.7 Completeness of PL . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
iii
CONTENTS iv
3 Variations and Deviations from PL 63

3.1 Alternate connectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.1.1 Symbolizing truth functions in propositional logic . . . 63
3.1.2 Inadequate connective sets . . . . . . . . . . . . . . . . . . . 67
3.1.3 Sheffer stroke . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.2 Polish notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3 Multi-valued logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.3.1 Łukasiewicz’s system . . . . . . . . . . . . . . . . . . . . . . . 72
3.3.2 Kleene’s “strong” tables . . . . . . . . . . . . . . . . . . . . . 74
3.3.3 Kleene’s “weak” tables (Bochvar’s tables) . . . . . . . . . . 76
3.3.4 Supervaluationism . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.4 Intuitionism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4 Predicate Logic 84
4.1 Grammar of predicate logic . . . . . . . . . . . . . . . . . . . . . . . 84
4.2 Semantics of predicate logic . . . . . . . . . . . . . . . . . . . . . . . 85
4.3 Establishing validity and invalidity . . . . . . . . . . . . . . . . . . . 90
5 Extensions of Predicate Logic 93

5.1 Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.1.1 Grammar for the identity sign . . . . . . . . . . . . . . . . . 93
5.1.2 Semantics for the identity sign . . . . . . . . . . . . . . . . 94
5.1.3 Symbolizations with the identity sign . . . . . . . . . . . 95
5.2 Function symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.2.1 Grammar for function symbols . . . . . . . . . . . . . . . . 99
5.2.2 Semantics for function symbols . . . . . . . . . . . . . . . . 100
5.3 Definite descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.3.1 Grammar for ι . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3.2 Semantics for ι . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.3.3 Eliminability of function symbols and definite descriptions106
5.4 Further quantifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.4.1 Generalized monadic quantifiers . . . . . . . . . . . . . . . 111
5.4.2 Generalized binary quantifiers . . . . . . . . . . . . . . . . . 113
5.4.3 Second-order logic . . . . . . . . . . . . . . . . . . . . . . . . 115
6 Propositional Modal Logic 118

6.1 Grammar of MPL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.2 Symbolizations in MPL . . . . . . . . . . . . . . . . . . . . . . . . . . 120
CONTENTS v
6.3 Semantics for MPL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

6.3.1 Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.3.2 Kripke models . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.3.3 Semantic validity proofs . . . . . . . . . . . . . . . . . . . . . 131
6.3.4 Countermodels . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.3.5 Schemas, validity, and invalidity . . . . . . . . . . . . . . . . 152
6.4 Axiomatic systems of MPL . . . . . . . . . . . . . . . . . . . . . . . . 154
6.4.1 System K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.4.2 System D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
6.4.3 System T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.4.4 System B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.4.5 System S4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.4.6 System S5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
6.4.7 Substitution of equivalents and modal reduction . . . . . 171
6.5 Soundness in MPL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.5.1 Soundness of K . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6.5.2 Soundness of T . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6.5.3 Soundness of B . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
6.6 Completeness of MPL . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
6.6.1 Canonical models . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.6.2 Maximal consistent sets of wffs . . . . . . . . . . . . . . . . 179
6.6.3 Definition of canonical models . . . . . . . . . . . . . . . . 180
6.6.4 Features of maximal consistent sets . . . . . . . . . . . . . 181
6.6.5 Maximal consistent extensions . . . . . . . . . . . . . . . . . 182
6.6.6 “Mesh” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.6.7 Truth and membership in canonical models . . . . . . . . 187
6.6.8 Completeness of systems of MPL . . . . . . . . . . . . . . 188
7 Variations on MPL 191

7.1 Propositional tense logic . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.1.1 The metaphysics of time . . . . . . . . . . . . . . . . . . . . 191
7.1.2 Tense operators . . . . . . . . . . . . . . . . . . . . . . . . . . 193
7.1.3 Syntax of tense logic . . . . . . . . . . . . . . . . . . . . . . . 195
7.1.4 Possible worlds semantics for tense logic . . . . . . . . . . 195
7.1.5 Formal constraints on ≤ . . . . . . . . . . . . . . . . . . . . . 197
7.2 Intuitionist propositional logic . . . . . . . . . . . . . . . . . . . . . . 199
7.2.1 Kripke semantics for intuitionist propositional logic . . 199
7.2.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
CONTENTS vi
7.2.3 Soundness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
8 Counterfactuals 207
8.1 Natural language counterfactuals . . . . . . . . . . . . . . . . . . . . 208
8.1.1 Not truth-functional . . . . . . . . . . . . . . . . . . . . . . . 208
8.1.2 Can be contingent . . . . . . . . . . . . . . . . . . . . . . . . . 208
8.1.3 No augmentation . . . . . . . . . . . . . . . . . . . . . . . . . 209
8.1.4 No contraposition . . . . . . . . . . . . . . . . . . . . . . . . . 210
8.1.5 Some implications . . . . . . . . . . . . . . . . . . . . . . . . . 210
8.1.6 Context dependence . . . . . . . . . . . . . . . . . . . . . . . 211
8.2 The Lewis/Stalnaker approach . . . . . . . . . . . . . . . . . . . . . 213
8.3 Stalnaker’s system (SC) . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
8.3.1 Syntax of SC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
8.3.2 Semantics of SC . . . . . . . . . . . . . . . . . . . . . . . . . . 214
8.4 Validity proofs in SC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
8.5 Countermodels in SC . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
8.6 Logical Features of SC . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
8.6.1 Not truth-functional . . . . . . . . . . . . . . . . . . . . . . . 230
8.6.2 Can be contingent . . . . . . . . . . . . . . . . . . . . . . . . . 230
8.6.3 No augmentation . . . . . . . . . . . . . . . . . . . . . . . . . 230
8.6.4 No contraposition . . . . . . . . . . . . . . . . . . . . . . . . . 231
8.6.5 Some implications . . . . . . . . . . . . . . . . . . . . . . . . . 231
8.6.6 No exportation . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
8.6.7 No importation . . . . . . . . . . . . . . . . . . . . . . . . . . 232
8.6.8 No hypothetical syllogism (transitivity) . . . . . . . . . . . 233
8.6.9 No transposition . . . . . . . . . . . . . . . . . . . . . . . . . . 234
8.7 Lewis’s criticisms of Stalnaker’s theory . . . . . . . . . . . . . . . . 234
8.8 Lewis’s system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
8.9 The problem of disjunctive antecedents . . . . . . . . . . . . . . . 241
9 Quantified Modal Logic 243

9.1 Grammar of QML . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
9.2 Symbolizations in QML . . . . . . . . . . . . . . . . . . . . . . . . . . 243
9.3 A simple semantics for QML . . . . . . . . . . . . . . . . . . . . . . . 246
9.4 Countermodels and validity proofs in SQML . . . . . . . . . . . . 248
9.5 Philosophical questions about SQML . . . . . . . . . . . . . . . . . 254
9.5.1 The necessity of identity . . . . . . . . . . . . . . . . . . . . 254
9.5.2 The necessity of existence . . . . . . . . . . . . . . . . . . . 256
CONTENTS vii
9.5.3 Necessary existence defended . . . . . . . . . . . . . . . . . 261

9.6 Variable domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
9.6.1 Countermodels to the Barcan and related formulas . . . 266
9.6.2 Expanding, shrinking domains . . . . . . . . . . . . . . . . 267
9.6.3 Strong and weak necessity . . . . . . . . . . . . . . . . . . . 269
10 Two-dimensional modal logic 272

10.1 Actuality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
10.1.1 Kripke models with designated worlds . . . . . . . . . . . 273
10.1.2 Semantics for @ . . . . . . . . . . . . . . . . . . . . . . . . . . 274
10.1.3 Establishing validity and invalidity . . . . . . . . . . . . . . 275
10.2 × . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
10.2.1 Two-dimensional semantics for × . . . . . . . . . . . . . . 277
10.3 Fixedly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
10.4 A philosophical application: necessity and a priority . . . . . . . . 282
A Answers to Selected Exercises 289
B Answers to Remaining Exercises 298
References 350
Chapter 1
Nature of Logic
ince you are reading this book, you are probably already familiar with
S some logic. You probably know how to translate English sentences into
symbolic notation—into propositional logic:
English Propositional logic

If snow is white then grass is green S→G
Either snow is white or grass is not green S∨∼G
and into predicate logic:
English Predicate logic

If Jones is happy then someone is happy H j →∃xH x
Any friend of Jones is either insane or ∀x[F x j →(I x ∨ ∀yF xy)]
friends with everyone
You are probably also familiar with some basic techniques for evaluating argu-
ments written out in symbolic notation. You have probably encountered truth
tables, and some form of proof theory (perhaps a “natural deduction” system;
perhaps “truth trees”.) You may have even encountered some elementary model
theory. In short: you had an introductory course in symbolic logic.
What you already have is: literacy in elementary logic. What you will
get out of this book is: literacy in the rest of logic that philosophers tend to
presuppose, plus a deeper grasp of what logic is all about.
So what is logic all about?
1
CHAPTER 1. NATURE OF LOGIC 2
1.1 Logical consequence and logical truth

Logic is about logical consequence. The statement “someone is happy” is a logical
consequence of the statement “Ted is happy”. If Ted is happy, then it logically
follows that someone is happy. Put another way: the statement “Ted is happy”
logically implies the statement “someone is happy”. Likewise, the statement
“Ted is happy” is a logical consequence of the statements “It’s not the case that
John is happy” and “Either John is happy or Ted is happy”. The first statement
follows from the latter two statements. If the latter two statements are true,
the former must be true. Put another way: the argument whose premises are
the latter two statements, and whose conclusion is the former statement, is a
logically correct one.1
Relatedly, logic is about logical truth. A logical truth is a sentence that is
“true purely by virtue of logic”. Examples might include: “it’s not the case that
snow is white and also not white”, “All fish are fish”, and “If Ted is happy then
someone is happy”. It is plausible that logical truth and logical consequence
are related thus: a logical truth is a sentence that is a logical consequence of
any sentences whatsoever.
1.2 Form and abstraction

Logicians focus on form. Consider again the following argument:
It’s not the case that John is happy
Ted is happy or John is happy (Argument A)
Therefore, Ted is happy
Argument A is logically correct—its conclusion is a logical consequence of its
premises. It is customary to say that this is so in virtue of its form—in virtue of
the fact that its form is:
It’s not the case that φ
φ or ψ
Therefore ψ
1
The word ‘valid’ is sometimes used for logically correct arguments, but I will reserve that
word for a different concept: that of a logical truth according to the semantic conception of
logical truth.
Likewise, we say that “it’s not the case that snow is white and snow is not white”
is a logical truth because it has the form: it’s not the case that φ and not-φ.
We need to think hard about the idea of form. Apparently, we got the
alleged form of Argument A by replacing some words with Greek letters and
leaving other words as they were. We replaced the sentences ‘John is happy’ and
‘Ted is happy’ with φ and ψ, respectively, but left the expressions ‘It’s not the
case that’ and ‘or’ as they were, resulting in the schematic form displayed above.
Let’s call that form, “Form 1”. What’s so special about Form 1? Couldn’t we
make other choices for what to leave and what to replace? For instance, if we
replace the predicate ‘is happy’ with the schematic letter α, leaving the rest
intact, we get this:
It’s not the case that John is α

Ted is α or John is α (Form 2)
Therefore, Ted is α
And if we replace the ‘or’ with the schematic letter γ and leave the rest intact,
then we get this:

Ted is happy γ John is happy (Form 3)
If we think of Argument A as having Form 1, then we can think of it as being

logically correct in virtue of its form, since every “instance” of Form 1 is
logically correct. That is, no matter what sentences we substitute in for the
greek letters φ and ψ in Form 1, the result is a logically correct argument.
Now, if we think of Argument A’s form as being Form 2, we can continue to
think of Argument A as being logically correct in virtue of its form, since, like
Form 1, every instance of Form 2 is logically correct: no matter what predicate
we change α to, Form 2 becomes a logically correct argument. But if we think
of Argument A’s form as being Form 3, then we cannot think of it as being
logically correct in virtue of its form, for not every instance of Form 3 is a
logically correct argument. If we change γ to ‘if and only if’, for example, then
we get the following logically incorrect argument:

Ted is happy if and only if John is happy
So, what did we mean, when we said that Argument A is logically correct in
virtue of its form? What is Argument A’s form? Is it Form 1, Form 2, or Form
3?
There is no such thing as the form of an argument. When we assign an
argument a form, what we are doing is focusing on certain words and ignoring
others. We leave intact the words we’re focusing on, and we insert schematic
letters for the rest. Thus, in assigning Argument A Form 1, we’re focusing on
the words (phrases) ‘it is not the case that’ and ‘or’, and ignoring other words.
More generally, in (standard) propositional logic, we focus on the phrases
‘if…then’, ‘if and only if’, ‘and’, ‘or’, and so on, and ignore others. We do this
in order to investigate the relations of logical consequence that hold in virtue
of these words’ meaning. The fact that Argument A is logically correct depends
just on the meaning of the phrases ‘it is not the case that’ and ‘or’; it does not
depend on the meanings of the sentences ‘John is happy’ and ‘Ted is happy’.
We can substitute any sentences we like for ‘φ’ and ‘ψ’ in Form 1 and still get
a valid argument.
In predicate logic, on the other hand, we focus on further words: ‘all’ and
‘some’. Broadening our focus in this way allows us to capture a wider range
of logical consequences and logical truths. For example “If Ted is happy then
someone is happy” is a logical truth in virtue of the meaning of ‘someone’, but
not merely in virtue of the meanings of the characteristic words of propositional
logic.
Call the words on which we’re focusing—that is, the words that we leave
intact when we construct the forms of sentences and arguments—the logical
constants. (We can speak of natural language logical constants—‘and’, ‘or’, etc.
for propositional logic; ‘all’ and ‘some’ in addition for predicate logic—as well
as symbolic logical constants: ∧, ∨, etc. for propositional logic; ∀ and ∃ in
addition for predicate logic.) What we’ve seen is that the forms we assign
depend on what we’re considering to be the logical constants.
We call these expressions logical constants because we interpret them in a
constant way in logic, in contrast to other terms. For example, ∧ is a logical
constant; in propositional logic, it always stands for conjunction. There are
fixed rules governing ∧, in proof systems (the rule that from P ∧Q one can
infer P , for example), in the rules for constructing truth tables, and so on.
Moreover, these rules are distinctive for ∧: there are different rules for other
logical constants such as ∨. In contrast, the terms in logic that are not logical
constants do not have fixed, particular rules governing their meanings. For
example, there are no special rules governing what one can do with a P as
opposed to a Q in proofs or truth tables. That’s because P doesn’t symbolize

any sentence in particular; it can stand for any old sentence.
There isn’t anything sacred about the choices of logical constants we make
in propositional and predicate logic; and therefore, there isn’t anything sacred
about the customary forms we assign to sentences. We could treat other words
as logical constants. We could, for example, stop taking ‘or’ as a logical constant,
and instead take ‘It’s not the case that John is happy’, ‘Ted is happy’, and ‘John
is happy’ as logical constants. We would thereby view Argument A as having
Form 3. This would not be a particularly productive choice (since it would not
help to explain the correctness of Argument A), but it’s not wrong simply by
virtue of the concept of form.
More interestingly, consider the fact that every argument of the following
form is logically correct:
α is a bachelor
Therefore, α is unmarried
Accordingly, we could treat the predicates ‘is a bachelor’ and ‘is unmarried’
as logical constants, and develop a corresponding logic. We could introduce
special symbolic logical constants for these predicates, we could introduce
distinctive rules governing these predicates in proofs. (The rule of “bachelor-
elimination”, for instance, might allow one to infer “α is unmarried” from “α
is a bachelor”.) As with the choices of the previous paragraph, this choice of
what to treat as a logical constant is also not ruled out by the concept of form.
And it would be more productive than the choices of the last paragraph. Still,
it would be far less productive than the usual choices of logical constants in
predicate and propositional logic. The word ‘bachelor’ doesn’t have as general
application as the words commonly treated as logical constants in propositional
and predicate logic; the latter are ubiquitous.
At least, this remark about “generality” is one idea about what should be
considered a “logical constant”, and hence one idea about the scope of what
is usually thought of as “logic”. Where to draw the boundaries of logic—and
indeed, whether the logic/nonlogic boundary is an important one to draw—is
an open philosophical question about logic. At any rate, in this course, one
thing we’ll do is study systems that expand the list of logical constants from
standard propositional and predicate logic.
1.3 Formal logic

Modern logic is “mathematical” or “formal” logic. This means simply that
one studies logic using mathematical techniques. More carefully: in order
to develop theories of logical consequence, and logical truth, one develops a
formal language (see below), one treats the sentences of the formal language as
mathematical objects; one uses the tools of mathematics (especially, the tools
of very abstract mathematics, such as set theory) to formulate theories about
the sentences in the formal language; and one applies mathematical standards
of rigor to these theories. Mathematical logic was originally developed to study
mathematical reasoning2 , but its techniques are now applied to reasoning of all
kinds.
Think, for example, of propositional logic (this will be our first topic below).
The standard approach to analyzing the logical behavior of ‘and’, ‘or’, and so
on, is to develop a certain formal language, the language of propositional logic.
The sentences of this language look like this:
P
(Q→R)∨(Q→∼S)
P ↔(P ∧Q)
The symbols ∧, ∨, etc., are used to represent the English words ‘and’, ‘or’, and
so on (the logical constants for propositional logic), and the sentence letters
P, Q, etc., are used to represent declarative English sentences.
Why ‘formal’? Because we stipulate, in a mathematically rigorous way,
a grammar for the language; that is, we stipulate a mathematically rigorous
definition of the idea of a sentence of this language. Moreover, since we are
only interested in the logical behavior of the chosen logical constants ‘and’,
‘or’, and so on, we choose special symbols (∧, ∨ . . . ) for these words only; we
use P, Q, R, . . . indifferently to represent any English sentence whose internal
logical structure we are willing to ignore.3
We go on, then, to study (as always, in a mathematically rigorous way) vari-
ous concepts that apply to the sentences in formal languages. In propositional
logic, for example, one constructs a mathematically rigorous definition of a
2
Notes
3
Natural languages like English also have a grammar, and the grammar can be studied
using mathematical techniques. But the grammar is much more complicated, and is discovered
rather than stipulated; and natural languages lack abstractions like the sentence letters.
tautology (“all Trues in the truth table”), and a rigorous definition of a prov-
able formula (e.g., in terms of a system of deduction, using rules of inference,
assumptions, and so on).
Of course, the real goal is to apply the notions of logical consequence and
logical truth to sentences of English and other natural languages. The formal
languages are merely a tool; we need to apply the tool.
1.4 Correctness and application

To apply the tools we develop for formal languages, we need to speak of a
formal system as being correct. What does that sort of claim mean?
As we saw, logicians use formal languages and formal structures to study
logical consequence and logical truth. And the range of structures that one
could in principle study is very wide. For example, I could introduce a new
notion of “provability” by saying “in Ted Logic, the following rule may be
used when constructing proofs: if you have P on a line, you may infer ∼P .
The annotation is ‘T’.” I could then go on to investigate the properties of
such a system. Logic can be viewed as a branch of mathematics, and we can
mathematically study any system we like, including a system (like Ted logic) in
which one can “prove” ∼P from P .
But no such formal system would shed light on genuine logical consequence
and genuine logical truth. It would be implausible to claim, for example, that
when we translate an English argument into symbols, the conclusion of the
resulting symbolic argument may be derived in Ted logic from its premises iff
the conclusion of the original English argument is a logical consequence of its
premises.
Thus, the existence of a coherent, specifiable logical system must be dis-
tinguished from its application. When we say that a logical system is correct,
we have in mind some application of that system. Here’s an oversimplified
account of one such correctness claim. Suppose we have developed a certain
formal system for constructing proofs in propositional logic. And suppose
we have specified some translation scheme from English into the language
of propositional logic. This translation schema would translate English ‘and’
into the logical ∧ , English ‘or’ into the logical ∨, and so on. Then, the claim
that the formal system gives a correct logic of English ‘and’, ‘or’, etc. might be
taken to be the claim that one English sentence is a logical consequence of
some other English sentences in virtue of ‘and’, ‘or’, etc., iff one can prove the
translation of the former English sentence from the translations of the latter
English sentences in the formal system.
In this book I won’t spend much time on philosophical questions about
which formal systems are correct. My goal is rather to introduce those for-
malisms that are ubiquitous in philosophy, to give you the tools you need to
address such philosophical questions yourself. Still, from time to time, we’ll dip
just a bit into these philosophical questions, in order to motivate our choices
of logical systems to study.
1.5 The nature of logical consequence

The previous section discussed what it means to say that a formal theory gives a
correct account of logical consequence (as applied to sentences of English and
other natural languages). But what is it for sentences to stand in the relation of
logical consequence? What is logical consequence?
The question here is a philosophical question, as opposed to a mathematical
one. Logicians define various notions concerning sentences of formal languages:
derivability in this or that proof-system, “all trues in the truth table”, and so
on. They thereby stipulatively introduce various formal concepts. These
formal concepts are good insofar as they correctly model logical truth and
logical consequence. But in what do logical truth and logical consequence—the
intuitive concepts, as opposed to the stipulatively introduced concepts—consist?
This is one of the core questions of philosophical logic.
This book is not primarily a book in philosophical logic, so we won’t spend
much time on the question. However, I do want to make clear that the question
is indeed a question. The question is sometimes obscured by the fact that terms
like ‘logical truth’ are often stipulatively defined in logic books. This can lead
to the belief that there are no genuine issues concerning these notions. It is also
obscured by the fact that one philosophical theory of these notions—the model-
theoretic one—is so dominant that one can forget that it is a nontrivial theory.
Stipulative definitions are of course not things whose truth can be questioned;
but stipulative definitions of logical notions are good insofar as the stipulated
notions accurately model the real, intuitive, nonstipulated notions of logical
consequence and logical truth. Further, the stipulated definitions generally
concern formal languages, whereas the ultimate goal is an understanding of
correct reasoning of the sort that we actually do, using natural languages.
Let’s focus just on logical consequence. Here is a quick survey of some
competing philosophical accounts of its nature. Probably the most standard

account is the semantic, or model-theoretic one. Intuitively, a logical truth is
“true no matter what”. The model theoretic account is one way of making this
slogan precise. It says that φ is a logical consequence of the sentences in set Γ
if the formal translation of φ is true in every model (interpretation) in which
the formal translations of the members of Γ are true. This account needs to be
spelled out in various ways. First, “formal translations” are translations into a
formal language; but which formal language? It will be a language that has a
logical constant for each English logical expression. But that raises the question
of which expressions of English are logical expressions. In addition to ‘and’,
‘or’, ‘all’, and so on, are any of the following logical expressions?
necessarily
it will be the case that
most
it is morally wrong that
Further, the notion of translation must be defined; further, an appropriate

definition of ‘model’ must be chosen.
Similar issues of refinement confront a second account, the proof-theoretic
account: φ is a logical consequence of the members of Γ iff the translation of
φ is provable from the translations of the members of Γ. We must decide what
formal language to translate into, and we must decide upon an appropriate
account of provability.
A third view is Quine’s: φ is a logical consequence of the members of Γ
iff there is no way to (uniformly) substitute new nonlogical expressions for
nonlogical expressions in φ and the members of Γ so that the members of Γ
become true and φ becomes false.
Three other accounts should be mentioned. The first account is a modal
one. Say that Γ modally implies φ iff it is not possible for φ to be false while the
members of Γ are true. (What does ‘possible’ mean here? There are many kinds
of possibility one might have in mind: so-called “metaphysical possibility”,
“absolute possibility”, “idealized epistemic possibility”…. Clearly the accept-
ability of the proposal depends on the legitimacy of these notions. We discuss
modality later in the book, beginning in chapter 6.) One might then propose
that φ is a logical consequence of the members of Γ iff Γ modally implies φ. (An
intermediate proposal: φ is a logical consequence of the members of Γ iff, in

virtue of the forms of φ and the members of Γ, Γ modally implies φ. More carefully:
φ is a logical consequence of the members of Γ iff Γ modally implies φ, and
moreover, whenever Γ0 and φ0 result from Γ and φ by (uniform) substitution of
nonlogical expressions, Γ0 modally implies φ0 . This is like Quine’s definition,
but with modal implication in place of truth-preservation.) Second, there is
a primitivist account, according to which logical consequence is a primitive
notion. Third, there is a pluralist account according to which there is no one
kind of genuine logical consequence. There are, of course, the various con-
cepts proposed by each account, each of which is trying to capture genuine
logical consequence; but in fact there is no further notion of genuine logical
consequence at all; there are only the proposed construals.
As I say, this is not a book on philosophical logic, and so we will not inquire
further into which (if any) of these accounts is correct. We will, rather, focus
exclusively on two kinds of formal proposals for modeling logical consequence
and logical truth: model-theoretic and proof-theoretic proposals.
1.6 Extensions, deviations, variations4

“Standard logic” is what is usually studied in introductory logic courses. It
includes propositional logic (logical constants: ∧, ∨, ∼, →, ↔), and predicate
logic (logical constants: ∀, ∃, variables). In this book we’ll consider various
modifications of standard logic:
1.6.1 Extensions
Here we add to standard logic. We add both:
· new symbols
· new cases of logical consequence and logical truth that we can model
We do this in order to get a better representation of logical consequence. There

is more to logic than that captured by plain old standard logic.
We extended propositional logic, after all, to get predicate logic. You can
do a lot with propositional logic, but you can’t capture the obvious fact that
4
See Gamut (1991a, pp. 156-158).
‘Ted is happy’ logically implies ‘someone is happy’ using propositional logic

alone. It was for this reason that we added quantifiers, variables, predicates,
etc., to propositional logic (added symbols), and added means to deal with
these new symbols in semantics and proof theory (new cases of logical conse-
quence and logical truth). But there is no need to stop with plain old predicate
logic. We will consider adding symbols for identity, function symbols, and
definite descriptions to predicate logic, for example, and we’ll add a sign for
“necessarily” when we get to modal logic. And in each case, we’ll introduce
modifications to our formal theories that let us account for logical truths and
logical consequences involving the new symbols.
1.6.2 Deviations
Here we change, rather than add. We retain the same symbols from standard
logic, but we alter standard logic’s proof theory and semantics. We therefore
change what we say about the logical consequences and logical truths that
involve the symbols.
Why do this? Perhaps because we think that standard logicians are wrong
about what the right logic for English is. If we want to correctly model logical
consequence in English, therefore, we must construct systems that behave
differently from standard logic.
For example, in the standard semantics for propositional logic, every for-
mula is either true or false. But some have argued that natural language sen-
tences like the following are neither true nor false:
The king of the United States is bald

Sherlock Holmes weighs more than 178 pounds
Bill Clinton is tall There will be a sea battle tomorrow
If this is correct, then perhaps we should abandon the standard semantics for
propositional logic in favor of multi-valued logic, in which formulas are allowed
to be neither true nor false.
1.6.3 Variations
Here we also change standard logic, but we change the notation without
changing the content of logic. We study alternate ways of expressing the same
thing.
For example, in intro logic we show how:
∼(P ∧Q)
∼P ∨∼Q
are two different ways of saying the same thing. We will study other ways of
saying what those two sentences say, including:
P |Q
∼∧P Q
In the first case, | is a new symbol for “not both”. In the second case (“Polish
notation”), the ∼ and the ∧ mean what they mean in standard logic; but instead
of going between the P and the Q, the ∧ goes before P and Q. The value of
this, as we’ll see, is that we no longer will need parentheses.
1.7 Metalogic, metalanguages, and formalization

In introductory logic, we learned how to use certain logical systems. We learned
how to do truth tables, construct derivations, and so on. But logicians do not
spend much of their time developing systems only to sit around all day doing
derivations in those systems. As soon as a logician develops a new system, he
or she will begin to ask questions about that system. For an analogy, imagine
people who make up games. They might invent a new version of chess. Now,
they might spend some time actually playing the new game. But if they were
like logicians, they would soon get bored with this and start asking questions
about the game, such as: “is the average length of this new game longer than the
average length of a game of standard chess?”. “Is there any strategy one could
pursue which will guarantee a victory?” Analogously, logicians ask questions
like: what things can be proved in such and such a system? Can you prove the
same things in this system as in system X? Proving things about logical systems
is part of “meta-logic”, which is an important part of logic.
One particularly important question of metalogic is that of soundness and
completeness. Standard textbooks introduce a pair of methods for characterizing
logical truth for the formulas of propositional logic. One is semantic: a formula
is a semantic logical truth iff the truth table for that formula has all “trues” in its
final column. Another is proof-theoretic: a sentence is a proof-theoretic logical
truth iff there exists a derivation of it (from no premises), where a derivation
is then appropriately defined. (Think: introduction- and elimination- rules,

conditional and indirect proof, and so on.) The question of soundness and
completeness is: how do these two methods for characterizing logical truth
relate to each other? The question is answered, in the case of propositional
logic, by the following metalogical results, which are proved in standard books
on metalogic:
Soundness of propositional logic: In propositional logic, any proof-theoretic

logical truth is a semantic logical truth
Completeness of propositional logic: In propositional logic, any semantic

logical truth is a proof-theoretic logical truth
These are really interesting claims! They show that the method of truth tables
and the method of constructing derivations amount to the same thing, as
applied to symbolic formulas of propositional logic. One can establish similar
results for standard predicate logic.
A couple remarks about proving things in metalogic.
First: what do we mean by “proving”? We do not mean: constructing a
derivation in the logical system we’re investigating. We’re trying to construct a
proof about the system. We do this in English, and we do it with informal (though
rigorous!) reasoning of the sort one would encounter in a mathematics book.
Logicians often distinguish the “object language” from the “metalanguage”.
The object language is the language that’s being studied—the language of
propositional logic, for example. Sentences of this object language look like
this:
P ∧Q
∼(P ∨Q)↔R
The metalanguage is the language we use to talk about the object language.
In the case of the present book, the metalanguage is English. Here are some
example sentences of the metalanguage:
‘P ∧Q’ is a sentence with three symbols, one of which is

a logical constant
Every sentence of propositional logic has the same num-
ber of left parentheses as right parentheses
If there exists a derivation of a formula, then its truth

table contains all “trues” in its final column (i.e., sound-
ness)
Thus, we formulate metalogical claims in the metalanguage, and our proofs in

metalogic take place in the metalanguage.
Second: to get anywhere in metalogic, we will have to get picky about a few
things about which one can afford to be lax in introductory logic. Let’s look at
soundness, for instance. To be able to prove this, in a mathematically rigorous
way, we’ll need to have the terms in it defined very carefully. In particular, we’ll
need to say exactly what we mean by ‘sentence of propositional logic’, ‘truth
tables’, and ‘derived’. Defining these terms precisely (another thing we’ll do
using English, the metalanguage!) is known as formalizing logic. Our first task
will be to formalize propositional logic.
1.8 Set theory5

As mentioned above, modern logic uses mathematical techniques to study
formal languages. The mathematical techniques in question are those of “set
theory”. Only the most elementary set-theoretic concepts and assumptions will
be needed, and you are probably already familiar with them; but nevertheless,
here is a brief overview.
Sets have members. Consider, for example, the set, N, of natural numbers.
Each natural number is a member of N: 1 is a member of N, 2 is a member of N,
and so on. We use the expression “∈” for this relationship of membership; thus,
we can say: 1 ∈ N, 2 ∈ N, and so on. We often name a set by putting names of
its members between braces: “{1, 2, 3, 4, . . . }” is another name of N.
Sets are not limited to sets of mathematical entities; anything can be a
member of a set. Thus, we may speak of the set of people, the set of cities,
or—to draw nearer to our intended purpose—the set of sentences in a given
language.
There is also the empty set, ∅. This is the one set with no members. That
is, for each object u, u is not a member of ∅ (i.e.: for each u, u ∈ / ∅.)
Though the notion of a set is an intuitive one, it is deeply perplexing. This
can be seen by reflecting on the Russell Paradox, discovered by Bertrand Russell,
the great philosopher and mathematician. Let us call R the set of all and only
5
Supplementary reading: the beginning of Enderton (1977)
those sets that are not members of themselves. For short, R is the set of non-
self-members. Russell asks the following question: is R a member of itself?
There are two possibilities:
· R∈/ R. Thus, R is a non-self-member. But R was said to be the set of all

non-self-members, and so we’d have R ∈ R. Contradiction.
· R ∈ R. So R is not a non-self-member. R, by definition, contains only

non-self-members. So R ∈/ R. Contradiction.
Thus, each possibility leads to a contradiction. But there are no remaining

possibilities—either R is a member of itself or it isn’t! So it looks like the very
idea of sets is paradoxical.
The modern discipline of axiomatic set theory arose in part to develop a
notion of sets that isn’t subject to this sort of paradox. This is done by imposing
rigid restrictions on when a given “condition” picks out a set. In the example
above, the condition “is a non-self-member” will be ruled out—there’s no set
of all and only the things satisfying this condition. The details of set theory are
beyond the scope of this course; for our purposes, we’ll help ourselves to the
existence of sets, and not worry about exactly what sets are, or how the Russell
paradox is avoided.
Various other useful set-theoretic notions can be defined in terms of the
notion of membership. We say that A is a subset of B (“A ⊆ B”) when every
member of A is a member of B. We say that the intersection of A and B (“A∩B”)
is the set that contains all and only those things that are in both A and B, and
that the union of A and B (“A∪B”) is the set containing all and only those things
that are members of either A or B.
Suppose we want to refer to the set of the so-and-sos—that is, the set
containing all and only objects, u, that satisfy the condition “so-and-so”. We’ll
do this with the term “{u: u is a so-and-so}”. Thus, we could write: “N = {u :
u is a natural number}”. And we could restate the definitions of ∩ and ∪ from
the previous paragraph as follows:
A ∩ B = {u : u ∈ A and u ∈ B}
A ∪ B = {u : u ∈ A or u ∈ B}
Sets have members, but they don’t contain them in any particular order. For
example, the set containing me and Bill Clinton doesn’t have a “first” member.
This is reflected in the fact that “{Ted, Clinton}” and “{Clinton, Ted}” are
two different names for the same set—the set containing just Clinton and Ted.
But sometimes we need to talk about a set-like thing containing Clinton and
Ted, but in a certain order. For this purpose, logicians use ordered sets. Two-
membered ordered sets are called ordered pairs. To name the ordered pair of
Clinton and Ted, we use: “〈Clinton, Ted〉”. Here, the order is significant, for
〈Clinton, Ted〉 and 〈Ted, Clinton〉 are not the same thing. The three-membered
ordered set of u, v, and w (in that order) is written: 〈u, v, w〉; and similarly for
ordered sets of any finite size. A n-membered ordered set is called an n-tuple.
Let’s even allow 1-tuples: let’s define the 1-tuple 〈u〉 as being the object u itself.
In addition to sets, and ordered sets, we’ll need a further related concept:
that of a function. A function is a rule that “takes in” an object or objects,
and “spits out” a further object. For example, the addition function is a rule
that takes in two numbers, and spits out their sum. As with sets and ordered
sets, functions are not limited to mathematical entities: they can “take in” and
“spit out” any objects whatsoever. We can speak of the father-of function, for
example, which is a rule that takes in a person, and spits out the father of that
person. And later in this book we will be considering functions that take in
and spit out linguistic entities: sentences and parts of sentences from formal
languages.
Each function has a fixed number of “places”: a fixed number of objects
it must take in before it is ready to spit out something. You need to give
the addition function two arguments (numbers) in order to get it to spit out
something, so it is called a two-place function. You only need to give the father-
of function one object, on the other hand, to get it to spit out something, so it
is a one-place function.
The objects that the function takes in are called its arguments, and the object
it spits out is called its value. Suppose f is an n-place function, and u1 . . . un are
n of its arguments; one then writes “ f (u1 . . . un )” for the value of function f as
applied to arguments u1 . . . un . f (u1 . . . un ) is the object that f spits out, if you
feed it u1 . . . un . For example, where f is the father-of function, since Ron is
my father, we can write: f (Ted) = Ron; and, where a is the addition function,
we can write: a(2, 3) = 5.
There’s a trick for “reducing” talk of both ordered pairs and functions to
talk of sets. One first defines 〈u, v〉 as the set {{u}, {u, v}}; one defines 〈u, v, w〉
as 〈u, 〈v, w〉〉, and similarly for n-membered ordered sets, for each positive
integer n. And, finally, one defines an n-place function as a set, f , of n + 1-
tuples obeying the constraint that if 〈u1 , . . . , un , v〉 and 〈u1 , . . . , un , w〉 are both
members of f , then v = w; f (u1 , . . . , un ) is then defined as the object, v, such

that 〈u1 , . . . , un , v〉 ∈ f . Thus, ordered sets and functions are defined as certain
sorts of sets. The trick of the definition of ordered pairs is that we put the
set together in such a way that we can look at the set and tell what the first
member of the ordered pair is: it’s the one that “appears twice”. Similarly, the
trick of the definition of a function is that we can take any arguments to the
function, look at the set that is identified with the function, and figure out
what value the function spits out for those arguments. But the technicalities of
these reductions won’t matter for us; I’ll just feel free to speak of ordered pairs,
triples, functions, etc., without defining them as sets.
Chapter 2
Propositional Logic
e begin with the simplest logic commonly studied: propositional logic.

W Despite its simplicity, it has great power and beauty.
2.1 Grammar of PL
Modern logic has made great strides by treating the language of logic as a
mathematical object. To do so, grammar needs to be developed rigorously.
(Our study of a new logical system will always begin with grammar.)
If all you want to do is understand the language of logic informally, and be
able to use it effectively, you don’t really need to get so careful about grammar.
For even if you haven’t ever seen the grammar of propositional logic formalized,
you can recognize that things like this make sense:
P →Q
R ∧ (∼S↔P )
Whereas things like this do not:
→P QR∼
(P ∼Q∼(∨
But to make any headway in metalogic, we will need more than an intuitive
understanding of what makes sense and what does not; we will need a precise
definition that has the consequence that only the strings of symbols in the first
group “make sense”.
18
CHAPTER 2. PROPOSITIONAL LOGIC 19
Grammatical formulas (i.e., ones that “make sense”) are called well-formed
formulas, or “wffs” for short. We define these by first carefully defining exactly
which symbols are allowed to occur in wffs (the “primitive vocabulary”), and
second, carefully defining exactly which strings of these symbols count as wffs:
Primitive vocabulary:
· Sentence letters: P, Q, R . . . , with or without numerical subscripts

· Connectives: →, ∼
· Parentheses: ( , )
Definition of wff:
i) Sentence letters are wffs

ii) If φ and ψ are wffs then φ→ψ and ∼φ are also wffs
iii) Only strings than can be shown to be wffs using i) and ii) are wffs
We will be discussing a number of different logical systems throughout this

book, with differing notions of grammar. What we have defined here is the
notion of a wff for one particular language, the language of PL. So strictly, we
should speak of PL-wffs. But I’ll just say “wff” when where there is no danger
of ambiguity.
Notice an interesting feature of this definition: the very expression we
are trying to define, ‘wff’, appears on the right hand side of clause ii) of the
definition. In a sense, we are using the expression ‘wff’ in its own definition.
Is that “circular”? Not in any objectionable way. This definition is what is
called a “recursive” definition, and recursive definitions are legitimate despite
this sort of circularity. The reason is that clause ii) defines the notion of a
wff for certain complex expressions (namely, ∼φ and φ→ψ) in terms of the
notion of a wff as applied to smaller expressions (φ and ψ). These smaller
expressions may themselves be complex, and therefore may have their statuses
as wffs determined, via clause ii), in terms of yet smaller expressions, and so on.
But eventually this procedure will lead us to clause i), not clause ii). And clause
i) is not circular: in that clause, we do not appeal to the notion of a wff in its
own definition; we rather say directly that sentence letters are wffs. Recursive
definitions always “bottom out” in this way; they always include a clause (called
the “base” clause) like i).
Think of this procedure in reverse: we begin with the smallest wffs (sentence
letters), and build up complex wffs using clause ii). Example: we can use clauses
i) and ii) to show that the expression (∼P →(P →Q)) is a wff:
· P is a wff (clause i))
· so, ∼P is a wff (clause ii))
· Q is a wff (clause i))
· so, since P and Q are both wffs, (P →Q) is also a wff (clause ii))
· so, since ∼P and (P →Q) are both wffs, (∼P →(P →Q)) is also a wff (clause
ii))
What’s the point of clause iii)? Clauses i) and ii) provide only sufficient
conditions for being a wff, and therefore do not on their own exclude non-
sense combinations of primitive vocabulary like P ∼Q∼R, or even strings like
(P ∨147)→⊕ that include disallowed symbols. Clause iii) rules these strings out,
since there is no way to build up either of these strings from clauses i) and ii),
in the way that we built up the wff (∼P →(P →Q)).
What happened to ∧, ∨, and ↔? Our definition of a wff mentions only
→ and ∼; it therefore counts expressions like P ∧Q, P ∨Q, and P ↔Q as not
being wffs. Answer: we can define the ∧, ∨, and ↔ in terms of ∼ and →:
Definitions of ∧, ∨, and ↔:
· “φ∧ψ” is short for “∼(φ→∼ψ)”

· “φ∨ψ” is short for “∼φ→ψ”
· “φ↔ψ” is short for “(φ→ψ)∧(ψ→φ)”
So, whenever we subsequently write down an expression that includes one of

the defined connectives, we can regard it as being short for an expression that
includes only the official primitive connectives, ∼ and →. (We will show below
that the above definitions are good ones; in short, they are good because they
generate the correct truth tables for ∧, ∨, and ↔.)
Our choice to begin with → and ∼ as our primitive connectives was arbitrary.
We could have started with ∼ and ∧, and defined the others as follows:
· “φ∨ψ” is short for “∼(∼φ∧∼ψ)”

· “φ→ψ” is short for “∼(φ∧∼ψ)”
· “φ↔ψ” is short for “(φ→ψ)∧(ψ→φ)”
And other alternate choices are possible. We’ll talk about this later.
So: → and ∼ are our primitive connectives; the others are defined. Why
do we choose only a small number of primitive connectives? Because, as we
will see, it makes meta-proofs easier.
2.2 The semantic approach to logic

In the next section we will introduce a “semantics” for propositional logic. A
semantics for a language is a way of assigning meanings to words and sentences
of that language. For us, the central notion of meaning will be that of truth.
Roughly speaking, our approach will be to define, for each wff of propositional
logic, the circumstances in which it is true.
Philosophers disagree over how to understand the notion of meaning in
general. But the idea that the meaning of a sentence has something to do with
truth-conditions is hard to deny, and at any rate has currency within logic. On
this approach, one explains the meaning of a sentence by showing how that
sentence depends for its truth or falsity on the way the world is.
We will provide a truth-conditional semantics for a symbolic (formal) lan-
guage in two stages. First, we will define mathematical models of the various
configurations the world could be in. Second, we will define the conditions
under which a sentence of the symbolic language is true in one of these mathe-
matical configurations.
These definitions will have two main benefits. First, they will illuminate
meaning. In logic, the symbols of symbolic languages are typically intended to
represent bits of natural language. The PL connectives, for example, represent
‘and’, ‘or’, and so on. If the definitions are well-constructed, then the ways
in which the configurations render symbolic sentences true and false will be
parallel to the ways in which the real world renders corresponding natural
language sentences true and false. The definitions will therefore shed light on
the meanings of the natural language sentences represented by our symbolic
language. Second, our definitions will allow us to construct a precise theory (of
the semantic/model-theoretic variety) of logical consequence and logical truth.
The semantic conception is a way of making precise the idea that a logical truth
is a sentence that is “true no matter what”, and the idea that one sentence is a
logical consequence of some other sentences iff there is “no way” for the latter
sentences to be true without the former sentence being true. We will use our
definitions to make these rough statements more precise: we will say that one
formula is a logical consequence of others iff there is no configuration in which
the latter formulas are true but the former is not, and that a formula is a logical
truth iff it is true in all configurations.
2.3 Semantics of PL
Our semantics for propositional logic is really just truth tables, only presented
a little more carefully than in introductory logic books. What a truth table of a
formula does is depict how the truth value of that formula is determined by the
truth values of its sentence letters, for each possible combination of truth values
for its sentence letters. To do this nonpictorially, we need to define a notion
corresponding to “a possible combination of truth values for sentence letters”.
Definition of interpretation: A PL-interpretation is is a function I , that

assigns either 1 or 0 to every sentence letter
As with the notion of a wff, we will have different definitions of interpretations

for different logical systems, so strictly we must speak of PL-interpretations.
But usually it will be fine to speak simply of interpretations when it’s clear
which system is at issue.
The numbers 0 and 1 are our truth values. So an interpretation assigns
truth values to sentence letters. Instead of saying “let P be false, and Q be
true”, we can say: let I be an interpretation such that I (P ) = 0 and I (Q) = 1.
Once we settle what truth values a given interpretation assigns to the sen-
tence letters, the truth values of complex sentences containing those sentence
letters are thereby fixed. The usual, informal, method for showing exactly how
those truth values are fixed is by giving truth tables for each connective. The
standard truth tables for the → and ∼ are the following:1
→ 1 0 ∼
1 1 0 1 0
0 1 1 0 1
What we will do, instead, is write out a formal definition of a function—the

valuation function—that assigns truth values to complex sentences as a function
of the truth values of their sentence letters—i.e., as a function of a given
intepretation I . But the idea is the same as the truth tables: truth tables are
really just pictures of the definition of a valuation function:
Definition of valuation: For any PL-interpretation, I , the PL-valuation

for I , VI , is defined as the function that assigns to each wff either 1 or 0, and
which is such that, for any sentence letter α and any wffs φ and ψ,
VI (α) = I (α)
VI (φ→ψ) = 1 iff either VI (φ) = 0 or VI (ψ) = 1
VI (∼φ) = 1 iff VI (φ) = 0
We have another recursive definition: the valuation function’s values for com-
plex formulas are determined by its values for smaller formulas; and this pro-
cedure bottoms out in the values for sentence letters, which are determined
directly by the interpretation function I .
Notice also that in the definition of a valuation function I use the English
logical connectives ‘either…or’, and ‘iff ’. I used these English connectives
rather than the logical connectives ∨ and ↔, because at that point I was not
writing down wffs of the language of study (in this case, the language of propo-
sitional logic). I was rather using sentences of English—our metalanguage, the
informal language we’re using to discuss the formal language of propositional
logic—to construct my definition of the valuation function. My definition
needed to employ the logical notions of disjunction and bi-implication, the
English words for which are ‘either…or’ and ‘iff’.
1
The → table, for example, shows what truth value φ→ψ takes on depending on the truth
values of its parts. Rows correspond to truth values for φ, columns to truth values for ψ. Thus,
to ascertain the truth value of φ→ψ when φ is 1 and ψ is 0, we look in the 1 row and the 0
column. The listed value there is 0—the conditional is false in this case. The ∼ table lacks
multiple columns because ∼ is a one-place connective.
One might again worry that something circular is going on. We defined the
symbols for disjunction and bi-implication, ∨ and ↔, in terms of ∼ and → in
section 2.1, and now we’ve defined the valuation function in terms of disjunction
and bi-implication. So haven’t we given a circular definition of disjunction
and bi-implication? No. When we define the valuation function, we’re not
trying to define logical concepts such as negation, conjunction, disjunction,
implication, and bi-implication, and so on, at all. A reductive definition of these
very basic concepts is probably impossible (though one can define some of them
in terms of the others). What we are doing is starting with the assumption that
we already understand the logical concepts, and then using those notions to
provide a formalized semantics for a logical language. This can be put in terms
of object- and meta-language: we use metalanguage connectives, such as ‘iff’
and ‘or’, which we simply take ourselves to understand, to provide a semantics
for the object language connectives ∼, →, etc.
Back to the definition of the valuation function. The definition applies
only to official wffs, which can contain only the primitive connectives → and
∼. But sentences containing ∧, ∨, and ↔ are abbreviations for official wffs,
and therefore they too are governed by the definition. In fact, given the
abbreviations defined in section 2.1, one can show that the definition assigns
the intuitively correct truth values to sentences containing ∧, ∨, and ↔; one
can show that for any PL-interpretation I , and any wffs ψ and χ ,
VI (ψ∧χ ) = 1 iff VI (ψ) = 1 and VI (χ ) = 1

VI (ψ∨χ ) = 1 iff either VI (ψ) = 1 or VI (χ ) = 1
VI (ψ↔χ ) = 1 iff VI (ψ) = VI (χ )
I’ll show that the first statement is true here; the others are exercises for the
reader. I’ll write out my proof in excessive detail, to make it clear exactly how
the reasoning works.
Proof that ∧ gets the right truth condition. Let ψ and χ be any wffs. The expres-
sion ψ∧χ is an abbreviation for the expression ∼(ψ→∼χ ). So we want to
show that, for any PL-interpretation I , VI (∼(ψ→∼χ )) = 1 iff VI (ψ) = 1 and
VI (χ ) = 1. Now, in order to show that a statement α holds iff a statement
β holds, we must first show that if α holds, then β holds (the “forwards ⇒
direction”); then we must show that if β holds then α holds (the “backwards
⇐direction”):
⇒: First assume that VI (∼(ψ→∼χ )) = 1. Then, by definition of the val-

uation function, clause for ∼, VI (ψ→∼χ ) = 0. So2 , VI (ψ→∼χ ) is not 1. But
then, by the clause in the definition of VI for the →, we know that it’s not the
case that: either VI (ψ) = 0 or VI (∼χ ) = 1. That is: VI (ψ) = 1 and VI (∼χ ) = 0.
From the latter, by the clause for ∼, we know that VI (χ ) = 1. That’s what we
wanted to show—that VI (ψ) = 1 and VI (χ ) = 1.
⇐: This is sort of like undoing the previous half. Suppose that VI (ψ) = 1
and VI (χ ) = 1. Since VI (χ ) = 1, by the clause for ∼, VI (∼χ ) = 0; but now since
VI (ψ) = 1 and VI (∼χ ) = 0, by the clause for → we know that VI (ψ→∼χ ) = 0;
then by the clause for ∼, we know that VI (∼(ψ→∼χ )) = 1, which is what we
were trying to show.
(The symbol marks the end of meta-language proofs—that is, arguments
I give, phrased in English, to establish facts about various formal languages
discussed in this book.)
Exercise 2.1 Given the definitions of the defined symbols ∨ and

↔, show that for the valuation function V of any PL-interpretation,
and any wffs ψ and χ ,
V(ψ∨χ ) = 1 iff either V(ψ) = 1 or V (χ ) = 1

V(ψ↔χ ) = 1 iff V (ψ) = V (χ )
Let’s reflect on what we’ve done so far. We have defined the notion of
a PL-interpretation, which assigns 1s and 0s to sentence letters of the sym-
bolic language of propositional logic. And we have also defined, for any PL-
interpretation, a corresponding PL-valuation function, which extends the
interpretation’s assignment of 1s and 0s to complex wffs of PL. Note that we
have been informally speaking of these assignments as assignments of truth
values. That’s because the assignments of 1s and 0s accurately models the truth
values of statements in English that are represented in the obvious way by
PL-wffs. For example, the ∼ of propositional logic is supposed to model the
English phrase ‘it is not the case that’. Accordingly, just as an English sentence
2
The careful reader will note that here (and henceforth), I treat “VI (α) = 0” and “VI (α) is
not 1” interchangeably (for any wff α). (Similarly for “VI (α) = 1” and “VI (α) is not 0”.) This is
justified as follows. First, if VI (α) is 0, then it can’t also be that VI (α) is 1—VI was stipulated
to be a function. Second, since it was stipulated that VI assigns either 0 or 1 to each wff, if
VI (α) is not 1, then VI (α) must be 0.
“It is not the case that φ” is true iff φ is false, one of our valuation functions
assigns 1 to ∼φ iff it assigns 0 to φ.
Semantics in logic, recall, generally defines two things: configurations and
truth-in-a-configuration. In the propositional logic semantics we have laid
out, the configurations are the interpretation functions, and the valuation
function defines truth-in-a-configuration. Each interpretation function gives
a complete assignment of truth values to the sentence letters. Thus, insofar
as the sentence letters are concerned, an interpretation function completely
specifies a possible configuration of the world. And for any interpretation
function, its corresponding valuation function specifies, for each complex wff,
what truth value that wff has in that interpretation. Thus, for each wff (φ) and
each configuration (I ), we have specified the truth value of that wff in that
configuration (VI (φ)).
Onward. We are now in a position to define the semantic versions of the
notions of logical truth and logical consequence for PL. The semantic notion
of a logical truth is that of a valid formula:
Definition of validity: A formula φ is PL-valid iff for every PL-interpretation,

I , VI (φ) = 1
We write “PL φ” for “φ is PL-valid”. (When it’s obvious which system

we’re talking about, we’ll omit the subscript on .) The valid formulas of
propositional logic are also called tautologies.
As for logical consequence, the semantic version of this notion is that of a
single formula’s being a semantic consequence of a set of formulas:
Definition of semantic consequence: Formula φ is a PL-semantic conse-
quence of the formulas in set Γ iff for every PL-interpretation, I , if VI (γ ) = 1
for each γ in Γ, then VI (φ) = 1
That is, φ is a PL-semantic consequence of Γ iff φ is true whenever each
member of Γ is true. We write “Γ PL φ” for “φ is a PL-semantic consequence
of Γ”. (As usual we’ll often omit the “PL” subscript; and further, let’s improve
readability by writing “φ1 , . . . , φn ψ” instead of “{φ1 , . . . , φn } ψ”. That is,
let’s drop the set braces when it’s convenient to do so.)
A parenthetical remark: now we can see the importance for setting up the
grammar for our system according to precise rules. If we hadn’t, the definition
of ‘truth value’ given here would have been impossible. In this definition we
defined truth values of complicated formulas based on their form. For example,
if a formula has the form (φ→ψ), then we assigned it an appropriate truth value
based on the truth values of φ and ψ. But suppose we had a formula in our
language that looked as follows:
P →P →P
and suppose that P has truth value 0. What is the truth value of the whole? We
can’t tell, because of the missing parentheses. For if the parentheses look like
this:
(P →P )→P
then the truth value is 0, whereas if the parentheses look like this:
P →(P →P )
then it is 1. Certain kinds of grammatical ambiguity, then, make it impossible to

assign truth values. We solve this problem in logic by pronouncing the original
string “P →P →P ” as ill-formed; it is missing parentheses. Thus, the precise
rules of grammar assure us that when it comes time to do semantics, we are
able to assign semantic values (in this case, truth values) in an unambiguous
way.
Notice also a fact about validity in propositional logic: it is mechanically
“decidable”—a computer program could be written that is capable of telling,
for any given formula, whether or not that formula is valid. The program
would simply construct a complete truth table for the formula in question.
We can observe that this is possible by noting the following: every formula
contains a finite number N of sentence letters, and so for any formula, there
are only a finite number of different “cases” one needs to check—namely, the
2N permutations of truth values for the contained sentence letters. But given
any assignment of truth values to the sentence letters of a formula, it’s clearly a
perfectly mechanical procedure to compute the truth value the formula takes
for those truth values—simply apply the rules for the ∼ and → repeatedly.
It’s worth being very clear about two assumptions in this proof (which, by
the way, is our first bit of metatheory—our first proof about a logical system).
They are: that every formula has a finite number of sentence letters, and that
the truth values of sentence letters not contained in a formula do not affect
the truth value of the formula. We need the latter assumption to be sure that
we only need to check a finite number of cases—namely, the permutations
of truth values of the contained sentence letters—to see whether a formula is

valid. After all, there are infinitely many interpretation functions (since there
are infinitely many sentence letters in the language of PL), and a valid formula
must be true in each one.
These two assumptions are obviously true, but it would be good to prove
them. I’ll prove the first assumption here (the second may be proved by a
similar method), and take this opportunity to introduce an important technique
for metalanguage proofs: proof by induction.
Proof that every wff contains a finite number of sentence letters. In this sort of proof
by induction, we’re trying to prove a statement of the form: every wff has prop-
erty P . The property P in this case is having a finite number of different sentence
letters. In order to do this, we must show two separate statements:
base case: we show that every atomic sentence has the property. This is
obvious—atomic sentences are just sentence letters, and each of them contains
one sentence letter, and thus finitely many different sentence letters.
induction step: we begin by assuming that if formulas φ and ψ have the
property, then so will the complex formulas one can form from φ and ψ by the
rules of formation, namely ∼φ and φ→ψ. So, we assume that φ and ψ have
finitely many different sentence letters; and we show that the same must hold
for ∼φ and φ→ψ. That’s obvious: ∼φ has as many different sentence letters
as does φ; since φ, by assumption, has only finitely many, then so does ∼φ.
As for φ→ψ, by hypothesis, φ and ψ have finitely many different sentence
letters, and so φ→ψ has, at most, n + m sentence letters, where n and m are
the number of different sentence letters in φ and ψ, respectively.
We’ve shown that every atomic formula has the property having a finite
number of different sentence letters; and we’ve shown that the property is inherited
by complex formulas built according to the recursion rules. But every wff is
either atomic, or built from atomics by a finite series of applications of the
recursion rules. Therefore, by induction, every wff has the property.
2.4 Natural deduction in propositional logic

We have investigated a semantic conception of the notions of logical truth and
logical consequence. An alternate conception is proof-theoretic, in which the
central conception is that of proof. On this conception, logical consequence
means “provable from”, and a logical truth is a sentence that can be proved
starting from no premises at all. A “proof” procedure, informally, is a method

of reasoning one’s way, step by step, according to mechanical rules, from some
premises to a conclusion. This all is, of course, informal; we must now make it
precise.
One method for characterizing proof is called the method of natural de-
duction. Any system in which one has assumptions for “conditional proof”,
assumptions for “indirect derivation”, etc. is a system of natural deduction.
This is the usual method in introductory logic books. Proofs in these systems
often look like this:
1 P →(Q→R)
2 P ∧Q
3 P 2, ∧E
4 Q 2, ∧E
5 Q→R 1, 3 →E
6 R 4, 5 →E
7 (P ∧Q)→R 2-6, →I
or like this:
1. P →(Q→R) Pr.
2. show (P ∧Q)→R CD
3. P ∧Q As.
4. show R DD
5. P 3, ∧E
6. Q 3, ∧E
7. Q→R 1, 5 →E
8. R 6, 7→E
We will implement natural deduction a little differently here, in order to reveal

what is really going on. Our derivations will therefore look a little different
from the derivations familiar from introductory books. Our version of the
above derivation will look like this:
1. P →(Q→R) ` P →(Q→R) RA
2. P ∧Q ` P ∧Q RA
3. P ∧Q ` P 2, ∧E
4. P ∧Q ` Q 2, ∧E
5. P →(Q→R), P ∧Q ` Q→R 1,3 →E
6. P →(Q→R), P ∧Q ` R 4,5 →E
7. P →(Q→R) ` (P ∧Q)→R 5, →I
It looks different, but the underlying idea is nevertheless the same.
2.4.1 Sequents
Natural deduction systems model the kind of reasoning one employs in everyday
life. How does that reasoning work? In its simplest form, one reasons in a
step-by-step fashion from premises to a conclusion, each step being sanctioned
by a rule of inference. For example, suppose that one already knows the premise
that P ∧(P →Q) is true. One can then reason one’s way to the conclusion that
Q is also true, as follows:
1. P ∧(P →Q) premise

2. P from line 1
3. P →Q from line 1
4. Q from lines 2 and 3
In this kind of proof, each step is a tiny, indisputably correct, logical inference.
Consider the moves from 1 to 2 and from 1 to 3, for example. These are
indisputably correct because a conjunctive statement clearly logically implies
either of its conjuncts. Likewise for the move from 2 and 3 to 4: it is clear
that a conditional statement, plus its antecedent, together imply its consequent.
Natural deduction systems consist in part of simple general principles like these
(“a conjunctive statement logically implies either of its conjuncts”); they are
known as rules of inference.
In addition to rules of inference, ordinary reasoning employs a further
technique: the use of assumptions. In order to establish a conditional claim “if P
then Q”, one would ordinarily i) assume P , ii) reason one’s way to Q, and then
iii) on that basis conclude that the conditional claim “if P then Q” is true. Once
P ’s assumption is shown to lead to Q, the conditional claim “if P then Q” may
be concluded. Another example: to establish a claim of the form “not-P ”, one
would ordinarily i) assume P , ii) reason one’s way to a contradiction, and iii)
on that basis conclude that “not-P ” is true. Once P ’s assumption is shown to
lead to a contradiction, “not-P ” may be concluded. The first sort of reasoning
is called conditional proof, the second, reductio ad absurdum.
When one reasons with assumptions, one writes down statements that one
does not know to be true. When you write down P as an assumption, with the
goal of proving the conditional “if P then Q”, you do not know P to be true.
You’re merely assuming P for the sake of establishing the conditional “if P
then Q”. Outside the context of this proof, the assumption need not hold; once
you’ve reasoned your way to Q on the basis of the assumption of P , and so
concluded that the conditional “if P then Q” is true, you stop assuming P . To
model this sort of reasoning formally, we need a way to keep track of how the
conclusions we establish depend on the assumptions we have made. Natural
deduction systems in introductory textbooks tend to do this geometrically (by
placement on the page), with special markers (e.g., ‘show’), and by drawing
lines or boxes around parts of the proof once the assumptions that led to those
parts are no longer operative. We will do it differently: we will keep track of the
dependence of conclusions on assumptions by writing down explicitly, for each
conclusion, which assumptions it depends on. We will do this by constructing
our derivations out of sequents.
A sequent looks like this:
Γ`φ
Γ is a set of formulas, called the premises of the sequent. φ is a single formula,

called the conclusion of the sequent. “`” is a sign that goes between the sequent’s
premises and its conclusion, to indicate that the whole thing is a sequent. We
will introduce sequent proofs below; and when you write down the sequent
Γ ` φ in one of them, the idea is that φ is an established conclusion, but it
was established by making the assumptions in Γ. Take away those assumptions,
and φ may no longer be established. In fact, one may think of a sequent as
“meaning” that its conclusion is a logical consequence of its premises.
Thus, our proofs will be proofs of sequents. It’s a bit weird at first to think in
terms of proving sequents, rather than formulas, since each sequent itself asserts
a relation of logical consequence between its premises and its conclusion, but
the idea nevertheless makes sense. Let’s introduce an informal notion of logical
correctness for sequents: the sequent Γ ` φ is logically correct if the formula φ
is a logical consequence of the formulas in Γ. Thus, one is entitled to conclude
the conclusion of a logically correct sequent from its premises. The idea, then,
of constructing a sequent proof of a sequent is to show that that sequent is
logically correct—to show, that is, that its consequent is a logical consequence
of its premises.
From our investigation of the semantics of propositional logic, we already
have the makings of a semantic criterion for when a sequent is logically correct:
the sequent Γ ` φ is logically correct iff φ is a semantic consequence of Γ.
What we will be doing in this section is giving a new, proof-theoretic, criterion
for the logical correctness of sequents.
2.4.2 Rules
The first step in developing our system is to write down sequent rules. A sequent
rule is a permission to move from certain sequents to certain other sequents. Our
goal is to construct rules with the following feature: if the “from” sequents are
all logically correct sequents, then any of the “to” sequents will be guaranteed
to be a logically correct sequent. Call such sequent rules “logical-correctness
preserving”.
Consider, as an example, the first rule of our system “∧ introduction”, or
“∧I” for short. We picture this sequent rule thus:
Γ`φ ∆`ψ
∧I
Γ, ∆ ` φ∧ψ
Above the line go the “from” sequents; below the line go the “to”-sequents.
(The comma between Γ and ∆ in the “to” sequent simply means that the
premises of this sequent are all the members of Γ plus all the members of ∆.
Strictly speaking it would be more correct to write this in set-theoretic notation
as: Γ ∪ ∆ ` φ∧ψ.) Thus, ∧I permits us to move from the sequents Γ ` φ and
∆ ` ψ to the sequent Γ, ∆ ` φ∧ψ. For any sequent rule, we say that any of the
“to” sequents (Γ, ∆ ` φ∧ψ in this case) follows from the “from” sequents (in this
case Γ ` φ and ∆ ` ψ) via the rule.
It seems intuitively clear that ∧I preserves logical correctness. For if some
assumptions Γ logically imply φ, and some assumptions ∆ logically imply ψ,
then (since φ∧ψ intuitively follows from φ and ψ taken together) the conclusion
φ∧ψ should indeed logically follow from all the assumptions together, the ones
in Γ and the ones in ∆.
Our next sequent rule is ∧E:
Γ ` φ∧ψ
∧E
Γ`φ Γ`ψ
This lets one move from the sequent Γ ` φ∧ψ to either the sequent Γ ` φ or
the sequent Γ ` ψ (or both). This, too, appears to preserve logical correctness.
If the members of Γ imply the conjunction φ∧ψ, then (since φ∧ψ intuitively
implies both φ and ψ individually) it must be that the members of Γ imply φ,
and they must also imply ψ.
The rule ∧I is known as an introduction rule for ∧, since it allows us to move
to a sequent whose major connective is the ∧. Likewise, the rule ∧E is known
as an elimination rule for ∧, since it allows us to move from a sequent whose
major connective is the ∧. In fact our sequent system contains introduction and
elimination rules for the other connectives as well: ∼, ∨, and → (let’s forget
the ↔ here.) We’ll present those rules in turn.
First ∨I and ∨E:
Γ`φ Γ ` φ∨ψ ∆1 , φ ` χ ∆2 , ψ ` χ
∨I ∨E
Γ ` φ∨ψ Γ ` ψ∨φ Γ, ∆1 , ∆2 ` χ
Let’s think about what ∨E means. Remember the intuitive meaning of a sequent:
its conclusion is a logical consequence of its premise. Another (related) way to
think of it is that Γ ` φ means that one can establish that φ if one assumes the
members of Γ. So, if the sequent Γ ` φ∨ψ is logically correct, that means we’ve
got the disjunction φ∨ψ, assuming the formulas in Γ. Now, suppose we can
reason to a new formula χ , assuming φ, plus perhaps some other assumptions
∆1 . And suppose we can also reason to χ from ψ, plus perhaps some other
assumptions ∆2 . Then, since either φ or ψ (plus the assumptions in ∆1 and ∆2 )
leads to χ , and we know that φ∨ψ is true (conditional on the assumptions in Γ),
we ought to be able to infer χ itself, assuming the assumptions we needed along
the way (∆1 and ∆2 ), plus the assumptions we needed to get φ∨ψ, namely, Γ.
Next, we have double negation:
Γ`φ Γ ` ∼∼φ
DN
Γ ` ∼∼φ Γ`φ
In connection with negation, we also have the rule of reductio ad absurdum:

Γ, φ ` ψ∧∼ψ
RAA
Γ ` ∼φ
That is, if φ (along with perhaps some other assumptions, Γ) leads to a contra-
diction, we can conclude that ∼φ is true (given the assumptions in Γ). RAA
and DN together are our introduction and elimination rules for ∼.
And finally we have →I and →E:
Γ, φ ` ψ Γ ` φ→ψ ∆ ` φ
→I →E
Γ ` φ→ψ Γ, ∆ ` ψ
→E is perfectly straightforward; it’s just the familiar rule of modus ponens.
But →I requires a bit more thought. →I is the principle of conditional proof.
Suppose you can get to ψ on the assumption that φ (plus perhaps some other
assumptions Γ.) Then, you should be able to conclude that the conditional
φ→ψ is true (assuming the formulas in Γ). Put another way: if you want to
establish the conditional φ→ψ, all you need to do is assume that φ is true, and
reason your way to ψ.
We add, finally, one more sequent rule, the rule of assumptions
RA
φ`φ
Note that this is the one sequent rule when no “from” sequent is required,
since there are no sequents above the line. The rule permits us to move from
no sequents at all to a sequent of the form φ ` φ. (Strictly, this sequent should
be written “{φ} ` φ”.) Call any such sequent an “assumption sequent”. Clearly,
any assumption sequent is a logically correct sequent, since clearly φ can be
proved if we assume φ itself.
2.4.3 Sequent proofs

We have assembled all the sequent rules; now we need to show how to use
those rules to provide a criterion for logically correct sequents. We do this by
first defining the notion of a “sequent proof”:
Definition of sequent proof: A sequent proof is a series of sequents, each of
which is either an assumption sequent, or follows from earlier sequents in the
series by some sequent rule.
So, for example, the following is a sequent proof
1. P ∧Q ` P ∧Q As
2. P ∧Q ` P 1, ∧E
3. P ∧Q ` Q 1, ∧E
4. P ∧Q ` Q∧P 2, 3 ∧I
Though it isn’t strictly required, we write a line number to the left of each
sequent in the series, and to the right of each line we write the sequent rule
that justifies it, together with the line or lines (if any) that contained the “from”
sequents required by the sequent rule in question. (The rule of assumptions
requires no “from” sequents, recall.)
(It’s important to distinguish what we’re now calling proofs, namely, sequent
proofs, from the kinds of informal arguments I gave in section 2.3, and will
give elsewhere in this book. Sequent proofs (and also the axiomatic proofs we
will introduce in section 2.5) are formalized object-language proofs. The sentences
in sequent proofs are sentences in the object language; they are wffs of PL.
Moreover, we gave a rigorous definition of what a sequent proof is. Moreover,
sequent proofs are restrictive in that only the system’s official rules may be
used. For contrast, consider the argument I gave in section 2.3 that any PL-
valuation assigns 1 to φ∧ψ iff it assigns 1 to φ and 1 to ψ. That argument was
an informal metalanguage proof. The sentences in the argument were sentences
of English, and the argument used informal (i.e., not formalized) techniques of
reasoning. “Informal” doesn’t imply lack of rigor. The argument was perfectly
rigorous: it conforms to the standards of good argumentation that generally
prevail in mathematics. We’re free to use any reasonable pattern of reasoning,
for example “universal proof” (to establish something of the form “everything
is thus-and-so”, we consider an arbitrary thing and show that it is thus-and-so).
We may “skip steps” if it’s clear how the argument is supposed to go. In short,
what we must do is convince a well-informed and mathematically sophisticated
reader that the result we’re after is indeed true.)
Next we introduce the notion of a “provable sequent”. The idea is that
each sequent proof culminates in the proof of some sequent. Thus we offer the
following definition:
Definition of provable sequent: A provable sequent is a sequent that is the

last line of some sequent proof
(Note that it would be equivalent to define a provable sequent as any line in

any sequent proof, because at any point in a sequent proof one may simply stop
adding lines, and the proof up until that point counts as a legal sequent proof.)
So, for example, the sequent proof given above establishes that P ∧Q ` Q∧P is
a provable sequent. (We call a sequent proof, whose last line is Γ ` φ, a sequent
proof of Γ ` φ.)
The definitions we have given in this section give us a way to make precise
the proof-theoretic conception of the core logical notions, as applied to propo-
sitional logic. Namely, we can say that φ is a logical truth, on this conception,
iff the sequent ∅ ` φ is a provable sequent, and that φ is a logical consequence
of the formulas in set Γ iff the sequent Γ ` φ is a provable sequent. The symbol
∅ stands for the “empty set”—the set containing no members. Thus, a logical
truth here is understood as a formula that is provable from no assumptions at
all.
2.4.4 Example sequent proofs

Let’s explore how to construct sequent proofs. You may find this a bit less
intuitive, initially, than constructing natural deduction proofs in the systems
familiar from introductory textbooks. But a little experimentation will show
that the techniques for proving things in the usual systems carry over to the
present system.
A first simple example: let’s return to the sequent proof of P ∧Q ` Q∧P :
1. P ∧Q ` P ∧Q As
2. P ∧Q ` P 1, ∧E
3. P ∧Q ` Q 1, ∧E
4. P ∧Q ` Q∧P 2, 3 ∧I
Notice the strategy. We first use the rule of assumptions to enter the premise
of the sequent we’re trying to prove: P ∧Q. We then use the rules of inference
to infer the consequent of that sequent: Q∧P . Since our initial assumption of
P ∧Q was dependent on the formula P ∧Q, our subsequent inferences remain
dependent on that same assumption, and so the final formula concluded, Q∧P ,
remains dependent on that assumption.
Let’s write our proofs out in a simpler way. Instead of writing out entire
sequents, let’s write out only their conclusions. We can indicate the premises
of the sequent using line numbers; the line numbers indicating the premises
of the sequent will go to the left of the number indicating the sequent itself.
Rewriting the previous proof in this way yields:
1 (1) P ∧Q As
1 (2) P 1, ∧E
1 (3) Q 1, ∧E
1 (4) Q∧P 2, 3 ∧I
Next, let’s have an example to illustrate conditional proof. Let’s construct a

sequent proof of P →Q, Q→R ` P →R:
1. P →Q ` P →Q As
2. Q→R ` Q→R As
3. P ` P As
4. P →Q, P ` Q 1,3 →E
5. P →Q, Q→R, P ` R 2,4 →E
6. P →Q, Q→R ` P →R 5,→I
It can be rewritten in the simpler style as follows:
1 (1) P →Q As
2 (2) Q→R As
3 (3) P As
1,3 (4) Q 1, 3 →E
1,2,3 (5) R 2, 4 →E
1,2 (6) P →R 5, →I
Let’s think about this example. We’re trying to establish P →R on the basis of
two formulas, P →Q and Q→R, so we start by assuming the latter two formulas.
Then, since the formula we’re trying to establish is a conditional, we assume
the antecedent of the conditional, in line 3. We then proceed, on that basis,
to reason our way to R, the consequent of the conditional we’re trying to
prove. (Notice how in lines 4 and 5, we add more line numbers on the very
left. Whenever we use → E, we increase dependencies: when we infer Q from
P and P →Q, our conclusion Q depends on all the formulas that P and P →Q
depended on, namely, the formulas on lines 1 and 3. Look back to the statement
of the rule → E: the conclusion ψ depends on all the formulas that φ and φ→ψ
depended on: Γ and ∆.) That brings us to line 5. At that point, we’ve shown
that R can be proven, on the basis of various assumptions, including P . The
rule →I (that is, the rule of conditional proof) then lets us conclude that the
conditional P →R follows merely on the basis of the other assumptions; that
rule, note, lets us in line 6 drop line 3 from the list of assumptions on which
P →R depends.
Next let’s establish an instance of DeMorgan’s Law, ∼(P ∨Q) ` ∼P ∧∼Q:
1 (1) ∼(P ∨Q) As
2 (2) P As (for reductio)
2 (3) P ∨Q 2, ∨I
1,2 (4) (P ∨Q)∧∼(P ∨Q) 1, 3 ∧I
1 (5) ∼P 4, RAA
6 (6) Q As (for reductio)
6 (7) P ∨Q 6, ∨I
1,6 (8) (P ∨Q)∧∼(P ∨Q) 1, 7 ∧I
1 (9) ∼Q 8, RAA
1 (10) ∼P ∧∼Q 5, 9∧I
Next let’s establish ∅ ` P ∨∼P :
1 (1) ∼(P ∨∼P ) As
2 (3) P ∨∼P 2, ∨I
2,1 (4) (P ∨∼P )∧∼(P ∨∼P ) 1, 3 ∧I
1 (5) ∼P 4, RAA
6 (6) ∼P As (for reductio)
6 (7) P ∨∼P 6, ∨I
6,1 (8) (P ∨∼P )∧∼(P ∨∼P ) 1, 7 ∧I
1 (9) ∼∼P 8, RAA
1 (10) ∼P ∧∼∼P 5, 9 ∧I
∅ (11) ∼∼(P ∨∼P ) 10, RAA
∅ (12) P ∨∼P 11, DN
Comment: my overall goal was to assume ∼(P ∨∼P ) and then derive a con-
tradiction. And my route to the contradiction was to separately establish ∼P
(lines 2-5) and ∼∼P (lines 6-9), each by reductio arguments.

Finally, let’s establish a sequent corresponding to a way that ∨ E is sometimes
formulated: P ∨Q, ∼P ` Q:
1 (1) P ∨Q As
2 (2) ∼P As
3 (3) Q As (for use with ∨E)
4 (4) P As (for use with ∨E)
5 (5) ∼Q As (for reductio)
4,5 (6) ∼Q∧P 4,5 ∧I
4,5 (7) P 6, ∧E
2,4,5 (8) P ∧∼P 2,7 ∧I
2,4 (9) ∼∼Q 8, RAA
2,4 (10) Q 9, DN
1,2 (11) Q 1,3,10 ∨E
The basic idea of this proof is to use ∨ E on line 1 to get Q. That calls, in
turn, for showing that each disjunct of line 1, P and Q, leads to Q. Showing
that Q leads to Q is easy; that was line 3. Showing that P leads to Q took lines
4-10; line 10 states the result of that reasoning, namely that Q follows from P
(as well as line 2). I began at line 4 by assuming P . Then my strategy was to
establish Q by reductio, so I assumed ∼Q in line 5. At this point, I basically
had my contradiction: at line 2 I had ∼P and at line 4 I had P . (You might
think I had another contradiction: Q at line 3 and ∼Q at line 5. But at the end
of the proof, I don’t want my conclusion to depend on line 3, whereas I don’t
mind it depending on line 2, since that’s one of the premises of the sequent
I’m trying to establish.) So I want to put P and ∼P together, to get P ∧∼P ,
and then conclude ∼∼Q by RAA. But there is a minor hitch. Look carefully at
how RAA is formulated. It says that if we have Γ,φ ` ψ∧∼ψ, we can conclude
Γ ` ∼φ. The first of these two sequents includes φ in its premises. That means
that in order to conclude ∼φ, the contradiction ψ∧∼ψ needs to depend on φ.
So in the present case, in order to finish the reductio argument and conclude
∼∼Q, the contradiction P ∧∼P needs to depend on the reductio assumption
∼Q (line 5.) But if I just used ∧I to put lines 2 and 4 together, the resulting
contradiction will only depend on lines 2 and 4. To get around this, I used
a little trick. Whenever you have a sequent Γ ` φ, you can always add any
formula ψ you like to the premises on which φ depends, using the following
method:3
i. Γ`φ (begin with this)
i + 1. ψ`ψ As (ψ is any chosen formula)
i + 2. Γ, ψ ` φ∧ψ ∧I
i + 3. Γ, ψ ` φ ∧E
Lines 4, 6 and 7 in the proof employ this trick: initially, at line 4, P depends
only on 4, but then by line 7, P also depends on 5. That way, the move from 8
to 9 by RAA is justified.
Exercise 2.2 Prove the following sequents:
a) P, Q, R ` P
b) P →(Q→R) ` (Q∧∼R)→∼P
c) P →Q, R→Q ` (P ∨R)→Q
2.5 Axiomatic proofs in propositional logic

Natural deduction proofs are comparatively easy to construct; that is their
great advantage. A different approach to proof theory, the axiomatic method,
offers different advantages. Like natural deduction, the axiomatic method is a
proof-theoretic approach to logic, based on the step-by-step reasoning model
in which each step is sanctioned by a rule of inference. But unlike natural
deduction systems, axiomatic systems do not allow reasoning by assumptions,
and they have very few rules of inference. Although these differences make
axiomatic proofs much harder to construct, there is a compensatory advantage
in metalogic: it is far easier to prove things about axiomatic systems.
Let’s first think about axiomatic systems informally. An axiomatic proof is
a series of formulas (not sequents—we no longer need them since we’re not
3
Adding arbitrary dependencies is not allowed in relevance logic, where a sequent is provable
only when all of its premises are, in an intuitive sense, relevant to its conclusion. Relevant
logicians modify various rules of classical logic, including the rule of ∧E.
reasoning with assumptions), the last of which is the conclusion of the proof.
Each line in the proof must be justified in one of two ways: it may be inferred
by a rule of inference from earlier lines in the proof, or it may be an axiom.
An axiom is a certain kind of formula, a formula that one is allowed to enter
into a proof without any justification at all. Axioms are the “starting points” of
proofs, the foundation on which proofs rest. Since axioms are to play this role,
the axioms in a good axiomatic system ought to be indisputable logical truths.
For example, “P →P ” would be a good axiom—it’s obviously a logical truth.
(As it happens, we won’t choose this particular axiom; we’ll instead choose
other axioms from which this one may be proved.) Similarly, for each rule of
inference in a good axiomatic system, there should be no question but that the
premises of the rule logically imply its conclusion.
Actually we’ll employ a slightly more general notion of a proof: a proof
from a given set of wffs Γ. A proof from Γ will be allowed to contain members
of Γ, in addition to axioms and wffs that follow from earlier lines by a rule.
Think of the members of Γ as premises, which in the context of a proof from
Γ are temporarily treated as axioms, in that they are allowed to be entered
into the proof without any justification. The intuitive point of a proof from
Γ is to demonstrate its conclusion on the assumption that the members of Γ are
true, in contrast to a proof simpliciter (i.e. a proof in the sense of the previous
paragraph), whose point is to demonstrate its conclusion unconditionally. (Note
that we can regard a proof simpliciter as a proof from the empty set ∅.)
Formally, to apply the axiomatic method, we must choose i) a set of rules,
and ii) a set of axioms. An axiom is simply any chosen sentence (though as we
saw, in a good axiomatic system the axioms will be clear logical truths.) A rule
is simply a permission to infer one sort of sentence from other sentences. For
example, the rule modus ponens can be stated thus: “From φ→ψ and φ you may
infer ψ”, and pictured as follows:
φ→ψ φ
MP
ψ
(Modus ponens is the analog of the sequent rule →E.) Given any chosen axioms
and rules, we can define the following concepts:
Definition of axiomatic proof from a set: Where Γ is a set of wffs and φ is
a wff, an axiomatic proof from Γ is a finite sequence of wffs whose last line is
φ, in which each line either i) is an axiom, ii) is a member of Γ, or iii) follows
from earlier wffs in the sequence via a rule.
Definition of axiomatic proof: An axiomatic proof of φ is an axiomatic proof

of φ from ∅ (i.e., a finite sequence of wffs whose last line is φ, in which each
line either i) is an axiom, or ii) follows from earlier wffs in the sequence via a
rule.)
It is common to write “Γ ` φ” to mean that φ is provable from Γ, i.e., that

there exists some axiomatic proof of φ from Γ, and to write “` φ” to mean that
∅ ` φ, i.e. that φ is provable, i.e., that there exists some axiomatic proof of φ
from no premises at all. (Formulas provable from no premises at all are often
called theorems.) This notation can be used for any axiomatic system, i.e. any
choice of axioms and rules. The symbol ` may be subscripted with the name
of the system in question. Thus, for our axiom system for PL below, we may
write: `PL . (We’ll omit this subscript when it’s clear which axiomatic system is
at issue.)
Here is an axiomatic system for propositional logic:4
Axiomatic system for PL:
· Rule: modus ponens

· Axioms: Where φ, ψ, and χ are wffs, anything that comes from the
following schemas are axioms
φ → (ψ→φ) (A1)
(φ→(ψ→χ )) → ((φ→ψ)→(φ→χ )) (A2)
(∼ψ→∼φ) → ((∼ψ→φ)→ψ) (A3)
Thus, a PL-theorem is any formula that is the last line of a sequence of formulas,
each of which is either an A1, A2, or A3 axiom, or follows from earlier formulas
in the sequence by modus ponens. And a formula is PL-provable from some
set Γ if it is the last line of a sequence of formulas, each of which is either a
member of Γ, an A1, A2, or A3 axiom, or follows from earlier formulas in the
sequence by modus ponens.
The axiom “schemas” A1-A3 are not themselves axioms. They are, rather,
“recipes” for constructing axioms. Take A1, for example:
φ→(ψ→φ)
4
See Mendelson (1987, p. 29).
This string of symbols isn’t itself an axiom because it isn’t a wff; it isn’t a wff
because it contains Greek letters, which aren’t allowed in wffs (since they’re
not on the list of PL primitive vocabulary). φ and ψ are variables of our
metalanguage; you only get an axiom when you replace these variables with
wffs. P →(Q→P ), for example, is an axiom; it results from A1 by replacing φ
with P and ψ with Q. (Note: since you can put in any wff for these variables,
and there are infinitely many wffs, there are infinitely many axioms.)
A few points of clarification about how to construct axioms from schemas.
First point: you can stick in the same wff for two different Greek letters. Thus
you can let both φ and ψ in A1 be P , and construct the axiom P →(P →P ).
(But of course, you don’t have to stick in the same thing for φ as for ψ.) Sec-
ond point: you can stick in complex formulas for the Greek letters. Thus,
(P →Q)→(∼(R→S)→(P →Q)) is an axiom (I put in P →Q for φ and ∼(R→S)
for ψ in A1). Third point: within a single axiom, you cannot substitute different
wffs for a single Greek letter. For example, P →(Q→R) is not an axiom; you
can’t let the first φ in A1 be P and the second φ be R. Final point: even though
you can’t substitute different wffs for a single Greek letter within a single axiom,
you can let a Greek letter become one wff when making one axiom, and let
it become a different wff when making another axiom; and you can use each
of these axioms within a single axiomatic proof. For example, each of the
following is an instance of A1, and one could include both in a single axiomatic
proof:
P →(Q→P )
∼P →((Q→R)→∼P )
In the first case, I made φ be P and ψ be Q; in the second case I made φ be
∼P and ψ be Q→R. This is fine because I kept φ and ψ constant within each
axiom.
The definitions we have given in this section constitute another way of
making precise the proof-theoretic conception of the core logical notions, as
applied to propositional logic. A logical truth, on this conception, is a PL-
theorem; one formula is a logical consequence of others iff it is PL-provable
from them.
2.5.1 Example axiomatic proofs

Axiomatic proofs are much harder than natural deduction proofs. Some are
easy, of course. Here is a proof of (P →Q)→(P →P ):
1. P →(Q→P ) (A1)
2. (P →(Q→P ))→((P →Q)→(P →P )) (A2)
3. (P →Q)→(P →P ) 1,2 MP
The existence of this proof shows that (P →Q)→(P →P ) is a theorem.
Building on the previous proof, we can construct a proof of P →P from
{P →Q}:
1. P →(Q→P ) (A1)
2. (P →(Q→P ))→((P →Q)→(P →P )) (A2)
3. (P →Q)→(P →P ) 1,2 MP
4. P →Q member of {(P →Q)}
5. P →P 3, 4 MP
Thus, we have shown that {P →Q} ` P →P .
(When we’re talking about provability from a set, let’s adopt the convention
of writing “φ1 . . . φn ` ψ” instead of “{φ1 . . . φn } ` ψ”, and writing “Γ, φ1 . . . φn ”
instead of “Γ ∪ {φ1 . . . φn }”. That is, let’s drop the set-braces on the left hand
side of ` in these circumstances. In this new notation, what we showed in the
previous paragraph was: P →Q ` P →P .)
The next example is a little harder: (R→P )→(R→(Q→P ))
1. [R→(P →(Q→P ))]→[(R→P )→(R→(Q→P ))] A2
2. P →(Q→P ) A1
3. [P →(Q→P )]→[R→(P →(Q→P ))] A1
4. R→(P →(Q→P )) 2,3 MP
5. (R→P )→(R→(Q→P )) 1,4 MP
Here’s how I approached this problem. What I was trying to prove, namely
(R→P )→(R→(Q→P )), is a conditional whose antecedent and consequent both
begin: (R→. That looks like the consequent of A2. So I wrote out an instance
of A2 whose consequent was the formula I was trying to prove; that gave me
line 1 of the proof. Then I tried to figure out a way to get the antecedent of
line 1; namely, R→(P →(Q→P )). And that turned out to be pretty easy. The
consequent of this formula, P →(Q→P ) is an axiom (line 2 of the proof). And
if you can get a formula φ, then you choose anything you like—say, R,—and
then get R→φ, by using A1 and modus ponens; that’s what I did in lines 3 and
4.
Exercise 2.3 Establish each of the following facts. For these prob-
lems, do not use the “toolkit” from the following sections; i.e.,
construct the axiomatic proofs “from scratch”. However, you may
use a fact you prove in an earlier problem in later problems.
a) ` P →P
b) ` (∼P →P )→P
c) ∼∼P ` P
In fact, we’ll regularly want to make a move like that of lines 3 and 4 from
the preceding proofs—whenever we have φ on its own, and we want to move
to ψ→φ. Let’s call this move “adding an antecedent”; this is how it is done:
1. φ (from earlier lines)

2. φ→(ψ→φ) A1
3. ψ→φ 1, 2 MP
In future proofs, instead of repeating such steps, let’s just move directly from φ
to ψ→φ, with the justification “adding an antecedent”.
The preceding proof was a bit tricky, and most proofs are trickier still.
Moreover, the proofs quickly get very long. Practically speaking, the best way
to make progress in an axiomatic system like this is by building up a toolkit.
The toolkit consists of theorems and techniques for doing bits of proofs which
are applicable in a wide range of situations. Then, when approaching a new
problem, one can look to see whether the problem can be reduced to a few
chunks, each of which can be accomplished by using the toolkit. Further, one
can cut down on writing by citing bits of the toolkit, rather than writing down
entire proofs.
So far, we have just one tool in our toolkit: “adding an antecedent”. Let’s
add another: the “MP technique”. Here’s what the technique will let us do.
Suppose we can separately prove φ→ψ and φ→(ψ→χ ). The MP technique
then shows us how to construct a proof of φ→χ . I call this the MP technique
because its effect is that you can do modus ponens “within the consequent of
the conditional φ→”. Here’s how the MP technique works:
1. φ→ψ from earlier lines

2. φ→(ψ→χ ) from earlier lines
3. (φ→(ψ→χ ))→((φ→ψ)→(φ→χ )) A2
4. (φ→ψ)→(φ→χ ) 2,3 MP
5. φ→χ 1,4 MP
Note that the lines in this “proof schema” are schemas (they contain Greek
letters), rather than wffs. It therefore isn’t a proof at all; rather, it becomes a
proof once you fill in wffs for the φ, ψ, and χ . We constructed a proof schema
because we want the MP technique to be applicable whenever we want to move
from formulas of the form φ→ψ and φ→(ψ→χ ), to a formula of the form
φ→χ , no matter what φ, ψ, and χ may be.
And while we’re on the topic of proof schemas, note also that whenever
one constructs a proof of a formula containing sentence letters, one could just
as well have constructed a similar proof schema. Corresponding to the proof
of (R→P )→(R→(Q→P )), for example, there is this proof schema:
1. [φ→(ψ→(χ →ψ))]→[(φ→ψ)→(φ→(χ →ψ))] A2

2. ψ→(χ →ψ) A1
3. [ψ→(χ →ψ)]→[φ→(ψ→(χ →ψ))] A1
4. φ→(ψ→(χ →ψ)) 2,3 MP
5. (φ→ψ)→(φ→(χ →ψ)) 1,4 MP
It’s usually more useful to think in terms of proof schemas, rather than proofs,
because they can go into our toolkit, if they have general applicability. The
proof schema we just constructed, for example, shows that anything of the
form (φ→ψ)→(φ→(χ →ψ)) is a theorem. As it happens, this is a fairly intuitive
theorem schema. Think of it as the principle of “weakening the consequent”.
χ →ψ is logically weaker than ψ, so if φ leads to ψ, φ must also lead to χ →ψ.
That sounds like a pattern that might well recur, so let’s put it into the toolkit,
under the label “weakening the consequent”. If we’re ever in the midst of a
proof and could really use a line of the form (φ→ψ)→(φ→(χ →ψ)), then we
can simply write that line down, and annotate on the right “weakening the
consequent”. Given the proof sketch above, we know that we could always
in principle insert a five-line proof of line; to save writing we simply won’t
bother. (Note that once we do this—omitting those five lines—the proofs we
are constructing will cease to be official proofs, since not every line will be
either an axiom or a line that follows from earlier lines by MP. They will be
instead proof sketches, which are in essence metalanguage arguments to the
effect that there exists some proof or other of the desired type. An ambitious
reader could always construct an official proof on the basis of the proof sketch,
by taking each of the bits, filling in the details using the toolkit, and assembling
the results into one proof.)
Next I want to add to our toolkit the principle of “strengthening the an-
tecedent”: [(φ→ψ)→χ ]→(ψ→χ ). The intuitive idea is that if φ→ψ leads to
χ , then ψ ought to lead to χ , since ψ is logically stronger than φ→ψ. This
proof will be harder still; we’ll need to break it into bits and use the toolkit to
complete it. Here’s a sketch of the overall proof:
a. [(φ→ψ)→χ ]→[ψ→(φ→ψ)] see below

b. [(φ→ψ)→χ ]→[(ψ→(φ→ψ))→(ψ→χ )] see below
c. [(φ→ψ)→χ ]→[ψ→χ ] a,b MP method
All that remains is to supply separate proofs of lines a and b. Step a is pretty
easy. Its consequent, ψ→(φ→ψ) is an instance of A1, so we can prove it in one
line, then use “adding an antecedent” to get a.
Line b is a bit harder. It has the form: (α→β)→[(γ →α)→(γ →β)]. Call
this “adding antecedents” (and put it into the toolkit too), since it lets you add
the same antecedent (γ ) to both the antecedent and consequent of a condi-
tional (α→β). The following proof sketch for adding antecedents uses the MP
technique again!
1. [γ →(α→β)]→[(γ →α)→(γ →β)] A2

2. (α→β)→[γ →(α→β)] A1
3. (α→β)→[γ →(α→β)]→[(γ →α)→(γ →β)] adding an antecedent to
line 1
4. (α→β)→[(γ →α)→(γ →β)] 2, 3, MP method
(For this use of the MP method, we let φ = α→β, ψ = γ →(α→β), and

χ =(γ →α)→(γ →β).) Since we’ve now provided proof sketches for parts a. and
b., we’re finished our proof sketch for strengthening the antecedent.
Next let’s add a tool to our toolkit, to the effect that conditionals are “tran-
sitive”. Here’s a proof sketch for ` (φ→ψ)→[(ψ→χ )→(φ→χ )]:
1. (ψ→χ )→[(φ→ψ)→(φ→χ )] adding antecedents

2. [(ψ→χ )→[(φ→ψ)→(φ→χ )]]→ A2
[[(ψ→χ )→(φ→ψ)]→[(ψ→χ )→(φ→χ )]]
3. [(ψ→χ )→(φ→ψ)]→[(ψ→χ )→(φ→χ )] 1, 2 MP
4. [[(ψ→χ )→(φ→ψ)]→[(ψ→χ )→(φ→χ )]]→ strengthening the
[(φ→ψ)→[(ψ→χ )→(φ→χ )]] antecedent
5. (φ→ψ)→[(ψ→χ )→(φ→χ )] 3, 4 MP
Given this theorem, we can always move from φ→ψ and ψ→χ to φ→χ thus:
1. φ→ψ from earlier lines

2. ψ→χ from earlier lines
3. (φ→ψ)→[(ψ→χ )→(φ→χ )] theorem just proved
4. (ψ→χ )→(φ→χ ) 1, 3 MP
5. φ→χ 2, 4 MP
So, such moves may henceforth be justified by appeal to “transitivity”.

With transitivity in our toolkit, we can really get moving:
` [φ→(ψ→χ )]→[ψ→(φ→χ )] (“swapping antecedents”):
1. [φ→(ψ→χ )]→[(φ→ψ)→(φ→χ )] A2
2. [(φ→ψ)→(φ→χ )]→[ψ→(φ→χ )] strengthening the antecedent
3. [φ→(ψ→χ )]→[ψ→(φ→χ )] 1,2 transitivity
` (∼ψ→∼φ)→(φ→ψ) (“contraposition 1”):
1. (∼ψ→∼φ)→[(∼ψ→φ)→ψ] A3
2. [(∼ψ→φ)→ψ]→(φ→ψ) strengthening the antecedent
3. (∼ψ→∼φ)→(φ→ψ) 1, 2 transitivity
` (φ→ψ)→(∼ψ→∼φ) (“contraposition 2”):

1. (φ→ψ)→[(ψ→∼∼ψ)→(φ→∼∼ψ)] transitivity
2. (ψ→∼∼ψ)→[(φ→ψ)→(φ→∼∼ψ)] 1, swapping
antecedents
3. ψ→∼∼ψ exercise 2.4c
4. (φ→ψ)→(φ→∼∼ψ) 2, 3 MP
5. ∼∼φ→φ exercise 2.4b
6. (∼∼φ→φ)→[(φ→∼∼ψ)→(∼∼φ→∼∼ψ)] transitivity
7. (φ→∼∼ψ)→(∼∼φ→∼∼ψ) 5, 6 MP
8. (∼∼φ→∼∼ψ)→(∼ψ→∼φ) contraposition 1
9. (φ→ψ)→(∼ψ→∼φ) 4, 7, 8 transitivity (2x)
` ∼φ→(φ→ψ) (“ex falso quodlibet”):
1. ∼φ→(∼ψ→∼φ) A1
2. (∼ψ→∼φ)→(φ→ψ) contraposition 1
3. ∼φ→(φ→ψ) 1, 2 transitivity
Exercise 2.4 Establish each of the following. For these you may
use the toolkit.
a) Show that ` P →[(P →Q)→Q]
b) Show that ` ∼∼P →P (Hint: convert your proof that ∼∼P `

P from problem 2.3c into a proof that ` ∼∼P →P , making use of
the MP technique.)
c) Show that ` P →∼∼P
2.6 Soundness of PL
In this chapter we have discussed two approaches to propositional logic: the
proof-theoretic approach and the semantic approach. In each case, we in-
troduced formal notions of logical truth and logical consequence. For the
semantic approach, these notions involved truth in PL-interpretations. For the
proof-theoretic approach, we considered two formal definitions, one involving

sequent-proofs, the other involving axiomatic proofs.
An embarrassment of riches! We have multiple formal accounts of our
logical notions. But in fact, it can be shown that each of our definitions yields
exactly the same results. That is, whether you define a logical truth (for example)
as a formula that is true in all PL-interpretations, or as a formula that is the
last line of some PL axiomatic proof, or as a formula φ for which the sequent
∅ ` φ is a provable sequent, exactly the same formulas turn out logical truths.
Proving this is a major accomplishment of meta-logic. We won’t prove this
here, not in full anyway, but we will discuss the issues a bit, to show how such
metalogical proofs proceed.
Let’s focus, for the moment, on just two of our notions, the notion of
a theorem (last line of an axiomatic proof) and the notion of a valid formula
(true in all PL-interpretations). An important accomplishment of metalogic is
the establishment of the following two important connections between these
notions:
Soundness: Every PL-theorem is PL-valid
Completeness: Every PL-valid wff is a PL-theorem
Proof of soundness. It’s pretty easy to establish soundness. We do this by induc-

tion. But our inductive proof here is slightly different. We’re not trying to
prove something of the form “Every wff has property P”. Instead, we’re trying
to prove something of the form “Every theorem has property P”. In this case,
the property P is: being a valid formula.
Here’s how induction works in this case. A theorem is the last line of any
proof. So, to show that every theorem has a certain property P, all we need
to do is show that every time one adds another line to a proof, that line has
property P. Now, there are two ways one can add to a proof. First, one can add
an axiom. The base case of the inductive proof must show that adding axioms
always means adding a line with property P. Second, one can add a formula that
follows from earlier lines by a rule. The inductive step of the inductive proof
must show that in this case, too, one adds a line with property P, provided all
the preceding lines have property P. OK, here goes:
base case: here we need to show that every PL-axiom is valid. This is tedious
but straightforward. Take A1, for example. Suppose for reductio that some in-
stance of A1 is invalid, i.e., for some PL-interpretation I , VI (φ→(ψ→φ)) = 0.
Thus, VI (φ) = 1 and VI (ψ→φ) = 0. Given the latter, VI (φ) = 0—contradiction.

Analogous proofs can be given that instances of A2 and A3 are also valid.
induction step: here we begin by assuming that every line in a proof up to a
certain point is valid (this is the “inductive hypothesis”); we then show that if
one adds another line that follows from earlier lines by the rule modus ponens,
that line must be valid too. I.e., we’re trying to show that “modus ponens
preserves validity”. So, assume the inductive hypothesis: that all the earlier
lines in the proof are valid. And now, consider the result of applying modus
ponens. That means that the new line we’ve added to the proof is some formula
ψ, which we’ve inferred from two earlier lines that have the forms φ→ψ and
φ. We must show that ψ is a valid formula, i.e., that VI (ψ) = 1 for every PL-
interpretation I . By the inductive hypothesis, all earlier lines in the proof are
valid, and hence both φ→ψ and φ are valid. Thus, VI (φ)=1 and VI (φ→ψ) = 1.
But if VI (φ) = 1 then VI (ψ) can’t be 0, for if it were, then VI (φ→ψ) would be
0, and it isn’t. Thus, VI (ψ) = 1.
We’ve shown that axioms are valid, and that modus ponens preserves validity.
So, by induction, every time one adds to a proof, one adds a valid formula.
So the last line in a proof is always a valid formula. Thus, every theorem is
valid.
Notice the general structure of this proof: we first showed that every axiom
has a certain property, and then we showed that the rule of inference preserves
the property. Given the definition of ‘theorem’, it followed that every theorem
has the property. We chose our definition of a theorem with just this sort of
proof in mind.
Remember that this is a proof in the metalanguage, about propositional
logic. It isn’t a proof in any system of derivation.
One nice thing about soundness is that it lets us establish facts of unprov-
ability. Soundness says “if ` φ then φ”. Equivalently, it says: “if 2 φ then
0 φ”. So, to show that something isn’t a theorem, it suffices to show that it
isn’t valid. Consider, for example, the formula (P →Q)→(Q→P ). There exist
PL-interpretations in which the formula is false, namely, PL-interpretations in
which P is 0 and Q is 1. So, (P →Q)→(Q→P ) is not valid (since it’s not true
in all PL-interpretations.) But then soundness tells us that it isn’t a theorem
either. In general: if we’ve established soundness, then in order to show that a
formula isn’t a theorem, all we need to do is find an interpretation in which it
isn’t true.
One could also prove soundness for the natural deduction system of section
2.4. The soundness proof, for instance, would proceed by proving by induction
that whenever sequent Γ ` φ is provable, φ is a semantic consequence of Γ.
The main thing would be to show that each rule of inference (As, RAA, ∧I, ∧E,
etc.) preserves semantic consequence. But note how much more involved this
proof would be, since there are so many rules of inference. The paucity of rules
in the axiomatic system made the construction of proofs within that system a
real pain in the neck, but now we see how it makes metalogical life easier.
Before we leave this section, let me summarize and clarify the nature of
proofs by induction. Induction is the method of proof to use whenever one
is trying to prove that each entity of a certain sort has a certain feature F ,
where each such entity is generated from certain “starting points” by a finite
number of successive “operations”. To do this, one establishes two things: a)
that the starting points have feature F , and b) that the operations preserve
feature F —i.e., that if the inputs to the operations have feature F then the
output also has feature F .
In logic, it is important to distinguish two different cases where proofs by
induction are needed. One case is where one is establishing a fact of the form:
every theorem has a certain feature F . (The proof of soundness is an example of
this case.) Here’s why induction is applicable: a theorem is defined as the last
line of a proof. So the fact to be established is that every line in every proof has
feature F . Now, a proof is defined as a finite sequence, where each member is
either an axiom or follows from earlier lines by the rule modus ponens. The
axioms are the “starting points” and modus ponens is the “operation”. So if
we want to show that every line in every proof has feature F , all we need to
do is show that a) the axioms all have feature F , and b) show that if you start
with formulas that have feature F , and you apply modus ponens, then what you
get is something with feature F . More carefully, b) means: if φ has feature F ,
and φ→ψ has feature F , then ψ has feature F . Once a) and b) are established,
one can conclude by induction that all lines in all proofs have feature F . When
one gives this first sort of inductive argument, for the conclusion that every
theorem φ has a certain feature, it is sometimes called “induction on the proof
of φ” or “induction on the length of φ’s proof”.
A second case in which induction may be used is when one is trying to
establish a fact of the form: every formula has a certain feature F . (The proof
that every wff has a finite number of sentence letters is an example of this
case.) Here’s why induction is applicable: all formulas are built out of sentence
letters (the “starting points”) by successive applications of the rules of formation
(“operations”) (the rules of formation, recall, say that if φ and ψ are formulas,
then so are (φ→ψ) and ∼φ.) So, to show that all formulas have feature F ,
we must merely show that a) all the sentence letters have feature F , and b)
show that if φ and ψ both have feature F , then both (φ→ψ) and ∼φ also
will have feature F . When one gives this second sort of inductive argument,
for the conclusion that every formula φ has a certain feature, it is sometimes
called “induction on the construction of φ”, or “induction on the number of
connectives in φ”.
Inductions in logic can take yet other forms, but these two are particularly
common.
If you’re ever proving something by induction, it’s important to identify
what sort of inductive proof you’re constructing. What are the entities you’re
dealing with? What is the feature F ? What are the starting points, and what
are the operations generating new entities from the starting points? If you’re
trying to construct an inductive proof and get stuck, you should return to these
questions and make sure you’re clear about their answers.
Exercise 2.5 Consider the following (strange) system of propo-

sitional logic. The definition of wffs is the same as for standard
propositional logic, and the rules of inference are the same (just one
rule: modus ponens); but the axioms are different. For any wffs φ
and ψ, the following are axioms:
φ→φ
(φ→ψ)→(ψ→φ)
Establish the following two facts about this system:
· every theorem of this system has an even number of “∼”s.
· soundness is false for this system—i.e., some theorems are

not valid formulas
Exercise 2.6 Back to normal propositional logic and its semantics

(for more practice on inductive proofs). Show that the truth value of
a formula depends only on the truth values of the sentence letters in
that formula. That is, let φ be any wff and let V and V 0 be valuations
that agree on the sentence letters in φ (i.e., for any sentence letter
α, if α is in φ then V (α) = V 0 (α)). Show that V (φ) = V 0 (φ).
Exercise 2.7 Prove the following form of soundness: for any set
of formulas, Γ , and any formula φ, if Γ ` φ then Γ φ (i.e., if φ is
provable from Γ then φ is a semantic consequence of Γ.)
Exercise 2.8 Prove the soundness of the sequent calculus. That

is, show that if Γ ` φ is a provable sequent, then Γ φ. (No need
to go through each and every detail of the proof once it becomes
repetitive. Hint: call a sequent Γ ` φ valid iff Γ φ. Your proof
should be a proof by induction, and it should use the notion of a
valid sequent.)
2.7 Completeness of PL
In this section we’ll prove completeness for propositional logic. It will be a bit
more difficult than the preceding sections, and may be skipped without much
loss. If you decide to work through the more difficult sections dealing with
metalogic later in the book (for example sections 6.5 and 6.6), you might first
return to this section.
Before we prove completeness, we’ll need to prove a helpful theorem (which
is interesting in its own right) and a Lemma.
As you learned in section 2.5.1 (perhaps to your dismay), constructing
axiomatic proofs is much harder than constructing sequent proofs. It’s hard to
prove things when you’re not allowed to use conditional proof! Nevertheless,
one can prove a metalogical theorem about our axiomatic system that is closely
related to conditional proof:
Deduction theorem: If Γ, φ ` ψ, then Γ ` φ→ψ
That is: whenever there exists a proof from (Γ and) φ to ψ, then there also
exists a proof of φ→ψ (from Γ).
Suppose we want to prove φ→ψ. Our axiomatic system does not allow
us to assume φ in a conditional proof of φ→ψ. But once we’ve proved the
deduction theorem, we’ll be able to do the next best thing. Suppose we write
down a proof of ψ from {φ}. That is, we write down a proof in which each line
is either i) a member of {φ} (that is, φ itself), or ii) an axiom, or iii) follows
from earlier lines in the proof by modus ponens. The deduction theorem then
lets us conclude that some proof of φ→ψ exists. We won’t have constructed such
a proof ourselves; we only constructed the proof from φ to ψ. Nevertheless
the deduction theorem assures us that it exists. More generally, whenever we
can construct a proof of ψ from φ plus some other premises (the formulas in
some set Γ), then the deduction theorem assures us that some proof of φ→ψ
from those other premises also exists.
Proof of the deduction theorem. Suppose Γ, φ ` ψ. That is, there is some proof,
call it “Proof A”, of ψ from Γ ∪ {φ}. Such a proof looks like this:
1. α1
2. α2
.
.
n. ψ
where each αi is either a member of Γ ∪ {φ}, an axiom, or follows from earlier
lines in the proof by MP. Our strategy will be to establish that:
(*) for each αi in proof A, Γ ` φ→αi
We already know that each line of proof A is provable from Γ ∪ φ; what (*) says
is that if you stick “φ→” in front of any of those lines, the result is provable
from Γ all by itself. Once we succeed in establishing (*) then we will have
proved the deduction theorem. For the last line of proof A is ψ; (*) then tells
us that φ→ψ is provable from Γ.
(*) says that each line of proof A has a certain feature, namely, the feature of:
being provable from Γ when prefixed with “φ→”. Just as in the proof of soundness,
this calls for the method of proof by induction, and in particular, induction on
φ’s proof. Here goes.
What we’re going to do is show that whenever a line is added to proof A,
then it has the feature—provided, that is, that all earlier lines in the proof have
the feature. There are three cases in which a line αi could have been added
to proof A. The first case is where αi is an axiom. We must show that αi has
the feature—that is, show that Γ ` φ→αi . Well, we can prove φ→αi from Γ as
follows:
1. αi axiom
2. φ→αi adding an antecedent
This is not an official proof, of course; it’s a proof sketch. And note that we
didn’t need to use any members of Γ in the proof. That’s OK; if you look back
at the definition of a proof from a set, you’ll see that this counts officially as a
proof from Γ.
The second case in which a line αi could have been added to proof A is
where αi is a member of Γ ∪ {φ}. This subdivides into two subcases. The first
is where αi is φ itself. Here, φ→αi is φ→φ, which is shown in exercise 2.3a to
be a theorem, i.e., provable from no premises at all. So it is obviously provable
from Γ. The second subcase is where αi ∈ Γ. But here we can prove φ→αi
from Γ as follows:
1. αi member of Γ
2. φ→αi adding an antecedent
The first two cases were “base” cases of our inductive proof, because we
didn’t need to assume anything about earlier lines in proof A. The third case in
which a line αi could have been added to proof A leads us to the inductive part
of our proof: the case in which αi follows from two earlier lines of the proof
by MP. Here we simply assume that those earlier lines of the proof have the
feature we’re interested in (this assumption is the “inductive hypothesis”; the
feature, recall, is: being provable from Γ when prefixed with “φ→”) and we show
that αi has the feature as well.
So: we’re considering the case where αi follows from earlier lines in the
proof by modus ponens. That means that the earlier lines have to have the
forms χ →αi and χ . Furthermore, the inductive hypothesis tells us that the
result of prefixing either of these earlier lines with “φ→” is provable from Γ.
That is, we know that some proof from Γ culminates in φ→(χ →αi ), and some
other proof from Γ culminates in φ→χ . We can then string these two proofs
together into a new proof, and then continue that new proof as follows:
.
.
k. φ→(χ →αi )
.
.
l. φ→χ
l + 1. φ→αi k, l , MP method
This is a proof of φ→αi from Γ.
Thus, in all three cases, whenever αi was added to proof A, there always
existed some proof of φ→αi from Γ. By induction, (*) is established; and this
in turn completes the proof of the deduction theorem.
Next we’ll prove the following lemma:
Lemma: Let I be any PL-interpretation, let φ be any wff, let s1 . . . sn be the
sentence letters in φ, and where ψ is any formula, define ψ0 as being ψ itself if
ψ is true in I , and as being ∼ψ if ψ is false in I . Then s10 . . . sn0 ` φ0
A very rough way of thinking about what Lemma says is this: the truth value of a
formula is provably settled by the truth values of its sentence letters (“provably”
in the sense of provability in our axiomatic system.)
Proof of Lemma. Let I , φ, and s1 . . . sn be as described in Lemma. We must
show that φ0 is provable from {s10 . . . sn0 }. We’ll show this by induction. With an
eye toward setting up the inductive proof correctly, think of Lemma as saying:
every formula has the feature being a formula whose primed version is provable from
the primed versions of its sentence letters. This makes it clear that the assertion
we’re trying to prove has the form “every formula has a certain feature”, and
thus calls for proof by induction on the formula’s construction (rather than
proof by induction on a proof, as in the previous two inductive proofs). So,
we’ll need to show that all sentence letters have the feature (base case), and
then show that if α and β have the feature (inductive hypothesis), then both
∼α and α→β must have the feature as well.
Base case: suppose φ is a sentence letter. Then there is just one si , which is
φ itself. So what we need to show is: φ0 ` φ0 . But that’s trivial; we can give a
one-line proof of φ0 from {φ0 }:
1. φ0 member of {φ0 }
Now assume the inductive hypothesis: that both α and β have the feature.
That is, where s1 . . . sn are the sentence letters in α, and t1 . . . t m are the sentence
letters in β, we are assuming:
(a) s10 . . . sn0 ` α0
(b) t10 . . . tn0 ` β0
First we must show that ∼α has the feature. Since ∼α has the same sentence
letters as α (namely, s1 . . . sn ), this means showing that s10 . . . sn0 ` (∼α)0 . Now,
(∼α)0 is either ∼α or ∼∼α depending on whether ∼α is true or false (in I ; I’ll
suppress this from now on). We’ll consider these cases separately.
· In the former case, α is false; and so α0 is ∼α. Then (a) tells us that
s10 . . . sn0 ` ∼α. Since (∼α)0 is ∼α in this case, we’ve already shown what
we wanted: s10 . . . sn0 ` (∼α)0 .
· In the latter case, α is true, and so α0 is just α. So what (a) tells us in this
case is: s10 . . . sn0 ` α. Furthermore, since in this case (∼α)0 is ∼∼α, what
we’re trying to establish is s10 . . . sn0 ` ∼∼α. So, to construct a proof of
∼∼α from {s10 . . . sn0 }, begin with a proof of α from {s10 . . . sn0 } (we know
that such a proof exists from what (a) told us). Then insert a proof of
α→∼∼α (exercise 2.4c). Finish the proof by concluding ∼∼α by MP.
Next we must show that α→β has the feature. The sentence letters in
α→β are s1 . . . sn , t1 . . . t m , so what we must show is: s10 . . . sn0 , t10 . . . t m0 ` (α→β)0 .
We’ll consider three separate cases:
· First case: β is true. Then α→β is also true, and so (α→β)0 is just α→β.
We may then construct the desired proof from {s10 . . . sn0 , t10 . . . t m0 } of α→β
as follows. Here β0 is just β, so (b) tells us that t10 . . . tn0 ` β. So we may
begin our desired proof with a proof of β from {t10 . . . tn0 } (ipso facto this
is a proof from {s10 . . . sn0 , t10 . . . t m0 }). We then use the technique of “adding
an antecedent” to move to α→β.
· Second case: α is false. Here α→β is again true, so again we must

construct a proof from {s10 . . . sn0 , t10 . . . t m0 } of α→β. α0 is now ∼α, so (a)
tells us that s10 . . . sn0 ` ∼α. So, begin with a proof from {s10 . . . sn0 } (and so
from {s10 . . . sn0 , t10 . . . t m0 }) of ∼α. Then insert a proof of ∼α→(α→β) (ex
falso quodlibet, from section 2.5.1), and then by modus ponens conclude
α→β.
· The remaining case is where α is true (hence α0 is α) and β is false (hence

β0 is ∼β). Here α→β is false, so we must construct a proof of ∼(α→β)
from {s10 . . . sn0 , t10 . . . t m0 }. Begin with a proof of α (guaranteed by (a)), and
continue with a proof of ∼β (guaranteed by (b)), and then continue as
follows:
.
.
i. α
.
.
j. ∼β
j +1. [∼∼(α→β)→∼β]→[(∼∼(α→β)→β)→∼(α→β)]
A3
j +2. ∼∼(α→β)→∼β j , adding an antecedent
j +3. (∼∼(α→β)→β)→∼(α→β) j +1, j +2, MP
j +4. α→[∼∼(α→β)→β] see below
j +5. ∼∼(α→β)→β i, j +4, MP
j +6. ∼(α→β) j +3, j +5, MP
As for step j +4, consider the following proof of β from {α, ∼∼(α→β)}:
1. α member of {α, ∼∼(α→β)}

2. ∼∼(α→β) member of {α, ∼∼(α→β)}
3. ∼∼(α→β)→(α→β) exercise 2.4b
4. α→β 2, 3 MP
5. β 1, 4 MP
The existence of this proof shows that α, ∼∼(α→β) ` β. The de-

duction theorem then allows us to conclude that α ` ∼∼(α→β)→β.
And another use of the deduction theorem allows us to conclude that
` α→[∼∼(α→β)→β]. So we know that there exists some proof of
α→[∼∼(α→β)→β]. Any such proof may be inserted into step j +4
above.
This completes the inductive proof of Lemma.

We’ll now use the deduction theorem to prove completeness:
Completeness: if φ then ` φ
Proof. Suppose φ. We’ll prove ` φ in a series of stages. Let s1 . . . sn be the

sentence letters in φ.
Stage 1a: Let I be any PL-interpretation in which sn is true. Lemma tells
us that s10 . . . sn0 ` φ0 . Since φ, we know that φ is true in I , so φ0 will be φ.
Also, sn0 is sn . So we’ve learned: s10 . . . sn−1
0
, sn ` φ. By the deduction theorem,
0 0
s1 . . . sn−1 ` sn →φ.
Stage 1b: Let’s now choose a different interpretation, just like I but in
which sn is false. If we apply Lemma again, sn0 is now ∼sn , and φ0 is still φ,
so we have: s10 . . . sn−1
0
, ∼sn ` φ, from which we may infer s10 . . . sn−1 0
` ∼sn →φ
by the deduction theorem. (Note: it’s legitimate to continue to use the same
names for the “primed” versions of s1 . . . sn−1 , because our new interpretation
function assigns the same truth values to these sentence letters as did I .)
Stage 1c: We’ve learned that s10 . . . sn−10
` sn →φ and s10 . . . sn−1
0
` ∼sn →φ.
Let’s now construct a proof of φ from {s1 , . . . sn−1 } by first beginning with a
0 0
proof of sn →φ, continuing with a proof of ∼sn →φ, and then continuing as
follows:
.
.
i. sn →φ
.
.
j. ∼sn →φ
j +1. ∼φ→∼sn i , contraposition 2, MP
j +2. ∼φ→∼∼sn j , contraposition 2, MP
j +3. (∼φ→∼∼sn )→[(∼φ→∼sn )→φ] A3
j +4. φ j + 1, j + 2, j + 3, MP (x2)
Thus we have shown that s10 . . . sn−1

0
` φ.
Stage 2 will show that s1 . . . sn−2 ` φ. That is, it will remove one further of
0 0
the si0 s on the left of the `. It proceeds like stage 1:

Stage 2a: first choose a new interpretation just like the one chosen at the
end of stage 1b, except that sn−1 is true in it. Following the strategy of stage 1a,
show that s10 . . . sn−2
0
` sn−1 →φ.
Stage 2b: choose a new interpretation just like the preceding, but in which
sn−1 is false. Following the strategy of 1b, show that s10 . . . sn−2
0
` ∼sn−1 →φ.
Stage 2c: Following the strategy of 1c, show on the basis of stages 2a and
2b that s10 . . . sn−2
0
` φ.
Stage 2 removed one more of the si0 s on the left of the `. Stage 3 will remove
another one, and so we will have: s10 . . . sn−3
0
` φ. Each stage removes another
one; and so after the last stage, stage nc, they will all be gone and we will have
shown that ` φ.
Chapter 3
Variations and Deviations from

Standard Propositional Logic
s promised, we will not stop with the standard logics familiar from in-
A troductory textbooks. In this chapter we examine some philosophically
important variations and deviations from standard propositional logic.
3.1 Alternate connectives

3.1.1 Symbolizing truth functions in propositional logic
Our propositional logic is in a sense “expressively complete”. To get at this idea,
let’s introduce the idea of a truth function. A truth function is a (finite-placed)
function that maps truth values (i.e., 0s and 1s ) to truth values. For example,
here is a truth function, f :
f (1) = 0
f (0) = 1
This is a called one-place function because it takes only one truth value as input.
In fact, we have a name for this truth function: negation. And we express that
truth function with our symbol ∼. So: negation is a truth function that we can
express in propositional logic.
63
CHAPTER 3. VARIATIONS AND DEVIATIONS FROM PL 64
Another truth function we can express is the conjunction truth function:
g (1, 1) = 1
g (1, 0) = 0
g (0, 1) = 0
g (0, 0) = 0
Conjunction is a two-place truth function, which means that it takes two truth
values as inputs. We have a symbol for this truth function as well: ∧.
Here’s another truth function:
i(1, 1) = 0
i(1, 0) = 1
i(0, 1) = 1
i(0, 0) = 1
Think of this truth function as “not both”. Unlike the negation and conjunction
truth functions, we don’t have a single symbol for this truth function. Never-
theless, it too can be expressed in propositional logic. If we want to express
“not-both (P , Q)”, we can just write:
∼(P ∧Q)
In fact, it’s not hard to show that any truth function (of any finite number
of places) can be expressed in propositional logic using just the ∧, ∨, and ∼.
Proof. The proof will be informal; but before giving it, we need a precise
definition of what it means to say that a truth function “can be expressed” in
propositional logic.
Definition of expressability: n-place truth function h can be expressed in
propositional logic iff there is some sentence of propositional logic, φ, con-
taining n sentence letters, P1 . . . Pn , which has the following feature: whenever
P1 . . . Pn have the truth values t1 . . . tn , respectively, then the whole sentence φ
has the truth value h(t1 . . . tn )
Now for the proof. I’ll begin by illustrating the idea with an example. Suppose
we want to express the following three-place truth-function:
f (1, 1, 1) = 0
f (1, 1, 0) = 1
f (1, 0, 1) = 0
f (1, 0, 0) = 1
f (0, 1, 1) = 0
f (0, 1, 0) = 0
f (0, 0, 1) = 1
f (0, 0, 0) = 0
We must construct a sentence with three sentence letters, P1 , P2 , and P3 , whose
truth table “matches” function f . Now, if we ignore everything but the numbers
in the above picture of function f , we can think of it as a kind of truth table
for the sentence we’re after. The first column of numbers represents the truth
values of P1 , the second column, the truth values of P2 , and the third column,
the truth values of P3 ; and the far right column represents the truth values that
the desired formula should have. Each row represents a possible combination
of truth values for these sentence letters. Thus, the second row (“ f (1, 1, 0) = 0”)
is the combination where P1 is 1, P2 is 1, and P3 is 0; the fact that the fourth
column in this row is 1 indicates that the desired formula should be true here.
Since function f returns the value 1 in just three cases (rows two, four, and
seven), the sentence we’re after should be true in exactly those three cases: (a)
when P1 , P2 , P3 take on the three truth values in the second row (i.e., 1, 1, 0);
(b) when P1 , P2 , P3 take on the three truth values in the fourth row (1, 0, 0); and
(c) when P1 , P2 , P3 take on the three truth values in the seventh row (0, 0, 1) .
Now, we can construct a sentence that is true in case (a) and false otherwise:
P1 ∧P2 ∧∼P3 . We can also construct a sentence that’s true in case (b) and false
otherwise: P1 ∧∼P2 ∧∼P3 . And we can also construct a sentence that’s true in
case (c) and false otherwise: ∼P1 ∧∼P2 ∧P3 . But then we can simply disjoin these
three sentences to get the sentence we want:
(P1 ∧P2 ∧∼P3 ) ∨ (P1 ∧∼P2 ∧∼P3 ) ∨ (∼P1 ∧∼P2 ∧P3 )
(Strictly speaking the three-way conjunctions, and the three-way disjunction,
need parentheses added, but since it doesn’t matter where they’re added—
conjunction and disjunction are associative—I’ve left them off.)
This strategy is in fact purely general. Any n-place truth function, f , can
be represented by a chart like the one above. Each row in the chart consists of
a certain combination of n truth values, followed by the truth value returned
by f for those n inputs. For each such row, construct a conjunction whose
i th conjunct is Pi if the i th truth value in the row is 1, and ∼Pi if the i th truth
value in the row is 0. Notice that the conjunction just constructed is true if and
only if its sentence letters have the truth values corresponding to the row in
question. The desired formula is then simply the disjunction of all and only
the conjunctions for rows where the function f returns the value 1.1 Since the
conjunction for a given row is true iff its sentence letters have the truth values
corresponding to the row in question, the resulting disjunction is true iff its
sentence letters have truth values corresponding to one of the rows where f
returns the value true, which is what we want.
Say that a set of connectives is adequate iff one can express all the truth
functions using a sentence containing only those connectives. What we just
showed was that the set {∧, ∨, ∼} is adequate. We can then use this fact to
prove that other sets of connectives are adequate. For example, it is easy to
prove that φ∨ψ has the same truth table as (is true relative to exactly the same
PL-interpretations as) ∼(∼φ∧∼ψ). But that means that for any sentence χ
whose only connectives are ∧, ∨, and ∼, we can construct another sentence χ 0
with the same truth table but whose only connectives are ∧ and ∼: simply begin
with χ and use the equivalence between φ∨ψ and ∼(∼φ∧∼ψ) to eliminate
all occurrences of ∨ in favor of occurrences of ∧ and ∼. But now consider
any truth function f . Since {∧, ∨, ∼} is adequate, f can be expressed by some
sentence χ ; but χ has the same truth table as some sentence χ 0 whose only
connectives are ∧ , and ∼; hence f can be expressed by χ 0 as well. So {∧, ∼} is
adequate.
Similar arguments can be given to show that other connective sets are
adequate as well. For example, the ∧ can be eliminated in favor of the → and
the ∼ (since φ∧ψ has the same truth table as ∼(φ→∼ψ)); therefore, since
{∧, ∼} is adequate, {→, ∼} is also adequate.
1
Special case: if there are no such rows—i.e., if the function returns 0 for all inputs—
then let the formula be simply any logically false formula containing P1 . . . Pn , for example
P1 ∧∼P1 ∧P2 ∧P3 ∧ · · · ∧Pn .
3.1.2 Inadequate connective sets

Can we show that certain sets of connectives are not adequate?
We can quickly answer yes, for a trivial reason. The set {∼} isn’t adequate,
for the simple reason that, since ∼ is a one-place connective, no sentence with
more than one sentence letter can be built using just ∼. So there’s no hope of
symbolizing all the n-place truth functions, for any n > 1, using just the ∼.
More interestingly, we can show that there are inadequate connective sets
containing two-place connectives. Let’s prove that {∧, →} is not an adequate set
of connectives. We’ll do this by proving that if those were our only connectives,
we couldn’t express the negation truth function. And we’ll demonstrate that by
proving the following fact:
(+) For any sentence, φ, containing just sentence letter P and the connectives
∧ and →, φ is true in any PL-interpretation in which P is true
We’ll again use the method of induction. We want to show that (+) holds for all
sentences. So we first prove that (+) is true for all sentence with no connectives
(i.e., for sentences containing just sentence letters.) This is the base case, and
is very easy here, since if φ has no connectives, then obviously φ is just the
sentence letter P itself, in which case, clearly, φ is true in any PL-interpretation
in which P is true. Next we assume the inductive hypothesis:
(ih) (+) is true for sentences φ and ψ. (That is, in any interpretation in which
P is true, both φ and ψ are true.)
And we try to show, on the basis of this assumption, that (+) is true for φ∧ψ
and for φ→ψ. This is easy to do. First we show that (+) is true for φ∧ψ—
that is, φ∧ψ is true in any interpretation in which P is true. But we know
by the inductive hypothesis that φ and ψ are individually true in any such
interpretation. But then, we know from the truth table for ∧ that φ∧ψ is also
true in any such interpretation. The reasoning is exactly parallel for φ→ψ: the
inductive hypothesis tells us that whenever P is true, so are φ and ψ, and then
we know that in this case φ→ψ must also then be true, by the truth table for
→. Therefore, by induction, the result is proved.
Exercise 3.1 Let’s define the connective % to have the following

truth table:
% 1 0
1 0 1
0 1 0
Can all the truth functions be expressed using just the %? Justify
your answer.
3.1.3 Sheffer stroke

We’ve seen how we can choose alternate sets of connectives. Some of these
choices are adequate (i.e., allow symbolization of all truth functions), others
are not.
As we saw, there are some truth functions that can be expressed in propo-
sitional logic, but not by a single connective (e.g., the not-both function i
discussed above.)
We could change this, by adding a new connective. Let’s use a new connec-
tive, the “Sheffer stroke”, |, to express not-both. φ|ψ is to mean that not both
φ and ψ are true, so let’s stipulate that φ|ψ will have the same truth table as
∼(φ∧ψ), i.e:
% 1 0
1 0 1
0 1 1
Now here’s an exciting thing about |: it’s an adequate connective all on its own.
You can express all the truth functions using just |!
Here’s how we can prove this. We showed above that {→, ∼} is adequate;
so all we need to do is show how to define the → and the ∼ using just the |.
Defining ∼ is easy; φ|φ has the same truth table as ∼φ. As for φ→ψ, think of
it this way. φ→ψ is equivalent to ∼(φ∧∼ψ), i.e., φ|∼ψ. But given the method
just given for defining ∼ in terms of |, we know that ∼ψ is equivalent to ψ|ψ.
Thus, φ→ψ has the same truth table as: φ|(ψ|ψ).
Exercise 3.2 For each of the following two truth functions, f and
g , first find a sentence that expresses it in standard propositional
logic (i.e., with ∼, ∧, ∨, ↔, →); then find a sentence that expresses
it using just the Sheffer stroke:
f (1, 1) = 1 g (1, 1, 1) = 1
f (1, 0) = 0 g (1, 1, 0) = 0
f (0, 1) = 0 g (1, 0, 1) = 1
f (0, 0) = 1 g (1, 0, 0) = 1
g (0, 1, 1) = 1
g (0, 1, 0) = 1
g (0, 0, 1) = 0
g (0, 0, 0) = 1
Feel free to avoid writing out extremely long formulas by making

abbreviations, saying things like “make such-and-such substitutions
throughout”, etc.
Exercise 3.3 Show that all truth functions can be defined using
just ↓ (nor). The truth table for ↓ is the following:
% 1 0
1 0 0
0 0 1
3.2 Polish notation

Alternate connectives, like the Sheffer stroke, are called “variations” of standard
logic because they don’t really change what we’re saying with propositional
logic; it’s just a change in notation.
Another fun change in notation is polish notation. The basic idea of polish
notation is that the connectives all go before the sentences they connect. Instead
of writing P ∧Q, we write ∧P Q. Instead of writing P ∨Q we write ∨P Q.
Formally, here is the definition of a wff:
Definition of wffs for Polish notation:
· sentence letters are wffs

· if φ and ψ are wffs, then so are:
∼φ
∧φψ ∨φψ →φψ ↔φψ
What’s the point? This notation eliminates the need for parentheses. With the
usual notation, in which we put the connectives between the sentences they
connect, we need parentheses to distinguish, e.g.:
(P ∧Q)→R
P ∧(Q→R)
But with Polish notation, these are distinguished without parentheses; they
become:
→∧P QR
∧P →QR
respectively.
Exercise 3.4 Translate each of the following into Polish notation:
a) P ↔∼P
b) (P →(Q→(R→∼∼(S∨T ))))
c) [(P ∧∼Q)∨(∼P ∧Q)]↔∼[(P ∨∼Q)∧(∼P ∨Q)]
3.3 Multi-valued logic2

Logicians have considered adding a third truth value to the usual two. In these
new systems, in addition to truth (1) and falsity (0) , we have a third truth value,
2
See Gamut (1991a, pp. 173-183).
#. There are a number of things one could take # to mean (e.g., “meaningless”,
or “undefined”, or “unknown”).
Standard logic is “bivalent”—that means that there are no more than two
truth values. So, moving from standard logic to a system that admits a third
truth value is called “denying bivalence”. One could deny bivalence, and go
even further, and admit four, five, or even infinitely many truth values. But
we’ll only discuss trivalent systems—i.e., systems with only three truth values.
Why would one want to admit a third truth value? There are various
philosophical reasons one might give. One concerns vagueness. A person with
one dollar is not rich. A person with a million dollars is rich. Somewhere in the
middle, there are some people that are hard to classify. Perhaps a person with
$100,000 is such a person. They seem neither definitely rich nor definitely
not rich. So there’s pressure to say that the statement “this person is rich” is
capable of being neither definitely true nor definitely false. It’s vague.
Others say we need a third truth value for statements about the future. If
it is in some sense “not yet determined” whether there will be a sea battle
tomorrow, then (it is argued) the sentence:
There will be a sea battle tomorrow

is neither true nor false. In general, statements about the future are neither
true nor false if there is nothing about the present that determines their truth
value one way or the other.3
Yet another case in which some have claimed that bivalence fails concerns
failed presupposition. Consider this sentence:
Ted stopped beating his dog.

In fact, I have never beaten a dog. I don’t even have a dog. So is it true that
I stopped beating my dog? Obviously not. But on the other hand, is this
statement false? Certainly no one would want to assert its negation: “Ted has
not stopped beating his dog”. The sentence presupposes that I was beating a dog;
since this presupposition is false, the question of the sentence’s truth does not
arise: the sentence is neither true nor false.
For a final challenge to bivalence, consider the sentence:
3
An alternate view preserves the “openness of the future” as well as bivalence: both ‘There
will be a sea battle tomorrow’ and ‘There will fail to be a sea battle tomorrow’ are false. This
combination is not contradictory, provided one rejects the equivalence of “It will be the case
tomorrow that ∼φ” and “∼ it will be the case tomorrow that φ”.
Sherlock Holmes has a mole on his left leg

‘Sherlock Holmes’ doesn’t refer to a real entity. Further, Sir Arthur Conan
Doyle does not specify in his Sherlock Holmes stories whether Holmes has
such a mole. Either of these reasons might be argued to result in a truth value
gap for the displayed sentence.
It’s an interesting philosophical question whether any of these arguments
for bivalence’s failing are any good. But we won’t take up that question. Instead,
we’ll look at the formal result of giving up bivalence. That is, we’ll introduce
some non-bivalent formal systems. We won’t ask whether these systems really
model English correctly.
These systems all give different truth tables for the Boolean connectives.
The original truth tables give you the truth values of complex formulas based
on whether their sentence letters are true or false (1 or 0) . The new truth
tables need to take into account cases where the sentence letters are # (neither
1 nor 0) .
3.3.1 Łukasiewicz’s system

Here are the new truth tables (let’s skip the ↔):
∼ ∧ 1 0 # ∨ 1 0 # → 1 0 #
1 0 1 1 0 # 1 1 1 1 1 1 0 #
0 1 0 0 0 0 0 1 0 # 0 1 1 1
# # # # 0 # # 1 # # # 1 # 1
Using these truth tables, one can calculate truth values of wholes based on
truth values of parts.
Example 3.1: Where P is 1, Q is 0 and R is #, calculate the truth value of
(P∨Q)→∼(R→Q). First, what is R→Q? Answer, from the truth table for →: #.
Next, what is ∼(R→Q)? From the truth table for ∼, we know that the negation
of a # is a #. So, ∼(R→Q) is #. Next, P ∨Q: that’s 1∨0—i.e., 0. Finally, the
whole thing: 0→#, i.e., 1.
We can formalize this a bit more by defining up new notions of an inter-
pretation, and of truth relative to an interpretation:
Definition of trivalent interpretation: A trivalent interpretation is a func-

tion that assigns to each sentence letter exactly one of the values: 1, 0, #.
Definition of trivalent valuation: For any trivalent interpretation, I , the

Łukasiewicz-valuation for I , ŁVI , is defined as the function that assigns to
each wff either 1, 0, or #, and which is such that, for any wffs φ and ψ,
ŁVI (φ) = I (φ) if φ is a sentence letter

 1 if ŁVI (φ) = 1 and ŁVI (ψ) = 1

ŁVI (φ∧ψ) = 0 if ŁVI (φ) = 0 or ŁVI (ψ) = 0


# otherwise

 1 if ŁVI (φ) = 1 or ŁVI (ψ) = 1

ŁVI (φ∨ψ) = 0 if ŁVI (φ) = 0 and ŁVI (ψ) = 0


# otherwise

 1 if ŁVI (φ) = 0, or ŁVI (ψ) = 1, or

 ŁVI (φ) = ŁVI (ψ) = #


ŁVI (φ→ψ) =



 0 ŁVI (φ) = 1 and ŁVI (ψ) = 0

# otherwise

 1 if ŁVI (φ) = 0

ŁVI (∼φ) = 0 if ŁVI (φ) = 1


# otherwise
Let’s define validity and semantic consequence for Łukasiewicz’s system much
like we did for standard PL:
Łukasiewicz definitions of validity and consequence:
· φ is Łukasiewicz-valid (“Ł φ”) iff for every trivalent interpretation I ,

ŁVI (φ) = 1
· φ is a Łukasiewicz-semantic-consequence of Γ (“Γ Ł φ”) iff for every
trivalent interpretation, I , if ŁVI (γ ) = 1 for each γ ∈ Γ, then ŁVI (φ) =
1
Notice that there are now two ways a formula can fail to be valid. It can
be 0 under some trivalent interpretation, or it can be # under some trivalent
interpretation. “Valid” (under this definition) means always true; it does not
mean never false. (Similarly, the defined notion of semantic consequence is
that of truth-preservation, not nonfalsity-preservation.) The definition leaves
it open that a formula might be never-false, and still not be always-true: such a
formula would be sometimes # and sometimes 1, but never 0.
Example 3.2: Is P ∨∼P Łukasiewicz-valid? Answer: no, it isn’t. Suppose P

is #. Then ∼P is #; but then the whole thing is # (since #∨# is #.)
Example 3.3: Is P →P Łukasiewicz-valid? Answer: yes. P could be either 1,

0 or #. From the truth table for →, we see that P →P is 1 in all three cases.
Exercise 3.5 Assuming Łukasiewicz’s tables, show that the → is

not definable in terms of the ∼, ∧, and ∨. That is, show that there
is no wff φ such that i) φ contains just the sentence letters P and
Q, plus the connectives ∼, ∧, and ∨ (plus parentheses), and ii) φ
has the same truth table as P →Q (i.e., φ is true in exactly the same
Łukasiewicz-valuations as P →Q).
Exercise 3.6 As noted, the definition of Łukasiewicz-validity leaves

it open that a formula might be never-false, and still not be always-
true. Give an example of such a formula.
3.3.2 Kleene’s “strong” tables

This system is like Łukasiewicz’s system, except that the truth table for the →
is different:
→ 1 0 #
1 1 0 #
0 1 1 1
# 1 # #
As with Łukasiewicz’s system, let’s continue to understand validity as truth in

all trivalent interpretations, and semantic consequence as the preservation of
truth in a given trivalent interpretation.
Here is the intuitive idea behind the Kleene tables. Let’s call the truth values
0 and 1 the “classical” truth values. If a formula’s halves have only classical truth
values, then the truth value of the whole formula is just the classical truth value
determined by the classical truth values of the halves. But if one or both halves
are #, then we must consider the result of turning each # into one of the classical
truth values. If the entire formula would sometimes be 1 and sometimes be 0
after doing this, then the entire formula is #. But if the entire formula always
takes the same truth value, X, no matter which classical truth value any #s are
turned into, then the entire formula gets this truth value X. Intuitively: if there
is “enough information” in the classical truth values of a formula’s parts to settle
on one particular classical truth value, then that truth value is the formula’s
truth value.
Take the truth table for φ→ψ, for example. When φ is 0 and ψ is #, the
whole formula is 1—because the false antecedent is sufficient to make the whole
formula true, no matter what classical truth value we convert ψ to. On the
other hand, when φ is 1 and ψ is #, then the whole formula is #. The reason is
that what classical truth value we substitute in for ψ’s # affects the truth value of
the whole. If the # becomes a 0 then the whole thing is 0; but if the # becomes
a 1 then the whole thing is 1.
There are two important differences between Łukasiewicz’s and Kleene’s
systems. The first is that, unlike Łukasiewicz’s system, Kleene’s system makes
the formula P →P invalid. The reason is that in Kleene’s system, #→# is #; thus,
P →P isn’t true in all valuations (it is # in the valuation where P is #.)
In fact, it’s easy to show that there are no valid formulas in Kleene’s system.
Proof. Consider the valuation that makes every sentence letter #. Here’s an
inductive proof that every wff is # in this interpretation. Base case: all the
sentence letters are # in this interpretation. (That’s obvious.) Inductive step:
assume that φ and ψ are both # in this interpretation. We need now to show
that φ∧ψ, φ∨ψ, and φ→ψ are all # in this interpretation. But that’s easy—just
look at the truth tables for ∧, ∨ and →. #∧# is #, #∨# is #, and #→# is #.
Even though there are no valid formulas in Kleene’s system, there are still
cases of semantic consequence. Semantic consequence for Kleene’s system is
defined as truth-preservation: Γ Kleene φ iff φ is true whenever every member
of Γ is true, given Kleene’s truth tables. Then P ∧Q Kleene P , since the only
way for P ∧Q to be true is for P to be true and Q to be true.
The second (related) difference is that in Kleene’s system, → is interde-
finable with the ∼ and ∨, in that φ→ψ has exactly the same truth table as
∼φ∨ψ. (Look at the truth tables to verify that this is true.) But that’s not true
for Łukasiewicz’s system. In Łukasiewicz’s system, when φ and ψ are both #,
then φ→ψ is 1, but ∼φ∨ψ is #.
3.3.3 Kleene’s “weak” tables (Bochvar’s tables)

This final system is based on a very different intuitive idea: that # is “infectious”.
That is, if any formula has a part that is #, then the entire formula is #. Thus,
the tables are as follows:
∼ ∧ 1 0 # ∨ 1 0 # → 1 0 #
1 0 1 1 0 # 1 1 1 # 1 1 0 #
0 1 0 0 0 # 0 1 0 # 0 1 1 #
# # # # # # # # # # # # # #
So basically, the classical bit of each truth table is what you’d expect; but
everything gets boring if any constituent formula is a #.
One way to think about these tables is to think of the # as indicating nonsense.
The sentence “The sun is purple and blevledgekl;rz”, one might naturally think,
is neither true nor false because it is nonsense. It is nonsense even though it
has a part that isn’t nonsense.
3.3.4 Supervaluationism
Recall the guiding thought behind the strong Kleene tables: if a formula’s
classical truth values fix a particular truth value, then that is the value that the
formula takes on. There is a way to take this idea a step further, which results
in a new and interesting way of thinking about three-valued logic.
According to the strong Kleene tables, we get a classical truth value for
φ ψ, where is any connective, only when we have “enough classical
information” in the truth values of φ and ψ to fix a classical truth value for
φ ψ. Consider φ∧ψ for example: if either φ or ψ is false, then since
falsehood of a conjunct is classically sufficient for the falsehood of the whole
conjunction, the entire formula is false. But if, on the other hand, both φ and
ψ are #, then neither φ nor ψ has a classical truth value, we do not have enough
classical information to settle on a classical truth value for φ∧ψ, and so the
whole formula is #.
But now consider a special case of the situation just considered, where φ is
P , ψ is ∼P , and P is #. According to the strong Kleene tables, the conjunction
P ∧∼P is #, since it is the conjunction of two formulas that are #. But there is a
way of thinking about truth values of complex sentences according to which
the truth value ought to be 0, not #: no matter what classical truth value P were
to take on, the whole sentence P ∧∼P would be 0—therefore, one might think,
P ∧∼P ought to be 0. If P were 0 then P ∧∼P would be 0∧∼0—that is 0; and if
P were 1 then P ∧∼P would be 1∧∼1—0 again.
The general thought here is this: suppose a sentence φ contains some
sentence letters P1 . . . Pn that are #. If φ would be false no matter how we assign
classical truth values to P1 . . . Pn —that is, no matter how we precisified φ—then
φ is in fact false. Further, if φ would be true no matter how we precisified it,
then φ is in fact true. But if precisifying φ would sometimes make it true and
sometimes make it false, then φ in fact is #.
The idea here can be thought of as an extension of the idea behind the
strong Kleene tables. Consider a formula φ ψ, where is any connective.
If there is enough classical information in the truth values of φ and ψ to fix on a
particular classical truth value, then the strong Kleene tables assign φ ψ that
truth value. Our new idea goes further, and says: if there is enough classical
information within φ and ψ to fix a particular classical truth value, then φ ψ
gets that truth value. Information “within” φ and ψ includes, not only the
truth values of φ and ψ, but also a certain sort of information about sentence
letters that occur in both φ and ψ. For example, in P ∧∼P , when P is #, there
is insufficient classical information in the truth values of P and of ∼P to settle
on a truth value for the whole formula P ∧∼P (since each is #). But when we
look inside P and ∼P , we get more classical information: we can use the fact
that P occurs in each to reason as we did above: whenever we turn P to 0, we
turn ∼P to 1, and so P ∧∼P becomes 0; and whenever we turn P to 1 we turn
∼P to 0, and so again, P ∧∼P becomes 0.
This new idea—that a formula has a classical truth value iff every way of
precisifying it results in that truth value—is known as supervaluationism. Let us
lay out this idea formally.
Where I is a trivalent interpretation and C is a PL-interpretation (i.e., a
bivalent interpretation in the sense of section 2.3), say that C is a precisification
of I iff: whenever I assigns a sentence letter a classical truth value (i.e., 1 or 0),
C assigns that sentence letter the same classical value. Thus, precisifications
of I agree with I on the classical truth values, but in addition—being PL-
interpretations—they also assign classical truth values to sentence letters to

which I assigns #. Each precisification of I “decides” each of I ’s #s in some
way or other; different precisifications decide those #s in different ways.
We can now say how the supervaluationist assigns truth values to complex
formulas relative to a trivalent interpretation.
Definition of supervaluation: When φ is any wff and I is a trivalent inter-

pretation, the supervaluation of φ relative to I , is the function SI (φ) which
assigns 0, 1, or # to each wff as follows:

 1 if VC (φ) = 1 for every precisification, C , of I

SI (φ) = 0 if VC (φ) = 0 for every precisification, C , of I


# otherwise
Here VC is the valuation for PL-interpretation C , as defined in section 2.3.

Some common terminology: when SI (φ) = 1, we say that φ is supertrue in I ,
and when SI (φ) = 0, we say that φ is superfalse in I . For the supervaluationist,
a formula is true when it is supertrue (i.e., true in all precisifications of I ), false
when it is superfalse (i.e., false in all precisifications of I ), and # when it is
neither supertrue nor superfalse (i.e., when it is true in some precisifications of
I but false in others.)
Supervaluational notions of validity and semantic consequence may be
defined thus:
Supervaluational validity and consequence:

· φ is supervaluationally valid (“SV φ”) iff φ is supertrue in every trivalent
interpretation
· φ is a supervaluational semantic consequence of Γ (“Γ SV φ”) iff φ is
supertrue in each trivalent interpretation in which every member of Γ is
supertrue
To return to the example considered above: the supervaluationist assigns a

different truth value to P ∧∼P , when P is #, than do the strong Kleene tables
(and indeed, than do all the other tables we have considered.) The strong
Kleene tables say that P ∧∼P is # in this case. But the supervaluationist says
that it is 0: each precisification of any trivalent interpretation that assigns P #
is by definition a PL-interpretation, and P ∧∼P is 0 in each PL-interpretation.
Let us note a few facts about supervaluationism.

First, note that every tautology (PL-valid formula) turns out to be superval-
uationally valid. For let φ be a tautology; and consider any trivalent interpreta-
tion I , and any precisification C of I . Precisifications are PL-interpretations;
so, since φ is a tautology, φ is true in C .
Second, note that according to supervaluationism, some formulas are nei-
ther true nor false in some trivalent interpretations. For instance, take the
formula P ∧Q, in any trivalent interpretation I in which P is 1 and Q is #. Any
precisification of I must continue to assign P 1. But some precisifications of
I will assign 1 to Q, whereas others will assign 0 to Q. Any of the former
precisifications will assign 1 to P ∧Q, whereas any of the latter will assign 0 to
P ∧Q. Hence P ∧Q is neither supertrue nor superfalse in I : SI (P ∧Q) = #.
Finally, notice that the propositional connectives are not truth-functional
according to supervaluationism. To say that a connective is truth-functional,
given a certain definition of truth in an interpretation, is to say that the truth
value4 according to that definition of a complex statement whose major connec-
tive is is a function of the truth values of the immediate constituents of that
formula—that is, any two such formulas whose immediate constituents have
the same true values must themselves have the same truth value. According to
all of the truth tables for three-valued logic we considered earlier (Łukasiewicz,
Kleene strong and weak), the propositional connectives are truth-functional.
Indeed, this is in a way trivial: if a connective weren’t truth-functional according
to some definition of truth in an interpretation, then one couldn’t give that
connective a truth table at all; what a truth table does is specify how a connective
determines the truth value of entire sentences as a function of the truth values
of its parts. But supervaluationism renders the connectives not truth functional.
Consider the following pair of sentences, in a trivalent interpretation in which
P and Q are both #:
P ∧Q
P ∧∼P
As we have seen, in such trivalent interpretations, the first formula is # and

the second formula is 0 (since it is superfalse). But each of these formulas is
a conjunction, each of whose conjuncts is #: the truth values of φ and ψ do
not determine the truth value of φ∧ψ. So in supervaluationism, the ∧ isn’t
4
I am counting #, in addition to 1 and 0, as a “truth value”.
truth-functional. Similar arguments can be given to show that the → and the
∨ aren’t truth-functional either, given supervaluationism.
Exercise 3.7 Show that if a formula is true in a trivalent interpre-

tation given the strong Kleene truth tables, then it is supertrue in
that interpretation.
Exercise 3.8 Suppose that a wff φ has no repetitions of sentence

letters (i.e., each sentence letter occurs at most once in φ.) Show that
φ is not valid according to supervaluationism—that is, show that
it’s not the case that: for each trivalent interpretation I , SI (φ) = 1.
3.4 Intuitionism
Intuitionism in the philosophy of mathematics is a view according to which
there are no mind-independent mathematical facts. Rather, mathematical facts
and entities are mental constructs that owe their existence to the mental activity
of mathematicians constructing proofs. This philosophy of mathematics leads
intuitionists to a distinctive form of logic: intuitionist logic.
Let P be the statement: The sequence 0123456789 occurs somewhere in the
decimal expansion of π. How should we think about its meaning? For the classical
mathematician, the answer is straightforward. P is a statement about a part of
mathematical reality, namely, the infinite decimal expansion of π. Either the
sequence 0123456789 occurs somewhere in that expansion, in which case P is
true, or it does not, in which case P is false and ∼P is true.
For the intuitionist, this whole picture is mistaken, premised as it is on the
reality of an infinite decimal expansion of π. Our minds are finite, and so only
the finite initial segment of π’s decimal expansion that we have constructed so
far is real. The intuitionist’s alternate picture of P ’s meaning, and indeed of
meaning generally (for mathematical statements) is a radical one.5
The classical mathematician, comfortable with the idea of a realm of mind-
independent entities, thinks of meaning in terms of truth and falsity. As we saw,
she thinks of P as being true or false depending on the facts about π’s decimal
5
One intuitionist picture, anyway, on which see Dummett (1973). What follows is a crude
sketch. It does not do justice to the actual intuitionist position, which is, as they say, subtle.
expansion. Further, she explains the meanings of the propositional connectives

in truth-theoretic terms: a conjunction is true iff each of its conjuncts are true;
a negation is true iff the negated formula is false; and so on. Intuitionists, on
the other hand, reject the centrality of truth to meaning, since truth is tied
up with the rejected picture of mind-independent mathematical reality. For
them, the central semantic concept is that of proof. They simply do not think
in terms of truth and falsity; in matters of meaning, they think in terms of the
conditions under which formulas have been proved.
Take P , for example. Intuitionists advise us: don’t think in terms of what
it would take for P to be true. Think, rather, in terms of what it would take
to prove P . And the answer is clear: we would need to actually continue our
construction of the decimal expansion of π to a point where we found the
sequence 0123456789.
What, now, of ∼P ? Again, thinking in terms of proof, not truth: what
would it take for ∼P to be proved? The answer here is less straightforward.
Since P said that there exists a number of a certain sort, it was clear how it
would have to be proved: by actually exhibiting (calculating) some particular
number of that sort. But ∼P says that there is no number of a certain sort; how
do we prove something like that? The intuitionist’s answer: by proving that the
assumption that there is a number of that sort leads to a contradiction. In general, a
negation, ∼φ, is proved by proving that φ leads to a contradiction.6
Similarly for the other connectives: the intuitionist explicates their meanings
by their role in generating proof conditions, rather than truth conditions. φ∧ψ
is proved by separately giving a proof of φ and a proof of ψ; φ∨ψ is proved
by giving either a proof of φ or a proof of ψ; φ→ψ is proved by exhibiting a
construction whereby any proof of φ can be converted into a proof of ψ.
Likewise, the intuitionist thinks of logical consequence as the preservation
of provability, not the preservation of truth. For example, φ∧ψ logically implies
φ because if one has a proof of φ∧ψ, then one has a proof of φ; and conversely,
if one has proofs of φ and ψ separately, then one has the materials for a proof
of φ∧ψ. So far, so classical. But ∼∼φ does not logically imply φ, for the
intuitionist. Simply having a proof of ∼∼P —a proof that the assumption that
0123456789 occurs nowhere in π’s decimal expansion leads to a contradiction—
wouldn’t give us a proof of P , since proving P would require exhibiting a
particular place in π’s decimal expansion where 0123456789 occurs.
6
Given the contrast with the classical conception of negation, a different symbol (often
“¬”) is sometimes used for intuitionist negation.
Likewise, intuitionists do not accept the law of the excluded middle, φ∨∼φ,
as a logical truth. To be a logical truth, according to an intuitionist, a sentence
should be provable from no premises whatsoever. But to prove P ∨∼P , for
example, would require either exhibiting a case of 0123456789 in π’s decimal
expansion, or proving that the assumption that 0123456789 occurs in π’s decimal
expansion leads to a contradiction. We’re not in a position to do either.
Though we won’t consider intuitionist predicate logic, one of its most
striking features is easy to grasp informally. Intuitionists say that an existentially
quantified sentence is proved iff one of its instances has been proved. Therefore
they reject the inference from ∼∀xF x to ∃x∼F x, for one might be able to
prove a contradiction from the assumption of ∀xF x without being able to
prove any instance of ∃x∼F x.
We have so far been considering a putative philosophical justification for
intuitionist propositional logic. That justification has been rough and ready;
but intuitionist propositional logic itself is easy to present, perfectly precise,
and is a coherent system regardless of what one thinks of its philosophical
underpinnings. Two simple modifications to the natural deduction system of
section 2.4 generate a natural deduction system for intuitionistic propositional
logic. First, we need to split up the double-negation rule, DN, into two halves,
“double-negation introduction” and “double-negation elimination”:
Γ ` ∼∼φ Γ`φ
DNE DNI
Γ`φ Γ ` ∼∼φ
In our original system from section 2.4 we were allowed to use both DNE and
DNI; but in the intuitionist system, we are only allowed to use DNI; DNE is
not allowed. Second, to make up for the dropped rule DNE, our intuitionist
system adds the rule “ex falso”:
Γ ` φ∧∼φ
EF
Γ`ψ
Note that EF can be proved in the original system: simply use RAA and then
DNE. So, intuitionist logic results from a system for classical logic by simply
dropping one rule (DNE) and adding another rule that was previously provable
(EF). It follows that every intuitionistically provable sequent is also classically
provable (because every intuitionistic proof can be converted to a classical
proof).
Notice how dropping DNE blocks proofs of various classical theorems the
intuitionist wants to avoid. The proof of ∅ ` P ∨∼P in section 2.4, for instance,
used DNE. Of course, for all we’ve said so far, there might be some other way
to prove this sequent. Only when we have a semantics for intuitionistic logic,
and a soundness proof relative to that semantics, can we show that this sequent
cannot be proven without DNE. We will discuss a semantics for intuitionism
in section 7.2.
It is interesting to note that even though intuitionists reject the inference
from ∼∼P to P , they accept the inference from ∼∼∼P to ∼P , since its proof
only requires the half of DN that they accept, namely the inference from P to
∼∼P :
1 (1) ∼∼∼P As
2 (3) ∼∼P 2, DN (accepted version)
1,2 (4) ∼∼P ∧ ∼∼∼P 1,3 ∧I
1 (5) ∼P 4, RAA
Note that you can’t use this sort of proof to establish ∼∼P ` P . Given the way
RAA is stated, its application always results in a formula beginning with the ∼.
Chapter 4
Predicate Logic
et’s now turn from propositional logic to the “predicate calculus” (PC),
L as it is sometimes called. As with propositional logic, we’re going to
formalize predicate logic. We’ll first do grammar, and then move to semantics.
We won’t consider proof theory at all.1
4.1 Grammar of predicate logic

As before, we start by specifying the kinds of symbols that may be used in
sentences of predicate logic—primitive vocabulary—and then go on to define
the well formed formulas as strings of primitive vocabulary that have the right
form.
· logical: →, ∼, ∀
· nonlogical:
· for each n > 0, n-place predicates F , G . . ., with or without subscripts

· variables x, y . . . with or without subscripts
· individual constants (names) a, b . . ., with or without subscripts
· parentheses
1
Proof systems for predicate logic, in both axiomatic and natural-deduction form, are
straightforward, and can be found in standard logic textbooks.
84
CHAPTER 4. PREDICATE LOGIC 85
No symbol of one type is a symbol of any other type. Let’s call any variable or
constant a term.
Definition of wff:
i) if Π is an n-place predicate and α1 . . . αn are terms, then Πα1 . . . αn is a wff

ii) if φ, ψ are wffs, and α is a variable, then ∼φ, (φ→ψ), and ∀αφ are wffs
iii) nothing else is a wff
We’ll call formulas that are wffs in virtue of clause i) “atomic” formulas. When
a formula has no free variables, we’ll say that it is a closed formula, or sentence;
otherwise it is an open formula. (“Free” means that the variable doesn’t “belong”
to any quantifier in the formula. For example, in ∀yRxy, the variable x is free,
whereas the variable y is “bound” to the quantifier ∀y. ‘Free’ and ‘bound’ can
be precisely defined, but I won’t bother.)
We have the same defined logical terms: ∧, ∨, ↔. We also add the following
definition of the existential quantifier:
Definition of ∃: “∃vφ” is short for “∼∀α∼φ” (where α is a variable and φ is

a wff)
4.2 Semantics of predicate logic

Recall from section 2.2 the truth-conditional conception of semantics. Se-
mantics is about meaning; meaning is about the way truth is determined by
the world; and the way we represent the dependence of truth on the world in
logic consists of i) defining certain abstract configurations, which represent
different ways the world could be, and ii) defining the notion of truth for
formulas in these configurations. We thereby shed light on meaning, and we
are thereby able to define precise versions of the notions of logical truth and
logical consequence.
In propositional logic, the configurations were the PL-interpretations:
assignments of truth or falsity (1 or 0) to sentence letters; and valuation functions
defined truth in a configuration. This procedure needs to get more complicated
for predicate logic. The reason is that the method of truth tables assumes that
we can calculate the truth value of a complex formula by looking at the truth
values of its parts. But take the sentence ∃x(F x∧Gx). You can’t calculate its
truth value by looking at the truth values of F x and Gx, since sentences like
F x don’t have truth values at all. The variable ‘x’ doesn’t stand for any one
thing, and so ‘F x’ doesn’t have a truth value.
The solution to this problem is due to the Polish logician Alfred Tarski. It
begins with a new conception of a configuration, that of a model:
Definition of model: A PC-model is an ordered pair 〈D, I 〉 such that:

· D is a non-empty set (“the domain”)
· I is a function (“the interpretation function”) obeying the following
constraints:
· if α is a constant then I (α) ∈ D
· if Π is an n-place predicate, then I (Π) = some set of n-tuples of
members of D.
(Recall the notion of an n-tuple from section 1.8.)

Models, as we have defined them, seem like good ways to represent con-
figurations of the world. The domain, D, contains, intuitively, the individuals
that exist in the configuration. I , the interpretation function, tells us what the
non-logical constants (names and predicates) mean in the configuration. I
assigns to each name a member of the domain—its denotation. For example,
if the domain is the set of persons, then the name ‘a’ might be assigned me.
One-place predicates get assigned sets of 1-tuples of D—that is, just sets of
members of D. So, a one-place predicate ‘F ’ might get assigned a set of persons.
That set is called the “extension” of the predicate—if the extension is the set
of males, then the predicate ‘F ’ might be thought of as symbolizing “is male”.
Two-place predicates get assigned sets of ordered pairs of members of D—that
is, binary relations over the domain. If a two place predicate ‘R’ is assigned
the set of persons 〈u, v〉 such that u is taller than v, we might think of ‘R’ as
symbolizing “is taller than”. Similarly, three-place predicates get assigned sets
of ordered triples…
Relative to any PC-model 〈D, I 〉, we want to define the notion of truth in
a model—the corresponding valuation function. But we’ll need some apparatus
first. It’s pretty easy to see what truth value a sentence like F a should have. I
assigns a member of the domain to a—call that member u. I also assigns a
subset of the domain to F —let’s call that subset S. The sentence F a should be
true iff u ∈ S—that is, iff the referent of a is a member of the extension of F .
That is, F a should be true iff I (a) ∈ I (F ). Similarly, Rab should be true iff
〈I (a), I (b )〉 ∈ I (R). And so on.
As before, we can give recursive clauses for the truth values of negations
and conditionals. φ→ψ, for example, will be true iff either φ is false or ψ is
true.
But this becomes tricky when we try to specify the truth value of ∀xF x. It
should, intuitively, be true if and only if ‘F x’ is true, no matter what we put
in in place of ‘x’. But this is vague. Do we mean “whatever name (constant)
we put in place of ‘x”’? No, because we don’t want to assume that we’ve got a
name for everything in the domain, and what if F x is true for all the objects we
have names for, but false for one of the nameless things! Do we mean, “true no
matter what object from the domain we put in place of ‘x”’? No; objects from
the domain aren’t part of our primitive vocabulary, so the result of replacing ‘x’
with an object from the domain won’t be a formula!2
Tarski’s solution to this problem goes as follows. Initially, we don’t consider
truth values of formulas absolutely. Rather, we let the variables refer to certain
things in the domain temporarily. Then, we’ll say that ∀xF x will be true iff
for all objects u in the domain D: F x is true while x temporarily refers to u.
We implement this idea of temporary reference with the idea of a “variable
assignment”:
Definition of variable assignment: g is a variable assignment for model

〈D, I 〉 iff g is a function that assigns to each variable some object in D.
The variable assignments give the “temporary” meanings to the variables; when
g (x) = u, then u is the temporary denotation of x.
We need a further bit of notation. Let u be some object in D, let g be some
variable assignment, and let α be a variable. We then define “ g uα ” to be the
variable assignment that is just like g , except that it assigns u to α. (If g already
assigns u to α then g uα will be the same function as g .)
Note the following important fact about variable assignments: g uα , when
applied to α, must give the value u. (Work through the definitions to see that
this is so.) That is:
g uα (α) = u
One more bit of apparatus. Given any model M (= 〈D, I 〉), and given any
variable assignment, g , and given any term (i.e., variable or name) α, we define
2
Unless the domain happens to contain members of our primitive vocabulary!
the denotation of α, relative to M and g , “[α]M ,g ” as follows:
I (α) if α is a constant
(
[α]M , g =
g (α) if α is a variable
The subscripts M and g on [ ] indicate that denotations are assigned relative

to a model (M ), and relative to a variable assignment (g ).
Now we are ready to define the valuation function. The valuation function
will assign truth values to formulas relative to variable assignments. (Relativization
to assignments is necessary because, as we noticed before, F x doesn’t have a
truth value absolutely. It only has a truth value relative to an assigned value to
the variable x—i.e., relative to a choice of an arbitrary denotation for x.)
Definition of valuation: The PC-valuation function, VM , g , for PC-model

M (= 〈D, I 〉) and variable assignment g , is defined as the function that assigns
to each wff either 0 or 1 subject to the following constraints:
i) for any n-place predicate Π and any terms α1 . . . αn , VM ,g (Πα1 . . . αn ) = 1

iff 〈[α1 ]M ,g . . . [αn ]M , g 〉 ∈ I (Π)
ii) for any wffs φ, ψ, and any variable α:
VM ,g (∼φ) = 1 iff VM ,g (φ) = 0

VM ,g (φ→ψ) = 1 iff either VM , g (φ) = 0 or VM ,g (ψ) = 1
VM , g (∀αφ) = 1 iff for every u ∈ D, VM , g α (φ) = 1
u
(In understanding clause i), recall that the one tuple containing just u, 〈u〉 is
just u itself. Thus, in the case where Π is F , some one place predicate, clause i)
says that VM ,g (F α) = 1 iff [α]M ,g ∈ I (F ).)
So far we have defined the notion of truth in a model relative to a variable
assignment. But what we really want is a notion of truth in a model, period—that
is, absolute truth in a model. (We want this because we want to define, e.g., a
valid formula as one that is true in all models.) So, let’s define absolute truth in
a model in this way:
Definition of truth in a model: φ is true in PC-model M iff VM ,g (φ) = 1,

for each variable assignment g
It might seem that this is too strict a requirement—why must φ be true relative
to each variable assignment? But in fact, it’s not too strict at all. The kinds of
formulas we’re really interested in are formulas without free variables (we’re
interested in formulas like F a, ∀xF x, ∀x(F x→Gx); not formulas like F x,
∀x Rxy, etc.) And if a formula has no free variables, then if there’s even a single
variable assignment relative to which it is true, then it is true relative to every
variable assignment. (And so, we could just as well have defined truth in a
model as truth relative to some variable assignment.) I won’t prove this fact,
but it’s not too hard to prove; one would simply need to prove (by induction)
that, for any wff φ and model M , if variable assignments g and h agree on all
variables free in φ, then VM , g (φ) = VM ,h (φ).
Now we can give semantic definitions of the core logical notions:
Definition of validity: φ is PC-valid (“PC φ”) iff φ is true in all PC-models
Definition of semantic consequence: φ is a PC-semantic consequence of

set of wffs Γ (“Γ PC φ”) iff for every PC-model M , if each member of Γ is
true in M then φ is also true in M
Since our new definition of the valuation function treats the propositional
connectives → and ∼ in the same way as the propositional logic valuation did,
it’s easy to see that it also treats the defined connectives ∧, ∨, and ↔ in the
same way:
VM ,g (φ∧ψ) = 1 iff VM ,g (φ) = 1 and VM ,g (ψ) = 1
VM ,g (φ∨ψ) = 1 iff VM ,g (φ) = 1 or VM , g (ψ) = 1
VM ,g (φ↔ψ) = 1 iff VM ,g (φ) = VM ,g (ψ)
Moreover, we can also prove that the valuation function treats ∃ as it should
(given its intended meaning):
VM ,g (∃αφ) = 1 iff there is some u ∈ D such that VM ,g α (φ) = 1
u
This can be established as follows. The definition of ∃αφ is: ∼∀α∼φ. So, we
must show that for any model, and any variable assignment g based on that
model, VM ,g (∼∀α∼φ) = 1 iff there is some u ∈ D such that VM ,g α (φ) = 1. (In
u
arguments like these, I’ll sometimes stop writing the subscript M in order to
reduce clutter. It should be obvious from the context what the relevant model
is.) Here’s the argument:
· Vg (∼∀α∼φ) = 1 iff Vg (∀α∼φ) = 0 (given the clause for ∼ in the defini-

tion of the valuation function)
· But, Vg (∀α∼φ) = 0 iff for some u ∈ D, Vg α (∼φ) = 0

u
· Given the clause for ∼, this can be rewritten as: “… iff for some u ∈ D,
Vg α (φ) = 1”
u
4.3 Establishing validity and invalidity

Given our definitions, we can establish that particular formulas are valid.
Example 4.1: Show that ∀xF x→F a is valid. That is, show that for any
model 〈D, I 〉, and any variable assignment g , Vg (∀xF x→F a) = 1:
i) Suppose otherwise; then Vg (∀xF x) = 1 and Vg (F a) = 0.
ii) Given the latter, that means that [a] g ∈

/ I (F )—that is, I (a) ∈
/ I (F ).
iii) Given the former, for any u ∈ D, Vg x (F x) = 1.

u
iv) I (a) ∈ D, and so Vg x (F x) = 1.

I (a)
v) By the truth condition for atomics, [x] g x ∈ I (F ).

I (a)
vi) By the definition of the denotation of a variable, [x] g x = gIx (a) (x)
I (a)
vii) but gIx (a) (x) = I (a). Thus, I (a) ∈ I (F ). Contradiction
The claim in step iv) that I (a) ∈ D comes from the definition of an inter-
pretation function: the interpretation of a constant is always a member of the
domain. Notice that “I (a)” is a term of our metalanguage; that’s why, when
I’m given that “for any u ∈ D” in step iii), I can set u equal to I (a) to obtain
step iv).
Example 4.2: Show that ∀x∀yRxy→∀xRx x:
i) Suppose for reductio that Vg (∀x∀yRxy→∀xRx x) = 0 (for some assign-

ment g in some model). Then Vg (∀x∀yRxy) = 1 and …
ii) …Vg (∀xRx x) = 0
iii) Given ii), for some u ∈ D, Vg x (Rx x) = 0, and so 〈[x] g x , [x] g x 〉 ∈

/ I (R)
u u u
iv) [x] g x is g xu (x), i.e., u. So 〈u, u〉 ∈

/ I (R)
u
v) Given i), for every member of D, and so for u in particular, Vg x (∀yRxy) =

u
1
vi) given v), for every member of D, and so for u in particular, Vg xy (Rxy) = 1
uu
vii) given vi), 〈[x] g xy , [y] g xy 〉 ∈ I (R)

uu uu
viii) But [x] g xy and [y] g xy are each just u. Hence 〈u, u〉 ∈ I (R), contradicting
uu uu
iv).
Exercise 4.1 Show that:
a) ∀x(F x→(F x∨Gx))
b) ∀x(F x∧Gx)→(∀xF x∧∀xGx)
c) ∀x(F x→Gx), ∀x(Gx→H x) ∀x(F x→H x)
d) ∃x∀yRxy→∀y∃xRxy
We’ve seen how to establish that particular formulas are valid. How do
we show that a formula is invalid? We need to simply exhibit a single model
in which the formula is false. (The definition of validity specifies that a valid
formula is true in all models; therefore, it only takes one model in which a
formula is false to make that formula invalid.) So let’s take one example; let’s
show that the formula (∃xF x∧∃xGx)→∃x(F x∧Gx) isn’t valid. To do this, we
must produce a model in which this formula is false. All we need is a single
model, since in order for the formula to be valid, it must be true in all models.
My model will contain letters in its domain:
D = {u, v}
I (F ) = {u}
I (G) = {v}
It is intuitively clear that the formula is false in this model. In this model,
something has F (namely, u), and something has G (namely, v), but nothing
has both.
One further example: let’s show that ∀x∃yRxy 2 ∃y∀xRxy. We must show
that the first formula does not semantically imply the second. So we must come
up with a model in which the first formula is true and the second is false. It helps
to think about natural language sentences that these formulas might represent.
If R symbolizes “respects”, then the first formula says that “everyone respects
someone or other”, and the second says that “there is someone whom everyone
respects”. Clearly, the first can be true while the second is false: suppose that
each person respects a different person, so that no one person is respected by
everyone. A simple case of this occurs when there are just two people, each of
whom respects the other, but neither of whom respects him/herself:
(
•h •
Here is a model based on this idea:
D = {u, v}
I (R) = {〈u, v〉, 〈v, u〉}
Exercise 4.2 Show that
a) 2 ∀x(F x→Gx)→∀x(Gx→F x)
b) 2 ∀x(F x∨∼Gx)→(∀xF x∨∼∃xGx)
c) Rab 2 ∃xRx x
d) ∀x∀y∀z[(Rxy∧Ry z)→Rx z], ∀x∃yRxy 2 ∃xRx x

Chapter 5
Extensions of Predicate Logic
he predicate logic we considered in the previous chapter is powerful.

T Much natural language discourse can be represented using it, in a way
that reveals logical structure. Nevertheless, it has its limitations. In this chapter
we consider some of its limitations, and corresponding additions to predicate
logic.
5.1 Identity
“Standard” predicate logic is usually taken to include the identity sign (“=”).
“a=b ” means that a and b are one and the same thing.
5.1.1 Grammar for the identity sign

We first need to expand our grammar of predicate logic to allow for the new
symbol =. Two changes are needed. First, we need to add = to the primitive
vocabulary of predicate logic. Then we need to the following clause to the
definition of a well-formed formula:
· If α and β are terms, then α=β is a wff
We need to beware of a potential source of confusion. We’re now using the

symbol ‘=’ as the object-language symbol for identity. But I’ve also been using
‘=’ as the metalanguage symbol for identity, for instance when I write things
93
CHAPTER 5. EXTENSIONS OF PREDICATE LOGIC 94
like “V(φ) = 1”. This shouldn’t generally cause confusion, but if there’s a
danger of misunderstanding, I’ll clarify by writing things like: “…= (i.e., is the
same object as)…”.
5.1.2 Semantics for the identity sign

This is easy. We keep the notion of a PC-model from the last chapter, and
simply add to our definition of truth-in-a-PC-model. All we need to add is a
clause to the definition of a valuation function telling it what truth values to
give to sentences containing the = sign. Here is the clause:
VM ,g (α=β) = 1 iff: [α]M ,g = (i.e., is the same object as) [β]M ,g
That is, the sentence α = β is true iff the terms α and β refer to the same
object.
Example 5.1: Show that the formula ∀x∃y x=y is valid. We need to show
that in any model, and any variable assignment g in that model, Vg (∀x∃y x=y) =
1. So:
i) So, suppose for reductio that for some g in some model, Vg (∀x∃y x=y) =
0.
ii) Given the clause for ∀, for some object in the domain, call it “u”, Vg x (∃y x=y) =
u
0.
iii) Given the clause for ∃, for every v in the domain, Vgu/x v/y (x=y) =0.
iv) Letting v in iii) be u, we have: Vg xy (x=y) = 0.

uu
v) So, given the clause for “=”, [x] g xy is not the same object as [y] g xy
uu uu
vi) but [x] g xy and [y] g xy are the same object. [x] g xy is g xuyu (x), i.e., u; and
uu uu uu
[y] g xy is g xuyu (y), i.e., u.
uu
Exercise 5.1 Demonstrate each of the following:
a) F ab ∀x(x=a→F x b )
b) ∃x∃y∃z(F x∧F y∧F z∧x6=y∧x6= z∧y6= z),

∀x(F x→(Gx∨H x) 2 ∃x∃y∃z(Gx∧Gy∧Gz ∧ x6=y∧x6= z∧y6= z)
5.1.3 Symbolizations with the identity sign

Why do we ever add anything to our list of logical constants? Why not stick
with the tried and true logical constants of propositional and predicate logic?
We generally add a logical constant when it has a distinctive inferential and
semantic role, and when it has very general application—when, that is, it occurs
in a wide range of linguistic contexts. We studied the distinctive semantic role
of ‘=’ in the previous section. In this section, we’ll look at the range of linguistic
contexts that can be symbolized using ‘=’.
The most obvious sentences that may be symbolized with ‘=’ are those
that explicitly concern identity, such as “Mark Twain is identical to Samuel
Clemens”:
t =c
and “Every man fails to be identical to George Sand”:
∀x(M x→∼x=s)
(It will be convenient to abbreviate ∼α=β as α6=β. Thus, the second symbol-
ization can be rewritten as: ∀x(M x→x6= s).) But many other sentences involve
the concept of identity in subtler ways.
Consider, for example, “Every lawyer hates every other lawyer”. The ‘other’
signifies nonidentity; we have, therefore:
∀x(Lx→∀y[(Ly∧x6=y)→H xy])
Consider next “Only Ted can change grades”. This means: “no one other than
Ted can change grades”, and may therefore be symbolized as:
∼∃x(x6= t ∧C x)
(letting ‘C x’ symbolize “x can change grades”.)

Another interesting class of sentences concerns number. We cannot sym-
bolize “There are at least two dinosaurs” as: “∃x∃y(D x∧D y)”, since this would
be true even if there were only one dinosaur: x and y could be assigned the
same dinosaur. The identity sign to the rescue:
∃x∃y(D x∧D y ∧ x6=y)

This says that there are two different objects, x and y, each of which are di-
nosaurs. To say “There are at least three dinosaurs” we say:
∃x∃y∃z(D x∧D y∧D z∧ x6=y ∧ x6= z ∧ y6= z)
Indeed, for any n, one can construct a sentence φn that symbolizes “there are
at least n F s”:
φn : ∃x1 . . . ∃xn (F x1 ∧ · · · ∧F xn ∧ δ)
where δ is the conjunction of all sentences “xi 6= x j ” where i and j are integers
between 1 and n (inclusive) and i < j . (The sentence δ says in effect that no
two of the variables x1 . . . xn stand for the same object.)
Since we can construct each φn , we can symbolize other sentences involving
number as well. To say that there are at most n F s, we write: ∼φn+1 . To say
that there are between n and m F s (where m > n), we write: φn ∧∼φ m+1 . To
say that there are exactly n F s, we write: φn ∧∼φn+1 .
These methods for constructing sentences involving number will always
work; but one can often construct shorter numerical symbolizations by other
methods. For example, to say “there are exactly two dinosaurs”, instead of
saying “there are at least two dinosaurs, and it’s not the case that there are at
least three dinosaurs”, we could say instead:
∃x∃y(D x∧D y ∧ x6=y ∧ ∀z[D z→(z=x∨z=y)])
Exercise 5.2 Symbolize each of the following, using predicate

logic with identity.
a) Everyone who loves someone else loves everyone
b) The only truly great player who plays in the NBA is Allen
Iverson
c) If a person shares a solitary confinement cell with a guard,

then they are the only people in the cell
d) There are at least five dinosaurs (What is the shortest sym-

bolization you can find?)
5.2 Function symbols

A singular term, such as ‘Ted’, ‘New York City’, ‘George W. Bush’s father’,
or ‘the sum of 1 and 2’, is a term that purports to refer to a single entity.
Notice that some of these have semantically significant structure. ‘George
W. Bush’s father’, for example, means what it does because of the meaning
of ‘George W. Bush’ and the meaning of ‘father’. But standard predicate
logic’s only (constant) singular terms are its names: a, b , c . . . , which do not
have semantically significant parts. Thus, using predicate logic’s names to
symbolize semantically complex English singular terms leads to an inadequate
representation.
Suppose, for example, that we give the following symbolizations:
“3 is the sum of 1 and 2”: a = b
“George W. Bush’s father was a politician”: P c
By symbolizing ‘the sum of 1 and 2’ as simply ‘b ’, the first symbolization ignores

the fact that ‘1’, 2’, and ‘sum’ are semantically significant constituents of ‘the
sum of 1 and 2’; and by symbolizing “George W. Bush’s father” as ‘c’, we ignore
the semantically significant occurrences of ‘George W. Bush’ and ‘father’. This
is a bad idea. We ought, rather, to produce symbolizations of these terms
that take account of their semantic complexity. The symbolizations ought to
account for the distinctive logical behavior of sentences containing the complex
terms. For example, the sentence “George W. Bush’s father was a politician”
logically implies the sentence “Someone’s father was a politician”. This ought
to be reflected in the symbolizations; the first sentence’s symbolization ought
to semantically imply the second sentence’s symbolization.
One way of doing this is via an extension of predicate logic: we add function
symbols to its primitive vocabulary. Think of “George W. Bush’s father” as
the result of plugging “George W. Bush” into the blank in “ ’s father”. “ ’s
father” is an English function symbol. Function symbols are like predicates in
some ways. The predicate “ is happy” has a blank in it, in which you can put
a name. “ ’s father” is similar in that you can put a name into its blank. But
there is a difference: when you put a name into the blank of a predicate, you
get a complete sentence, whereas when you put a name into the blank of “ ’s
father”, you get a noun phrase, such as “George W. Bush’s father”.
Corresponding to English function symbols, we’ll add logical function

symbols. We’ll symbolize “ ’s father” as f ( ). We can put names into the
blank here. Thus, we’ll symbolize “George W. Bush’s father” as “ f (a)”, where
“a” stands for “George W. Bush”.
We need to add two more complications. First, what goes into the blank
doesn’t have to be a name—it could be something that itself contains a function
symbol. E.g., in English you can say: “George W. Bush’s father’s father”. We’d
symbolize this as: f ( f (a)).
Second, just as we have multi-place predicates, we have multi-place function
symbols. “The sum of 1 and 2” contains the function symbol “the sum of
and —”. When you fill in the blanks with the names “1” and “2”, you get the
noun phrase “the sum of 1 and 2”. So, we symbolize this using the two-place
function symbol, “s( ,—). If we let “a” symbolize “1” and “b ” symbolize “2”,
then “the sum of 1 and 2” becomes: s (a, b ).
The result of plugging names into function symbols in English is a noun
phrase. Noun phrases combine with predicates to form complete sentences.
Function symbols function analogously in logic. Once you combine a function
symbol with a name, you can take the whole thing, apply a predicate to it, and
get a complete sentence. Thus, the sentence “George W. Bush’s father was a
politician” becomes:
P f (a)
And “3 is the sum of 1 and 2” becomes:
c = s(a, b )
(here “c” symbolizes “3”). We can put variables into the blanks of function
symbols, too. Thus, we can symbolize “Someone’s father was a politician” as
∃xP f (x)
Example 5.2: Symbolize the following sentences using predicate logic with
identity and function symbols:
Everyone loves his or her father

∀xLx f (x)
No one’s father is also his or her mother
∼∃x f (x)=m(x)
No one is his or her own father

∼∃x x= f (x)
A person’s maternal grandfather hates that person’s pa-
ternal grandmother
∀x H f (m(x)) m( f (x))
Every even number is the sum of two prime numbers
∀x(E x→∃y∃z(P y∧P z∧x=s(y, z)))

logic with identity and function symbols.
a) The product of an even number and an odd number is an

even number.
b) If the square of a number that is divisible by each smaller

number is odd, then that number is greater than all numbers. (I
know, the sentence is silly.)
5.2.1 Grammar for function symbols

We need to update our definition of a wff to allow for function symbols. First,
we need to add to our vocabulary. So, the new definition starts like this (the
new bit is in boldface):
· logical: →, ∼, ∀, =
· nonlogical:

· for each n > 0, n-place function symbols f , g ,…, with or with-
out subscripts
· parentheses
The definition of a wff, actually, stays the same; all that needs to change is the
definition of a “term”. Before, terms were just names or variables. Now, we
need to allow for f (a), f ( f (a)), etc., to be terms. This is done by the following
recursive definition of a term:1
Definition of terms:
· names and variables are terms

· if f is an n-place function symbol, and α1 . . . αn are terms, then
f (α1 . . . αn ) is a term
· nothing else is a term
5.2.2 Semantics for function symbols

We now need to update our definition of a PC-model by saying what the
interpretation of a function symbol is. That’s easy: the interpretation of an
n-place function symbol ought to be an n-place function defined on the model’s
domain—i.e., a rule that maps any n members of the model’s domain to another
member of the model’s domain. For example, in a model in which the one-place
function symbol f ( ) is to represent “ ’s father”, the interpretation of f will
be the function that assigns to any member of the domain that object’s father.
Here’s the new general definition of a model (a “PC+FS-model”, for “predicate
calculus plus function symbols”):
Definition of model: A PC+FS-model is an ordered pair 〈D, I 〉 such that:
· D is a non-empty set (“the domain”)

· I is a function (“the interpretation function”) obeying the following
constraints:

members of D.
1
Note that complex terms formed from function symbols with more than one place do not,
officially, contain commas. But I’ll informally include the commas to improve readability. I
will write, for example, f (x, y) instead of f (x y).
· If f is an n-place function symbol, then I (f ) is an n-place

(total) function defined on D.
(“Total” simply means that the function must yield an output for any n members
of D.)
The definition of a valuation function stays the same; all we need to do is
update the definition of denotation to accommodate our new complex terms.
Since we now can have arbitrarily long terms (not just names or variables), we
need a recursive definition:
Definition of denotation: For any model M (= 〈D, I 〉), variable assignment
g , and term α, the denotation of α relative to M and g , [α]M ,g , is defined as
follows:

 I (α) if α is a constant

[α]M ,g = g (α) if α is a variable


I (f )([α1 ]M ,g . . . [αn ]M ,g ) if α is a complex term f (α1 . . . αn )
Note the recursive nature of this definition: the denotation of a complex term
is defined in terms of the denotations of its smaller parts. Let’s think carefully
about what the final clause says. It says that, in order to calculate the denotation
of the complex term f (α1 . . . αn ) (relative to assignment g ), we must first figure
out what I ( f ) is—that is, what the interpretation function I assigns to the
function symbol f . This object, the new definition of a model tells us, is an
n-place function on the domain. We then take this function, I ( f ), and apply
it to n arguments: namely, the denotations (relative to g ) of the terms α1 . . . αn .
The result is our desired denotation of f (α1 . . . αn ).
It may help to think about a simple case. Suppose that f is a one-place
function symbol; suppose our domain consists of the set of natural numbers;
suppose that the name a denotes the number 3 in this model (i.e., I (a) = 3),
and suppose that f denotes the successor function (i.e., I ( f ) is the function,
successor, that assigns to any natural number n the number n + 1.) In that case,
the definition tells us that:
[ f (a)] g = I ( f )([a] g )
= I ( f )(I (a))
= successor(3)
=4
Example 5.3: Here’s a sample metalanguage argument that makes use of

the new definitions. As mentioned earlier, ‘George W. Bush’s father was a
politician’ logically implies ‘ ‘Someone’s father was a politician’. Let’s show that
these sentences’ symbolizations stand in the relation of semantic implication.
That is, let’s show that P f (c) ∃xP f (x)—i.e., that in any model in which
P f (c) is true, ∃xP f (x) is true:
i) Suppose that P f (c) is true in a model 〈I , D〉—i.e., Vg (P f (c)) = 1 (where

V is the valuation for this model), for each variable assignment g .
ii) Suppose for reductio that ∃xP f (x) is false in this model; i.e., for some
variable assignment g , Vg (∃xP f (x)) = 0
iii) By line i), Vg (P f (c)) = 1, and so [ f (c)] g ∈ I (P ). [ f (c)] g is just I ( f )([c] g ),

and [c] g is just I (c). So I ( f )(I (c)) ∈ I (P ).
iv) By ii), for every object u ∈ D,Vg x (P f (x)) = 0.

u
v) I (c) ∈ D. So, by line iv), Vg x (P f (x)) = 0, and hence, [ f (x)] g x ∈

/ I (P )
I (c)) I (c)
vi) [ f (x)] g x is just I ( f )([x] g x ), and [x] g x is just gI (c)x (x)—i.e., I (c).
I (c) I (c)) I (c)
vii) So I ( f )(I (c)) ∈

/ I (P ), which contradicts line iii)
Exercise 5.4 Demonstrate each of the following:
a) ∀xF x→F f (a)
b) {∀x f (x)6= x} 2 ∃x∃y( f (x)=y ∧ f (y)=x)
5.3 Definite descriptions

Our logic has gotten more powerful with the addition of function symbols,
but it still isn’t perfect. Function symbols let us “break up” certain complex
singular terms—e.g., “Bush’s father”. But there are others we still can’t break
up—e.g., “The black cat”. Even with function symbols, the only candidate for
a direct symbolization of this phrase into the language of predicate logic is a

simple name, “a” for example. But this symbolization ignores the fact that “the
black cat” contains “black” and “cat” as semantically significant constituents.
It therefore fails to provide a good model of this term’s distinctively logical
behavior. For example, ‘The black cat is happy’ logically implies ‘Some cat
is happy’. But the simple-minded symbolization of the first sentence, H a,
obviously does not semantically imply the obvious symbolization of the second:
∃x(C x∧H x).
One response is to introduce another extension of predicate logic. We
introduce a new symbol, ι, to stand for “the”. The grammatical function of
“the” in English is to turn predicates into noun phrases. “Black cat” is a predicate
of English; “the black cat” is a noun phrase that refers to the thing that satisfies
the predicate “black cat”. Similarly, in logic, given a predicate F , we’ll let ιxF x
be a term that means: the thing that is F .
We’ll want to let ιx attach to complex predicates, not just simple predi-
cates. To symbolize “the black cat”—i.e., the thing that is both black and a
cat—we want to write: ιx(B x∧C x). In fact, we’ll let ιx attach to wffs with
arbitrary complexity. To symbolize “the fireman who saved someone”, we’ll
write: ιx(F x∧∃yS xy).
5.3.1 Grammar for ι

Just as with function symbols, we need to add a bit to the primitive vocabulary,
and revise the definition of a term.
· logical: →, ∼, ∀, =, ι
· nonlogical:
· for each n > 0, n-place function symbols f , g ,…, with or without
subscripts
· parentheses
Definition of terms and wffs:

i) names and variables are terms

ii) if φ is a wff and α is a variable then ιαφ is a term
iii) if f is an n-place function symbol, and α1 . . . αn are terms, then f (α1 . . . αn )
is a term
iv) if Π is an n-place predicate and α1 . . . αn are terms, then Πα1 . . . αn is a wff
v) If α and β are terms, then α=β is a wff
vi) if φ, ψ are wffs, and α is a variable, then ∼φ, (φ→ψ), and ∀αφ are wffs
vii) nothing else is a wff or term
Notice how we needed to combine the recursive definitions of term and wff
into a single recursive definition of wffs and terms together. The reason is that
we need the notion of a wff to define what counts as a term containing the ι
operator (clause ii); but we need the notion of a term to define what counts as
a wff (clause iv). The way we accomplish this is not circular. The reason it isn’t
is that we can always decide, using these rules, whether a given string counts as
a wff or term by looking at whether smaller strings count as wffs or terms. And
the smallest strings are said to be wffs or terms in non-circular ways.
5.3.2 Semantics for ι

We need to update the definition of denotation so that ιxφ will denote the one
and only thing in the domain that is φ. This is a little tricky, though. What is
there is no such thing? Suppose that ‘K’ symbolizes “king of” and ‘a’ symbolizes
“USA”. Then, what should ‘ιxK xa’ denote? It is trying to denote the king of
the USA, but there is no such thing. Further, what if more than one thing
satisfies the predicate? In short, what do we say about “empty descriptions”?
One approach would be to say that every atomic sentence with an empty
description is false. One way to do this is to include in each model an “emptiness
marker”, E , which is an object we assign as the denotation for each empty de-
scription. The emptiness marker shouldn’t be thought of as a “real” denotation;
when we assign it as the denotation of a description, this just marks the fact that
the description has no real denotation. We will stipulate that the emptiness
marker is not in the domain; this ensures that it is not in the extension of any
predicate, and hence that atomic sentences containing empty descriptions are
always false. Here’s how the semantics looks (“PC+DD”—“predicate calculus
plus definite descriptions”):
Definition of model: A PC+DD-Model is an ordered triple 〈D, I , E 〉 such

that:
· D is a non-empty set
/D
· E∈
· I is a function obeying the following constraints:
members of D
· If f is an n-place function symbol, then I ( f ) is some n-place total
function on D ∪ {E } that maps u1 . . . un to a member of D if each
ui ∈ D and otherwise maps them to E
Definition of denotation and valuation: The denotation and valuation

functions, [ ]M ,g and VM ,g , for PC+DD-model M (=〈D, I 〉) and variable as-
signment g , are defined as the functions that satisfy the following constraints:
i) VM , g assigns to each wff either 0 or 1

ii) For any term α, [α]M , g is:
· I (α) if α is a constant
· g (α) if α is a variable
· I ( f )([α1 ]M ,g . . . [αn ]M ,g ) if α is a complex term f (α1 . . . αn )
· the unique u ∈ D such that V g β (φ) = 1 if α is a complex term ιβφ
u
and there is a unique such u
· E if α is a complex term ιβφ and there is no such u
iii) for any n-place predicate Π and any terms α1 . . . αn ,VM ,g (Πα1 . . . αn ) = 1
iff 〈[α1 ]M ,g . . . [αn ]M , g 〉 ∈ I (Π)
iv) VM , g (α=β) = 1 iff: [α]M ,g = (i.e., is the same object as) [β]M ,g
v) for any wffs φ, ψ, and any variable α:
VM ,g (∼φ) = 1 iff VM ,g (φ) = 0

VM ,g (φ→ψ) = 1 iff either VM , g (φ) = 0 or VM ,g (ψ) = 1
VM , g (∀αφ) = 1 iff for every u ∈ D, VM , g α (φ) = 1
u
As with the grammar, we need to mix together the definition of denotation

and the definition of the valuation function. The reason is that we need to
define the denotation of definite descriptions using the valuation function (in
clause ii), but we need to define the valuation function using the concept of
denotation (in clauses iii and iv). As before, this is not circular.
Note a few things about these definitions. First, note that in the definition
of a model, function symbols denote functions that “stay within D”. That is, if
you feed one of these functions an n-tuple consisting only of members of D,
it spits out another member of D. But if you feed it an n-tuple, one of whose
members is the emptiness marker E , then it spits back out E . Second, note
that the denotation of any term is either E , or a member of D (exercise 5.6).
Third, since E is not in the domain, and extensions of predicates are defined to
be sets of n-tuples drawn from the domain, it follows that E cannot be present
in predicate extensions.
An alternate approach to using the emptiness marker E would appeal to
three-valued logic. We could leave the denotation of ιxφ undefined if there is
no object in the domain such that φ. We could then treat any atomic sentence
that contains a denotationless term as being neither true nor false—i.e., #. We
would then need to update the other clauses to allow for #s, using one of the
three-valued approaches to propositional logic from chapter ??. I won’t pursue
this option further.
Exercise 5.5 Establish the following:
a) ∀xL(x, ιyF xy)→∀x∃yLxy
b) 2 GιxF x→F ιxGx
Exercise 5.6 Show that the denotation of any term is either E , or

a member of D.
5.3.3 Eliminability of function symbols and definite descrip-

tions
In a sense, we don’t really need function symbols or the ι. Let’s return to
the English singular term ‘the black cat’. Introducing the ι gave us a way
to symbolize this singular term in a way that takes into account its semantic
structure (namely: ιx(B x∧C x).) But even without the ι, there is a way to
symbolize whole sentences containing ‘the black cat’, using just standard predicate
plus identity. We could, for example, symbolize “The black cat is happy” as:
∃x[ (B x∧C x) ∧ ∀y[(B y∧C y)→y=x] ∧ H x]
That is, “there is something such that: i) it is a black cat, ii) nothing else is a
black cat, and iii) it is happy”.
This method for symbolizing sentences containing ‘the’ is called “Russell’s
theory of descriptions”, in honor of its inventor Bertrand Russell, the 19th and
20th century philosopher and logician.2 The general idea is to symbolize: “the
φ is ψ” as ∃x[φ(x) ∧ ∀y(φ(y)→x=y) ∧ ψ(x)]. This method can be iterated so
as to apply to sentences with two or more definite descriptions, such as “The
8-foot tall man drove the 20-foot long limousine”, which becomes, letting ‘E’
stand for ‘is eight feet tall’ and ‘T ’ stand for ‘is twenty feet long’:
∃x[E x∧M x ∧ ∀z([E z∧M z]→x=z) ∧

∃y[T y∧Ly ∧ ∀z([T z∧Lz]→y=z) ∧ D xy]]
An interesting problem arises with negations of sentences involving definite

descriptions, when we use Russell’s method. Consider “The president is not
bald”. Does this “The president is such that he’s non-bald”, which is symbolized
as follows:
∃x[P x∧∀y(P y→x=y)∧∼B x]
? Or does it mean “It is not the case that the President is bald”, which is
symbolized thus:
∼∃x[P x∧∀y(P y→x=y)∧B x]
? According to Russell, the original sentence is simply ambiguous. Symbolizing

it the first way is called “giving the description wide scope (relative to the ∼)”,
since the ∼ is inside the scope of the ∃x. Symbolizing it in the second way is
called “giving the description narrow scope (relative to the ∼)”, because the ∃x
is inside the scope of the ∼.
What is the difference in meaning between these two symbolizations? The
first says that there really is a unique president, and adds that he is not bald.
2
See Russell (1905).
So the first implies that there’s a unique president. The second merely denies
that there is a unique president, who is bald. That doesn’t imply that there’s
a unique president. It would be true if there’s a unique president who is not
bald, but it would also be true in two other cases: the case in which there are
no presidents at all, and the case in which there is more than one president.
A similar issue arises with the sentence “The round square does not exist”.
We might think to symbolize it:
∃x[Rx∧S x∧∀y([Ry∧S y]→x=y) ∧ ∼E x]
letting “E” stands for “exists”. In other words, we might give the description
wide scope. But this is wrong, because it says there is a certain round square that
doesn’t exist, and that’s a contradiction. This way of symbolizing the sentence
corresponds to reading the sentence as saying “The thing that is a round square
is such that it does not exist”. But that isn’t the most natural way to read the
sentence. The sentence would usually be interpreted to mean: “It is not true
that the round square exists”, —that is, as the negation of “the round square
exists”:
∼∃x[Rx∧S x∧∀y([Ry∧S y]→x=y) ∧ E x]
with the ∼ out in front. Here we’ve given the description narrow scope. Notice
also that saying that x exists at the end is redundant, so we could simplify to:
∼∃x[Rx∧S x∧∀y([Ry∧S y]→x=y)]
Again, notice the moral of these last two examples: if a definite description
occurs in a sentence with a ‘not’, the sentence may be ambiguous: does the
‘not’ apply to the entire rest of the sentence, or merely to the predicate?
If we are willing to use Russell’s method for translating definite descriptions,
we can drop ι from our language. We would, in effect, not be treating “the F ”
as a referring phrase. We would instead be paraphrasing sentences that contain
“the F ” into sentences that don’t. “The black cat is happy” got paraphrased
as: “there is something that is a black cat, is such that nothing else is a black
cat, and is happy”. See?—no occurrence of “the black cat” in the paraphrased
sentence.
In fact, once we use Russell’s method, we can get rid of function symbols too.
Given function symbols, we treated “father” as a function symbol, symbolized
it with “ f ”, and symbolized the sentence “George W. Bush’s father was a
politician” as P f (b ). But instead, we could treat ‘father of’ as a two-place

predicate, F , and regard the whole sentence as meaning: “The father of George
W. Bush was a politician.” Given the ι, this could be symbolized as:
P ιxF x b
But given Russell’s method, we can symbolize the whole thing without using
either function symbols or the ι:
∃x(F x b ∧ ∀y(F y b →y=x) ∧ P x)
We can get rid of all function symbols this way, if we want. Here’s the method:
· Take any n-place function symbol f
· Introduce a corresponding n + 1-place predicate R
· In any sentence containing the term “ f (α1 . . . αn )”, replace each occur-
rence of this term with “the x such that R(x, α1 . . . αn )”.
· Finally, symbolize the resulting sentence using Russell’s theory of de-

scriptions
For example, let’s go back to: “Every even number is the sum of two prime
numbers”. Instead of introducing a function symbol s (x, y) for “the sum of x
and y”, let’s introduce a predicate letter R(z, x, y) for “z is a sum of x and y”.
We then use Russell’s method to symbolize the whole sentence thus:
∀x(E x → ∃y∃z[P y∧P z ∧ ∃w(Rwy z ∧ ∀w1 (Rw1 y z→w1 =w) ∧ x=w)])
The end of the formula (beginning with ∃w) says “the product of y and z is
identical to x”—that is, that there exists some w such that w is a product of y
and z, and there is no other product of y and z other than w, and w = x.

logic with identity, function symbols, and the ι operator. (Do not
eliminate descriptions using Russell’s method.)
a) If a person commits a crime, then the judge that sentences

him/her wears a wig.
b) The tallest spy is a spy. (Use a two-place predicate to sym-

bolize “is taller than”.)
Exercise 5.8 For the sentence “The ten-feet-tall man is not

happy”, first symbolize with the ι operator. Then symbolize two
readings using Russell’s method. Explain the intuitive difference
between those two readings. Which gives truth conditions like the
ι symbolization?
5.4 Further quantifiers

Standard logic contains just the quantifiers ∀ and ∃. As we have seen, using just
these quantifiers, plus the rest of standard predicate logic, one can represent
the truth conditions of a great many sentences of natural language. But not all.
For instance, there is no way to symbolize the following sentences in predicate
logic:
Most things are massive

Most men are brutes
There are infinitely many numbers
Some critics admire only one another
Like those sentences that are representable in standard logic, these sentences
involve quantificational notions: most things, some critics, and so on. In this
section we introduce a broader conception of what a quantifier is, and new
quantifiers that allow us to symbolize these sentences.
5.4.1 Generalized monadic quantifiers

We will generalize the idea behind the standard quantifiers ∃ and ∀ in two
ways. To approach the first, think about the clauses in the definition of truth in
a PC-model, M , with domain D, for ∀ and ∃:
VM ,g (∀αφ) = 1 iff for every u ∈ D, VM ,g α (φ) = 1

u
VM ,g (∃αφ) = 1 iff for some u ∈ D, VM ,g α (φ) = 1

u
Let’s introduce the following bit of terminology. For any PC-model, M (=

〈D, I 〉), and wff, φ, let’s introduce the name “φM ,g ,α ” for (roughly speaking)
the set of members of M ’s domain of which φ is true:
Definition: φM , g ,α = {u : u ∈ D and VM ,g α (φ) = 1}
u
Thus, if we begin with any variable assignment g , then φM ,g ,α is the set of

things u in D such that φ is true, relative to variable assignment g uα ). Given
this terminology, we can rewrite the clauses for ∀ and ∃ as follows:
VM ,g (∀αφ) = 1 iff φM ,g ,α = D
VM ,g (∃αφ) = 1 iff φM ,g ,α 6= ∅
But if we can rewrite the semantic clauses for the familiar quantifiers ∀ and
∃ in this way—as conditions on φM ,g ,α —then why not introduce new symbols
of the same grammatical type as ∀ and ∃, whose semantics is parallel to ∀ and
∃ except in laying down different conditions on φM ,g ,α ? These would be new
kinds of quantifiers. For instance, for any integer n, we could introduce a
quantifier ∃n , to be read as “there exists at least n”. That is, ∃n φ means: “there
are at least n φs.” The definitions of a wff, and of truth in a model, would be
updated with the following clauses:
· if α is a variable and φ is a wff, then ∃n αφ is a wff
· VM , g (∃n αφ) = 1 iff |φM , g ,α | ≥ n
The expression |A| stands for the “cardinality” of set A—i.e., the number of
members of A. Thus, this definition says that ∃n αφ is true iff the cardinality of
φM ,g ,α is greater than or equal to n—i.e., this set has at least n members.
Now, the introduction of the symbols ∃n do not increase the expressive
power of predicate logic, for as we saw in section 5.1.3, we can symbolize
“there are at least n F s” using just standard predicate logic (plus “=”). The
new notation is merely a space-saver. But other such additions are not mere
space-savers. For example, by analogy with the symbols ∃n , we can introduce a
symbol ∃∞ , meaning “there are infinitely many”:
· if α is a variable and φ is a wff, then “∃∞ α φ” is a wff

· VM , g (∃∞ α φ) = 1 iff |φM , g ,α | is infinite
As it turns out (though I won’t prove it here), the addition of ∃∞ genuinely

enhances predicate logic: no sentence of standard (first-order) predicate logic
has the same truth condition as does ∃∞ xF x.
Another generalized quantifier that is not symbolizable using standard
predicate logic is most:
· If α is a variable and φ is a wff, then “most α φ” is a wff
· VM , g (most α φ) = 1 iff |φM ,g ,α | > |D − φM , g ,α |
The minus-sign in the second clause is the symbol for set-theoretic difference:
A − B is the set of things that are in A but not in B. Thus, the definition says
that most α φ is true iff more things in the domain D are φ than are not φ.
One could add all sorts of additional “quantifiers” Q in this way. Each
would be, grammatically, just like ∀ and ∃, in that each would combine with
a variable, α, and then attach to a sentence φ, to form a new sentence Qαφ.
Each of these new quantifiers, Q, would be associated with a relation between
sets, RQ , such that Qαφ would be true in a PC-model, M , with domain D,
relative to variable assignment g , iff φM ,g ,α bears RQ to D.
If such an added symbol Q is to count as a quantifier in any intuitive sense,
then the relation RQ can’t be just any relation between sets. It should be a
relation concerning the relative “quantities” of its relata. It shouldn’t, for
instance, “concern particular objects” in the way that the following symbol,
∃Ted-loved , concerns particular objects:
VM , g (∃Ted-loved αφ) = 1 iff φM ,g ,α ∩ {u : u ∈ D and Ted loves u} 6= ∅
So we should require the following of RQ . Consider any set, D, and any one-
one function, f , from D onto another set D 0 . Then, if a subset X of D bears
RQ to D, the set f [X ] must bear RQ to D 0 . ( f [X ] is the image of X under
function f —i.e., {u : u ∈ D 0 and u = f (v), for some v ∈ D}. It is the subset of
D 0 onto which f “projects” X .)
Exercise 5.9 Let the quantifier ∃prime mean “there are a prime
number of”. Using the notation of generalized quantifiers, write
out the semantics of this quantifier.
5.4.2 Generalized binary quantifiers

We have seen how the standard quantifiers ∀ and ∃ can be generalized in
one way: syntactically similar symbols may be introduced and associated with
different relations between sets. Our second way of generalizing the standard
quantifiers is to allow two-place, or binary quantifiers. ∀ and φ are monadic
in that ∀α and ∃α attach to a single open sentence φ. Compare the natural
language monadic quantifiers ‘everything’ and ‘something’:
Everything is material
Something is spiritual
Here, the predicates (verb phrases) ‘is material’ and ‘is spiritual’ correspond
to the open sentences of logic; it is to these that ‘everything’ and ‘something’
attach.
But in fact, monadic quantifiers in natural language are atypical. ‘Every’
and ‘some’ typically occur as follows:
Every student is happy

Some fish are tasty
The quantifiers ‘every’ and ‘some’ attach to two predicates. In the first, ‘every’
attaches to ‘[is a] student’ and ‘is happy’; in the second, ‘some’ attaches to ‘[is
a] fish’ and ‘[is] tasty’. In these sentences, we may think of ‘every’ and ‘some’ as
binary quantifiers. (Indeed, one might think of ‘everything’ and ‘something’ as
the result of applying the binary quantifiers ‘every’ and ‘some’ to the predicate
‘is a thing’.) A logical notation can be introduced which exhibits a parallel
structure, in which ∀ and ∃ attach to two open sentences. In this notation, the
form of quantified sentences is:
(∀α:φ)ψ
(∃α:φ)ψ
The first is to be read: “all φs are ψ”; the second is to be read “there is a φ that
is a ψ”. The clauses for these new binary quantifiers in the definition of the
valuation function for a PC-model are:
VM ,g ((∀α:φ)ψ) = 1 iff φM ,g ,α ⊆ ψM ,g ,α
VM ,g ((∃α:φ)ψ) = 1 iff φM ,g ,α ∩ ψM ,g ,α 6= ∅
A further important binary quantifier is the:
· if φ and ψ are wffs and α is a variable, then (the α:φ)ψ is a wff

· VM , g ((the α:φ)ψ) = 1 iff |φM ,g ,α | = 1 and φM , g ,α ⊆ ψM , g ,α
That is, (the α:φ)ψ is true iff i) there is exactly one φ, and ii) every φ is a
ψ. This truth condition, notice, is exactly the truth condition for Russell’s
symbolization of “the φ is a ψ”; hence the name the.
As with the introduction of the monadic quantifiers ∃n , the introduction of
the binary existential and universal quantifiers, and of the, does not increase the
expressive power of first order logic, for the same effect can be achieved with
monadic quantifiers. (∀α:φ)ψ, (∃α:φ)ψ, and (the α:φ)ψ become, respectively:
∀α(φ→ψ)
∃α(φ∧ψ)
∃α(φ ∧ ∀β(φβ →β=α) ∧ ψ)
(where φβ is φ with free αs changed to βs.) But, as with the monadic quantifiers
∃∞ and most, there are binary quantifiers one can introduce that genuinely
increase expressive power. For example, most occurrences of ‘most’ in English
are binary, e.g.:
Most fish swim
To symbolize such sentences, we can introduce a binary quantifier most2 . The

sentence (most2 α:φ)ψ is to be read “most φs are ψs”. The semantic clause for
most2 is:
VM , g ((most2 α:φ)ψ) = 1 iff |φM ,g ,α ∩ ψM ,g ,α | > |φM ,g ,α − ψM ,g ,α |

The binary most2 increases our expressive power, even relative to the monadic
most: not every sentence expressible with the former is equivalent to a sentence
expressible with the latter.3
Exercise 5.10 Symbolize the following sentence:
The number of people multiplied by the num-

ber of cats that bite at least one dog is 198.
You may invent any generalized quantifiers you need, provided you
write out their semantics.
5.4.3 Second-order logic

All the predicate logic we have considered so far is known as first-order. We’ll
now briefly look at second-order predicate logic, a powerful extension to first-
order predicate logic. The distinction has to do with how variables behave, and
has syntactic and semantic aspects.
The syntactic part of the idea concerns the grammar of variables. All
the variables in first-order logic are grammatical terms. That is, they behave
grammatically like names: you can combine them with a predicate to get a
wff; you cannot combine them solely with other terms to get a wff; etc. In
second-order logic, on the other hand, variables can occupy predicate position.
Thus, each of the following sentences is a well-formed formula in second-order
logic:
∃X X a
∃X ∃yXy
Here we see the variable X occupying predicate position. Predicate variables,

like the normal predicates of standard first-order logic, can be one-place, two-
place, three place, etc.
The semantic part of the idea concerns the interpretation of variables. In
first-order logic, a variable-assignment assigns to each variable a member of the
domain. A variable assignment in second-order logic assigns to each standard
3
Westerståhl (1989, p. 29).
(first-order) variable α a member of the domain, as before, but assigns to each

n-place predicate variable a set of n-tuples drawn from the domain. (This is
what one would expect: the semantic value of a n-place predicate is its extension,
a set of n-tuples, and variable assignments assign temporary semantic values.)
Then, the following clauses to the definition of truth in a PC-model must be
added:
· If π is an n-place predicate variable and α1 . . . αn are terms, then VM ,g (πα1 . . . αn ) =

1 iff 〈[α1 ]M ,g . . . [αn ]M , g 〉 ∈ g (π)
· If π is a predicate variable and φ is a wff, then VM ,g (∀πφ) = 1 iff for
every set U of n-tuples from D, VM ,g π (φ) = 1
U
(where gUπ is the variable assignment just like g except in assigning U to π.)
Notice that, as with the generalized monadic quantifiers, no alteration to the
definition of a PC-model is needed. All we need to do is change grammar and
the definition of the valuation function.
Second-order logic is different from first-order logic in many ways. For
instance, one can define the identity predicate in second-order logic:
Second-order definition of identity: “x=y” is short for: ∀X (X x↔Xy)
This can be seen to work correctly as follows. A one-place second-order variable

X gets assigned a set of things. Thus, the atomic sentence X x says that the
object (currently assigned to) x is a member of the set (currently assigned to)
X . Thus, ∀X (X x↔Xy) says that x and y are members of exactly the same
sets. But since x and only x is a member of {x} (i.e., x’s unit set), that means
that the only way for this to be true is for y to be identical to x.
More importantly, the metalogical properties of second-order logic are dra-
matically different from those of first-order logic. For instance, the axiomatic
method cannot be fully applied to second order logic. One cannot write down a
set of axioms for second-order logic that are both sound and complete—unless,
that is, one resorts to cheap tricks like saying “let every valid wff be an axiom”.
This trick is “cheap” because one would have no way of telling what an axiom
is.4
4
More precisely, the resulting set of axioms would fail to be recursive. For a rigorous
statement and proof of this and other metalogical results about second-order logic, see, e.g.,
Boolos and Jeffrey (1989, chapter 18).
Second-order logic also allows us to express claims that cannot be expressed

in first-order logic. Consider the “Geach-Kaplan sentence”:5
Some critics admire only one another

It can be shown that there is no way to symbolize (one reading of) the sentence
using just first-order logic and predicates for ‘is a critic’ and ‘admires’. The
sentence (on the desired reading) says that there is a group of critics such that
the members of that group admire only other members of the group, but one
cannot say this in first-order logic. However, the sentence can be symbolized
in second-order logic:
∃X [∃xX x ∧ ∀x(X x→C x) ∧ ∀x∀y([X x∧Axy]→[Xy∧x6=y)] (GK2 )
(GK2 ) “symbolizes” (GK) in the sense that it contains no predicates other than
C and A, and for every model 〈D, I 〉, the following is true:
(*) (GK2 ) is true in 〈D, I 〉 iff D has a nonempty subset, X , such that i)
X ⊆ I (C ), and ii) whenever 〈u, v〉 ∈ I (A) and u ∈ X , then v ∈ X as
well and v is not u.
No first-order sentence symbolizes the Geach-Kaplan sentence in this sense.

However, one can in a sense symbolize the Geach-Kaplan sentence using a first-
order sentence, provided the sentence employs, in addition to the predicates C
and A, a predicate ∈ for set-membership:
∃z[∃x x∈z ∧ ∀x(x∈z→C x) ∧ ∀x∀y([x∈z∧Axy]→[y∈z∧x6=y)] (GK1 )
(GK1 ) doesn’t “symbolize” (GK) in the sense of satisfying (*) in every model,
for in some models the two-place predicate ∈ doesn’t mean set-membership.
Nevertheless, if we just restrict our attention to models 〈D, I 〉 in which ∈ does
mean set-membership (restricted to the model’s domain, of course—that is,
I (∈) = {〈u, v〉 : u, v ∈ D and u ∈ v}), and in which D contains each subset of
I (C ) as a member, then (GK1 ) will indeed satisfy (*). In essence, the difference
between (GK1 ) and (GK2 ) is that it is hard-wired into the definition of truth in
a model that second-order predications Xy express set-membership, whereas
this is not hard-wired into the definition of the first-order predication y ∈ z.6
5
The sentence and its significance were discovered by Peter Geach and David Kaplan. See
Boolos (1984).
6
For more on second-order logic, see Boolos (1975, 1984, 1985).
Chapter 6
Propositional Modal Logic
odal logic is the logic of necessity and possibility. In it we treat words

M like “necessary”, “could be”, “must be”, etc. as logical constants. Here
are our new symbols:
2φ: “It is necessary that φ”, “Necessarily, φ”, “It must be that φ”
3φ: “It is possible that φ”, “Possibly, φ”, “It could be that φ”, “It can be that
φ”, “It might be that φ”
The phrase “φ is possible” is sometimes used in the following sense: “φ

could be true, but then again, φ could be false”. For example, if one says “it
might rain tomorrow”, one might intend to say not only that there is a possibility
of rain, but also that there is a possibility that there will be no rain. This is not
the sense of ‘possible’ that we symbolize with the 3. In our intended sense,
“possibly φ” does not imply “possibly not-φ”. To get into the spirit of this sense,
note the naturalness of saying the following: “well of course 2 + 2 can equal 4,
since it does equal 4”. Here, ‘can’ is used in our intended sense: it is presumably
not possible for 2 + 2 to fail to be 4, and so in this case, ‘it can be the case that
2 + 2 equals 4’ does not imply ‘it can be the case that 2 + 2 does not equal 4’.
It is helpful to think of the 2 and the 3 in terms of possible worlds. A
possible world is a complete and possible scenario. Calling a scenario “possible”
means simply that it’s possible that the scenario happen, i.e., be actual. This
requirement disqualifies scenarios in which, for example, it is both raining and
also not raining (at the same time and place)—such a thing couldn’t happen, and
118
CHAPTER 6. PROPOSITIONAL MODAL LOGIC 119
so doesn’t happen in any possible world. But within this limit, we can imagine
all sorts of possible worlds: possible worlds with talking donkeys, possible
worlds in which I am ten feet tall, and so on. “Complete” means simply that
no detail is left out—possible worlds are completely specific scenarios. There
is no possible world in which I am “somewhere between ten and eleven feet
tall” without being some particular height.1 Likewise, in any possible world in
which I am exactly ten feet, six inches tall (say), I must have some particular
weight, must live in some particular place, and so on. One of these possible
worlds is the actual world—this is the complete and possible scenario that in
fact obtains. The rest of them are merely possible—they do not obtain, but
would have obtained if things had gone differently.
In terms of possible worlds, we can think of our modal operators thus:
“2φ” is true iff φ is true in all possible worlds

“3φ” is true iff φ is true in at least one possible world
It is necessarily true that all bachelors are male; in every possible world, every
bachelor is male. There might have existed a talking donkey; some possible
world contains a talking donkey. Possible worlds provide, at the very least,
a vivid way to think about necessity and possibility. How much more than a
vivid guide they provide is an open philosophical question. Some maintain that
possible worlds are the key to the metaphysics of modality, that what it is for a
proposition to be necessarily true is for it to be true in all possible worlds.2
Whether this view is defensible is a question beyond the scope of this book;
what is important for present purposes is that we distinguish possible worlds as
a vivid heuristic from possible worlds as a concern in serious metaphysics.
Our first topic in modal logic is the addition of the 2 and the 3 to proposi-
tional logic; the result is modal propositional logic (“MPL”). A further step will be
be modal predicate logic (chapter 9).
6.1 Grammar of MPL

We need a new language: the language of propositional modal logic. The
grammar of this language is just like the grammar of propositional logic, except
that we add the 2 as a new one-place sentence connective:
1
This is not to say that possible worlds exclude vagueness.
2
Sider (2003) presents an overview of this topic.
· Sentence letters: P, Q, R . . . , with or without numerical subscripts
· Connectives: →, ∼, 2
· Parentheses
Definition of wff:
· Sentence letters are wffs
· If φ and ψ are wffs then φ→ψ, ∼φ, and 2φ are also wffs
· nothing else is a wff
The 2 is the only new primitive connective. But just as we were able to
define ∧, ∨, and ↔, we can define new nonprimitive modal connectives:
· “3φ” (“Possibly φ”) is short for “∼2∼φ
· “φ⇒ψ” (“φ strictly implies ψ”) is short for “2(φ→ψ)”
6.2 Symbolizations in MPL

Modal logic allows us to symbolize a number of sentences we couldn’t symbolize
before. The most obvious cases are sentences that overtly involve “necessarily”,
“possibly”, or equivalent expressions:
Necessarily, if snow is white, then snow is white or grass

is green
2[S→(S∨G)]
I’ll go if I must
2G→G
It is possible that Bush will lose the election
3L
Snow might have been either green or blue
3(G∨B)
If snow could have been green, then grass could have
been white
3G→3W
‘Impossible’ and related expressions signify the lack of possibility:
It is impossible for snow to be both white and not white

∼3(W ∧∼W )
If grass cannot be clever then snow cannot be furry
∼3C →∼3F
God’s being merciful is inconsistent with imperfection’s
being incompatible with your going to heaven.
∼3(M ∧∼3(I ∧H ))
(M = “God is merciful”, I = “You are imperfect”, H =

“You go to heaven”)
As for the strict conditional, it arguably does a decent job of representing

certain English conditional constructions:
Snow is a necessary condition for skiing

∼W ⇒∼K
Food and water are required for survival
∼(F ∧W )⇒∼S
Thunder implies lightning
T ⇒L
Once we add modal operators, we can expose an important ambiguity in

certain English sentences. The surface grammar of a sentence like “if Ted is a
bachelor, then he must be unmarried” is misleading: it suggests the symboliza-
tion:
B→2U
But since I am in fact a bachelor, it would follow from this symbolization that
the proposition that I am unmarried is necessarily true. But clearly I am not
necessarily a bachelor—I could have been married! The sentence is not saying
that if I am in fact a bachelor, then the following is a necessary truth: I am
married. It is rather saying that, necessarily, if I am a bachelor then I am
married:
2(B→U )
It is the relationship between my being a bachelor and my being unmarried that

is necessary. Think of this in terms of possible worlds: the first symbolization
says that if I am a bachelor in the actual world, then I am unmarried in every
possible world (which is absurd); whereas the second one says that in each
possible world, w, if I am a bachelor in w, then I am unmarried in w (which
is quite sensible). The distinction between φ→2ψ and 2(φ→ψ) is called
the distinction between the “necessity of the consequent” (first sentence) and
the “necessity of the consequence” (second sentence). It is important to keep
the distinction in mind, because of the fact that English surface structure is
misleading.
English modal words are ambiguous in a systematic way. For example,
suppose I say that I can’t attend a certain conference in Cleveland. What is the
force of “can’t” here? Probably I’m saying that my attending the conference
is inconsistent with honoring other commitments I’ve made at that time. But
notice that another sentence I might utter is: “I could attend the conference;
but I would have to cancel my class, and I don’t want to do that.” Now I’ve
said that I can attend the conference; have I contradicted my earlier assertion
that I cannot attend the conference? No—what I mean now is perhaps that I
have the means to get to Cleveland on that date. I have shifted what I mean by
“can”.
In fact, there are a lot of things one could mean by a modal word like ‘can’.
Examples:
I can come to the party, but I can’t stay late. (“can” = “is not
inconvenient”)
Humans can travel to the moon, but not Mars. (“can” = “is
achievable with current technology”)
Objects can move almost as fast as the speed of light, but nothing
can travel faster than light. (“can” = “is consistent with
the laws of nature”)
Objects could have traveled faster than the speed of light (if
the laws of nature had been different), but no matter what the
laws had been, nothing could have traveled faster than itself.
(“can” = “metaphysical possibility”)
You can borrow but you can’t steal. (“can” = “morally ac-
ceptable”)
So when representing English sentences using the 2 and the 3, one should
keep in mind that these expressions can be used to express different strengths
of necessity and possibility. (Though we won’t do this, one could introduce
different symbols for different sorts of possibility and necessity.)
The different strengths of possibility and necessity can be made vivid by
thinking, again, in terms of possible worlds. As we saw, we can think of the 2
and the 3 as quantifiers over possible worlds (the former a universal quantifier,
the latter an existential quantifier). The very broad sort of possibility and
necessity, metaphysical possibility and necessity, can be thought of as a completely
unrestricted quantifier: a statement is necessarily true iff it is true in all possible
worlds whatsoever. The other kinds of possibility and necessity can be thought
of as resulting from various restrictions on the quantifiers over possible worlds.
Thus, when ‘can’ signifies achievability given current technology, it means:
true in some possible world in which technology has not progressed beyond where it has
progressed in fact at the current time; when ‘can’ means moral acceptability, it
means: true in some possible world in which nothing morally forbidden occurs; and so
on.
6.3 Semantics for MPL

As usual, let’s consider semantics first. As always, our goal is to model how
statements involving the 2 and 3 are made true by the world, in order to shed
light on the meaning of these connectives, and in order to provide semantic
definitions of the notions of logical truth and logical consequence.
In constructing a semantics for MPL, we face two main challenges, one
philosophical, the other technical. The philosophical challenge is simply that
it isn’t wholly clear which formulas of MPL are indeed logical truths. It’s hard
to construct an engine to spit out logical truths if you don’t know which logical
truths you want it to spit out. With a few exceptions, there is widespread
agreement over which formulas of nonmodal propositional and predicate logic
are logical truths. But for modal logic, this is less clear, especially for sentences
that contain iterations of modal operators. Is 2P →22P a logical truth? It’s
hard to say.
A quick peek at the history of modal logic is in order. Modal logic arose
from dissatisfaction with the material conditional → of standard propositional
logic. The material conditional φ→ψ is true whenever φ is false or ψ is true;

but in expressing the conditionality of ψ on φ, we sometimes want to require a
tighter relationship: we want it not to be a mere accident that either φ is false or
ψ is true. To express this tighter relationship, C. I. Lewis introduced the strict
conditional φ⇒ψ, which he defined, as above, as 2(φ→ψ).3 Thus defined,
φ⇒ψ isn’t automatically true just because φ is false or ψ is true. It must be
necessarily true that either φ is false or ψ is true.
Lewis then asked: what principles govern this new symbol 2? Certain
principles seemed clearly appropriate, for instance: 2(φ→ψ)→(2φ→2ψ).
Others were less clear. Is 2φ→22φ a logical truth? What about 32φ→φ?
Lewis’s solution to this problem was not to choose. Instead, he formulated
several different modal systems. He did this axiomatically, by formulating differ-
ent systems that differed from one another by containing different axioms and
hence different theorems.
We will follow Lewis’s approach, and construct several different modal
systems. Unlike Lewis, we’ll do this semantically at first (the semantics for
modal logic we will study was published by Saul Kripke in the 1950s, long
after Lewis was writing), by constructing different definitions of a model for
modal logic. The definitions will differ from one another in ways that result
in different sets of valid formulas. In section 6.4 we’ll study Lewis’s axiomatic
systems, and in sections 6.5 and 6.6 we’ll discuss the relationship between the
semantics and the axiom systems.
Formulating multiple systems does not answer the philosophical question
of which formulas of modal logic are logically true; it merely postpones it.
The question re-arises when we want to apply Lewis’s systems; when we ask
which system is the correct system—i.e., which one correctly mirrors the logical
properties of the English words ‘possibly’ and ‘necessarily’? (Note that since
there are different sorts of necessity and possibility, different systems might
correctly represent different sorts of necessity.) But we won’t try to address
such philosophical questions here.
The technical challenge to constructing a semantics for MPL is that the
modal operators 2 and 3 are not truth functional. A (sentential) connective is an
expression that combines with sentences to make new sentences. A one-place
connective combines with one sentence to form a new sentence. ‘It is not the
case that’ is a one-place connective of English—the ∼ is a one-place connective
in the language of PL. A connective is truth-functional iff whenever it combines
3
See Lewis (1918); Lewis and Langford (1932).
with sentences to form a new sentence, the truth value of the resulting sentence
is determined by the truth value of the component sentences. Many think
that ‘and’ is truth-functional, since they think that an English sentence of the
form “φ and ψ” is true iff φ and ψ are both true. But ‘necessarily’ is not truth-
functional. Suppose I tell you the truth value of φ; will you be able to tell me
the truth value of this sentence? Well, if φ is false then presumably you can (it
is false), but if φ is true, then you still don’t know. If φ is “Ted is a philosopher”
then “Necessarily φ” is false, but if φ is “Either Ted is a philosopher or he isn’t
a philosopher” then “Necessarily φ” is true. So the truth value of “Necessarily
φ” isn’t determined by the truth value of φ. Similarly, ‘possibly’ isn’t truth-
functional either: ‘I might have been six feet tall’ is true, whereas ‘I might have
been a round square’ is false, despite the fact that ‘I am six feet tall’ and ‘I am a
round square’ each have the same truth value (they’re both false.)
Since the 2 and the 3 are supposed to represent ‘necessarily’ and ‘possibly’,
respectively, and since the latter aren’t truth-functional, we can’t use the method
of truth tables to construct the semantics for the 2 and the 3. For the method
of truth tables assumes truth-functionality. Truth tables are just pictures of truth
functions: they specify what truth value a complex sentence has as a function of
what truth values its parts have. Imagine trying to construct a truth table for
the 2. It’s presumably clear (though see the discussion of systems K, D, and T
below) that 2φ should be false if φ is false, but what about when φ is true?:
2
1 ?
0 0
There’s nothing we can put in this slot in the truth table, since when φ is true,
sometimes 2φ is true and sometimes it is false.
Our challenge is clear: we need a semantics for the 2 and the 3 other than
the method of truth tables.
6.3.1 Relations
Before we investigate how to overcome this challenge, a digression is necessary,
to introduce the concept of a relation. A relation is just a feature of multiple
objects taken together. The taller-than relation is one example: when one
person is taller than another, that’s a feature of those two objects taken together.
Another example is the the less-than relation for numbers. When one number
is less than another, that’s a feature of those two numbers taken together.
“Binary” relations apply to two objects at a time. The taller-than and less-
than relations are binary relations, or “two-place” relations as we might say.
We can also speak of three-place relations, four-place relations, and so on.
An example of a three-place relation would be the betweenness relation for
numbers: the relation that holds between 2, 5, and 23 for example.
Recall our discussion of ordered sets from section 1.8. In addition to their
use in constructing models, ordered sets are also useful for giving an official
definition of what a relation is.
Definition of relation: An n-place relation is defined as a set of n-tuples. So

a binary (two-place) relation is a set of ordered pairs.
For example, the taller-than relation may be taken to be the set of ordered pairs
〈u, v〉 such that u is a taller person than v. The less-than relation for positive
integers is the set of ordered pairs 〈m, n〉 such that m is a positive integer less
than n, another positive integer. That is, it is the following set:
{〈1, 2〉, 〈1, 3〉, 〈1, 4〉 . . . 〈2, 3〉, 〈2, 4〉 . . . }
When 〈u, v〉 is a member of relation R, we say, equivalently, that u and v “stand

in” R, or R “holds between” u and v, or that u “bears” R to v. Most simply,
we write “Ruv”. (This notation is like that of predicate logic; but here I’m
speaking the metalanguage, not displaying sentences of a formalized language.)
Some more definitions.
Definition of domain, range, over: Let R be any binary relation.
· The domain of R (“dom(R)”) is the set {u: for some v, Ruv}

· The range of R (“ran(R)”) is the set {u: for some v, Rv u}
· R is over A iff dom(R) ⊆ A and ran(R) ⊆ A
In other words, the domain of R is the set of all things that bear R to something;
the range is the set of all things that something bears R to; and R is over A iff
the members of the ’tuples in R are all drawn from A.
Binary relations come in different types, depend on the patterns in which
they hold. Here are some types of binary relations that we will need to think
about:
Definition of kinds of binary relations: Let R be any binary relation over

A.
· R is serial (in A) iff for every u ∈ A, there is some v ∈ A such that Ruv.
· R is reflexive (in A) iff for every u ∈ A, Ru u
· R is symmetric iff for all u, v, if Ruv then Rv u
· R is transitive iff for any u, v, w, if Ruv and Rv w then Ruw
· R is an equivalence relation (in A) iff R is symmetric, transitive, and
reflexive (in A)
· R is total (in A) iff for every u, v ∈ A, Ruv
Notice that we relativize some of these relation types to a given set A. The
notion of reflexivity is defined as being relative to a set, for example. We do
this because the alternative would be to say that a relation is reflexive simpliciter
if everything bears R to itself; but that would require the domain and range
of any reflexive relation to be the set of absolutely all objects. It’s better to
introduce the notion of being reflexive relative to a set, which is applicable to
relations with smaller domains and ranges. (I will sometimes omit the qualifier
‘in A’ when it is clear which set that is.) Why don’t symmetry and transitivity
have to be relativized to a set?—because they only say what must happen if
R holds among certain things. Symmetry, for example, says merely that if R
holds between u and v, then it must also hold between v and u, and so we can
say that a relation is symmetric absolutely, without implying that everything is
in its domain and range.
6.3.2 Kripke models

Now we’re ready to introduce a semantics for MPL. As we saw, we can’t
construct truth tables for the 2 or the 3. Instead, we will pursue an approach
called possible-worlds semantics. The intuitive idea is to count 2φ as being true
iff φ is true in all possible worlds, and 3φ as being true iff φ is true in some
possible worlds. More carefully: we are going to develop models for modal
propositional logic. These models will contain objects we will call “possible
worlds”. And formulas are going to be true or false “at” these worlds—that
is, we are going to assign truth values to formulas in these models relative to
possible worlds, rather than absolutely. Truth values of propositional-logic
compound formulas—that is, negations and conditionals—will be determined
by truth tables within each world; ∼φ, for example, will be true at a world iff φ
is false at that world. But the truth value of 2φ at a world won’t be determined
by the truth value of φ at that world; the truth value of φ at other worlds will
also be relevant.
Specifically, 2φ will count as true at a world iff φ is true at every world that
is “accessible” from the first world. What does “accessible” mean? Each model
will come equipped with a binary relation, R, that holds between possible
worlds; we will say that world v is “accessible from” world w when R wv. The
intuitive idea is that R wv if and only if v is possible relative to w. That is, if you
live in world w, then from your perspective, the events in world v are possible.
The idea that what is possible might vary depending on what possible
world you live in might at first seem strange, but it isn’t really. “It is physically
impossible to travel faster than the speed of light” is true in the actual world,
but false in worlds where the laws of nature allow faster-than-light travel.
On to the semantics. We first define a general notion of a MPL model,
which we’ll then use to give a semantics for each of our systems:
Definition of model: An MPL-model is an ordered triple, 〈W , R, I 〉, where:
· W is a non-empty set of objects (“possible worlds”)
· R is a binary relation over W (“accessibility relation”)
· I is a two-place function that assigns a 0 or 1 to each sentence letter,
relative to (“at”, or “in”) each world—that is, for any sentence letter α,
and any w ∈ W , I (α, w) is either 0 or 1. (“interpretation function”)
Each MPL-model contains a set W of possible worlds, and an accessibility
relation R. 〈W , R〉 is sometimes called the model’s frame. Think of the frame
as a map of the “structure” of the model’s space of possible worlds: it contains
information about how many worlds there are, and which worlds are accessible
from which. In addition to a frame, each model also contains an interpretation
function I , which assigns truth values to sentence letters.
A model’s interpretation function assigns truth values only to sentence
letters. But the sum total of all the truth values of sentence letters relative to
worlds determines the truth values of all complex wffs, again relative to worlds.
It is the job of the model’s valuation function to specify exactly how these truth
values get determined:
Definition of valuation: Where M (= 〈W , R, I 〉) is any MPL-model, the
valuation for M , VM , is defined as the two-place function that assigns either
0 or 1 to each wff relative to each member of W , subject to the following

constraints, where α is any sentence letter, φ and ψ are any wffs, and w is any
member of W :
VM (α, w) = I (α, w)
VM (∼φ, w) = 1 iff VM (φ, w) = 0
VM (φ→ψ, w) = 1 iff either VM (φ, w) = 0 or VM (ψ, w) = 1
VM (2φ, w) = 1 iff for each v ∈ W , if R wv, then VM (φ, v) = 1
What about the truth values for complex formulas that contain ∧, ∨, ↔, and
3? Given the definition of these defined connectives in terms of the primitive
connectives, it is easy to prove that the following derived conditions hold:
VM (φ∧ψ, w) = 1 iff VM (φ, w) = 1 and VM (ψ, w) = 1
VM (φ∨ψ, w) = 1 iff VM (φ, w) = 1 or VM (ψ, w) = 1
VM (φ↔ψ, w) = 1 iff VM (φ, w) = VM (ψ, w)
VM (3φ, w) = 1 iff for some v ∈ W , R wv and VM (φ, v) = 1
So far, we have introduced a general notion of an MPL model, and have
defined the notion of a wff’s being true at a world in an MPL model. Next, let
us consider how to define validity.
Remember that our overall strategy is C. I. Lewis’s: we want to construct
different modal systems, since it isn’t obvious which formulas ought to count as
logical truths. The systems will be named: K, D, T, B, S4, S5. Each system will
come with its own definition of a model. As a result, different formulas will
come out valid in the different systems. For example, as we’ll see, the formula
2P →22P is going to come out valid in S4 and S5, but not in the other systems.
Here are the definitions:
Definition of validity for modal systems:

· A K-model is defined as any MPL-model
· A D-model is any MPL-model whose accessibility relation is serial (i.e.,
any model 〈W , R, I 〉 in which R is serial in W )
· A T-model is any MPL-model whose accessibility relation is reflexive (in
W)
· A B-model is any MPL-model whose accessibility relation is reflexive (in
W ) and symmetric
· An S4-model is any MPL model whose accessibility relation is reflexive

(in W ) and transitive
· An S5-model is any MPL model whose accessibility relation is reflexive
(in W ), symmetric, and transitive
· A formula φ is valid in model M (= 〈W , R, I 〉 iff for every w ∈ W ,

VM (φ, w) = 1
· A formula is valid in system S (where S is either K, D, T, B, S4, or S5) iff
it is valid in every S-model
Notice that for each system, the valid formulas are defined as the formulas
that are valid in every model in which the accessibility relation has a certain
formal feature. The systems differ from from one another by what that formal
feature is. For T it is reflexivity: a formula is T-valid iff it is valid in every
model in which the accessibility relation is reflexive. For S4 the formal feature
is reflexivity + transitivity. Other systems correspond to other formal features.
As before, we’ll use the notation for validity. But since we have many
modal systems, if we claim that a formula is valid, we’ll need to indicate which
system we’re talking about. Let’s do that by subscripting with the name of
the system; thus, “T φ” means that φ is T-valid.
It’s important to get clear on the status of possible-worlds lingo here. Where
〈W , R, I 〉 is a model, we call the members of W “worlds”, and we call R the
“accessibility” relation. Now, there is no question that “possible worlds” is a
vivid way to think about necessity and possibility. But officially, W is nothing
but a nonempty set, any old nonempty set. Its members needn’t be the kinds
of things metaphysicians call possible worlds: they can be numbers, people,
bananas—whatever you like. Similarly, R is just defined to be any old binary
relation on R; it needn’t have anything to do with the metaphysics of modality.
Officially, then, the possible-worlds talk we use to describe our models is just
talk, not heavy-duty metaphysics. Still, models are usually intended to model
something—to depict some aspect of the dependence of truth on the world.
So if modal sentences of English containing ‘necessarily’ and ‘possibly’ aren’t
made true by anything like possible worlds, it’s hard to see why possible worlds
models would shed any light on their meaning, or why truth-in-all-possible-
worlds-models would be a good way of modeling (genuine) validity for modal
statements. At any rate, this philosophical issue should be kept in mind. Back,
now, to the formalism.
6.3.3 Semantic validity proofs

Given our definition of validity, one can now show that a certain formula is
valid in a given system. First, a very simple example.
Example 6.1: The formula 2(P ∨∼P ) is K-valid. To show this formula is
K-valid, we must show that it is valid in every MPL-model, since validity-in-all-
MPL-models is the definition of K-validity. Being valid in a model means being
true at every world in the model. So, consider any MPL-model 〈W , R, I 〉,
and let w be any world in W . We must prove that VM (2(P ∨∼P ), w) = 1. (As
before, I’ll start to omit the subscript M on VM when it’s clear which model
we’re talking about.)
i) Suppose for reductio that V(2(P ∨∼P ), w) = 0
ii) So, by the truth condition for the 2 in the definition of the valuation
function, there is some world, v, such that R wv and V(P ∨∼P, v) = 0
iii) Given the truth condition for the ∨, V(P, v) = 0 and V(∼P, v) = 0
iv) Since V(∼P, v) = 0, given the truth condition for the ∼, V(P, v) = 1. But
that’s impossible; V(P, v) can’t be both 0 and 1.
Thus, K 2(P ∨∼P ).
Note that similar reasoning would establish K φ, for any propositional-

logic tautology φ. The reason is this: within any world, the truth values of
complex statements of propositional logic are determined by the truth values
of their constituents in that world by the usual truth tables. So if φ is a PL-
tautology, it will be true in any world in any model; hence 2φ will turn out
true in any world in any model.
Example 6.2: Show that T (32(P →Q)∧2P ) → 3Q. We must show that
V((32(P →Q)∧2P )→3Q, w) = 1 for the valuation V for an arbitrarily chosen
model and world w in that model.
i) Assume for reductio that V((32(P →Q)∧2P )→3Q, w) = 0
ii) So V(32(P →Q)∧2P, w) = 1 and …
iii) …V(3Q, w) = 0
iv) From ii), 32(P →Q) is true at w, and so V(2(P →Q), v) = 1, for some
world, call it v, such that R wv
v) From ii), V(2P, w) = 1. So, by the truth condition for the 2, P is true in
every world accessible from w; since R wv, it follows that V(P, v) = 1.
vi) From iv), P →Q is true in every world accessible from v; since our model
is a T-model, R is reflexive. So R v v; and so V(P →Q, v) = 1
vii) From v) and vi), by the truth condition for the →, V(Q, v) = 1
viii) Given iii), Q is false at every world accessible from w; this contradicts
vii)
The last example just showed that the formula (32(P →Q)∧2P )→3Q is
valid in T. Suppose we were interested in showing that this formula is also valid
in S4. What more would we have to do? Nothing! To be S4-valid is to be
valid in every S4-model; but a quick look at the definitions shows that every
S4-model is a T-model. So, since we already know that the the formula is valid
in all T-models, we already know that it must be valid in all S4-models (and
hence, S4-valid), without doing a separate proof.
Think of it another way. To do a proof that the formula is S4-valid, we need
to do a proof in which we are allowed to assume that the accessibility relation is
both transitive and reflexive. And the proof above did just that. We didn’t ever
use the fact that the accessibility relation is transitive—we only used the fact
that it is reflexive (in line 9). But we don’t need to use everything we’re allowed
to assume.
In contrast, the proof above doesn’t establish that this formula is, say, K-valid.
To be K-valid, the formula would need to be valid in all models. But some
models don’t have reflexive accessibility relations, whereas the proof we gave
assumed that the accessibility relation was reflexive. And in fact the formula
isn’t in fact K-valid, as we’ll show how to demonstrate in the next section.
Consider the following diagram of systems:
= S5 `@@
||| @@
| @@
||
S4 aB ~>
B
BB ~
BB ~
B ~~~
TO
DO
An arrow from one system to another indicates that validity in the first system
implies validity in the second system. For example, if a formula is D-valid, then
it’s also T-valid. The reason is that if something is valid in all D-models, then,
since every T-model is also a D-model (since reflexivity implies seriality), it
must be valid in all T-models as well.
S5 is the strongest system, since it has the most valid formulas. (That’s
because it has the fewest models—it’s easier to be S5-valid because there are
fewer potentially falsifying models.)
Notice that the diagram isn’t linear. That’s because of the following. Both B
and S4 are stronger than T; each contains all the T-valid formulas. But neither
B nor S4 is stronger than the other—each contains valid formulas that the
other doesn’t. (They of course overlap, because each contains all the T-valid
formulas.) S5 is stronger than each; S5 contains all the valid formulas of each.
These relationships between the systems will be exhibited below.
Suppose you are given a formula, and for each system in which it is valid,
you want to give a semantic proof of its validity. This needn’t require multiple
semantic proofs—as we have seen, one semantic proof can do the job. To prove
that a certain formula is valid in a number of systems, it suffices to prove that it
is valid in the weakest possible system. Then, that very proof will automatically
be a proof that it is valid in all stronger systems. For example, a proof that a
formula is valid in K would itself be a proof that the formula is D, T, B, S4,
and S5-valid. Why? Because every model of any kind is a K-model, so K-valid
formulas are always valid in all other systems.
In general, then, to show what systems a formula is valid in, it suffices to
give a single semantic proof of it, namely, a semantic proof in the weakest
system in which it is valid. There is an exception, however, since neither B

nor S4 is stronger than the other. Suppose a formula is not valid in T, but one
has given a semantic proof its validity in B. This proof also establishes that the
formula is also valid in S5, since every S5 model is a B-model. But one still
doesn’t yet know whether the formula is S4-valid, since not every S4-model is a
B-model. Another semantic proof may be needed: of the formula’s S4-validity.
(Of course, the formula may not be S4-valid.)
So: when a wff is valid in both B and S4, but not in T, two semantic proofs
of its validity are needed.
We are now in a position to do validity proofs. But as we’ll see in the next
section, it’s often easier to do proofs of validity when one has failed to construct
a counter-model for a formula.
Exercise 6.1 Use validity proofs to demonstrate the following:
a) D [2P ∧2(∼P ∨Q)]→3Q
b) S4 33(P αQ)→3Q
6.3.4 Countermodels
We have a definition of validity for the various systems, and we’ve shown how
to establish validity of particular formulas. Now we’ll investigate establishing
invalidity.
Let’s show that the formula 3P →2P is not K-valid. A formula is K-valid if
it is valid in all K-models, so all we must do is find one K-model in which it
isn’t valid. What follows is a procedure for doing this:4
Place the formula in a box

The goal is to find some model, and some world in the model, where the
formula is false. Let’s start by drawing a box, which represents some chosen
world in the model we’ll construct. The goal is to make the formula false in
this world. In these examples I’ll always call this first world “r”:
r 3P →2P
4
This procedure is from Cresswell and Hughes (1996).
Now, since the box represents a world, we should have some way of representing
the accessibility relation. What worlds are possible, relative to r; what worlds
does r “see”? Well, to represent one world (box) seeing another, we’ll draw
an arrow from the first to the second. However in the case of this particular
model, we don’t need to make this world r see anything. After all, we’re trying
to construct a K-model, and the accessibility relation of a K-model doesn’t
even need to be serial—no world needs to see any worlds at all. So, we’ll forget
about arrows for the time being.
Make the formula false in the world

We will indicate a formula’s truth value (1 or 0) by writing it above the formula’s
major connective. So to indicate that 3P →2P is to be false in this model, we’ll
put a 0 above its arrow:
0
3P →2P
r
Enter in forced truth values

If we want to make the 3P →2P false in this world, the definition of a valuation
function requires us to assign certain other truth values. Whenever a conditional
is false at a world, its antecedent is true at that world and its consequent is false
at that world. So, we’ve got to enter in more truth values; a 1 over the major
connective of the antecedent (3P ), and a 0 over the major connective of the
consequent (2P ):
1 0 0
3P →2P
r
Enter asterisks
When we assign a truth value to a modal formula, we thereby commit ourselves
to assigning certain other truth values to various formulas at various worlds.
For example, when we make 3P true at r, we commit ourselves to making P
true at some world that r sees. To remind ourselves of this commitment, we’ll
put an asterisk (*) below 3P . An asterisk below indicates a commitment to there
being some world of a certain sort. Similarly, since 2P is false at r, this means
that P must be false in some world P sees (if it were true in all such worlds,
then by the semantic clause for the 2, 2P would be true at r). We again have a
commitment to there being some world of a certain sort, so we enter an asterisk
below 2P as well:
1 0 0
r 3P →2P
∗ ∗
Discharge bottom asterisks

The next step is to fulfill the commitments we incurred by adding the bottom
asterisks. For each, we need to add a world to the diagram. The first asterisk
requires us to add a world in which P is true; the second requires us to add a
world in which P is false. We do this as follows:
1 0 0
r 3P →2P
∗ ∗
??
??
??
??
??

a 1 0
b
P P
What I’ve done is added two more worlds to the diagram: a and b. P is true in
a, but false in b. I have thereby satisfied my obligations to the asterisks on my
diagram, for r does indeed see a world in which P is true, and another in which
P is false.
The official model

We now have a diagram of a K-model containing a world in which 3P →2P
is false. But we need to produce an official model, according to the official
definition of a model. A model is an ordered triple 〈W , R, I 〉, so we must
specify the model’s three members.
The set of worlds We first must specify the set of worlds, W . W is simply
the set of worlds I invoked:
W = {r, a, b}
But what are r, a, and b? Let’s just take them to be the letters ‘r’, ‘a’, and ‘b’. No
reason not to—the members of W , recall, can be any things whatsoever.
The accessibility relation Next, for the accessibility relation. This is

represented on the diagram by the arrows. In our model, there is an arrow
from r to a, an arrow from r to b, and no other arrows. Thus, the diagram
represents that r sees a, that r sees b, and that there are no further cases of
seeing. Now, remember that the accessibility relation, like all relations, is a set
of ordered pairs. So, we simply write out this set:
R = {〈r, a〉, 〈r, b〉}
That is, we write out the set of all ordered pairs 〈w1 , w2 〉 such that w1 “sees”
w2 .
The interpretation function Finally, we need to specify the interpreta-

tion function, I , which assigns truth values to sentence letters at worlds. In
our model, I must assign 1 to P at world a, and 0 to P at world b. Now, our
official definition requires an interpretation to assign a truth value to each of
the infinitely many sentence letters at each world; but so long as P is true at
world a and false at world b, it doesn’t matter what other truth values I assigns.
So let’s just (arbitrarily) choose to make all other sentence letters false at all
worlds in the model. We have, then:
I (P, a) = 1
I (P, b) = 0
I (α, w) = 0 for all other sentence letters α and worlds w
That’s it—we’re done. We have produced a model in which 3P →2P is false

at some world; hence this formula is not valid in all models; and hence it’s not
K-valid: 2K 3P →2P .
Check the model

At the end of this process, it’s a good idea to check to make sure that your
model is correct. This involves various things. First, make sure that you’ve
succeeded in producing the correct kind of model. For example, if you’re trying
to produce a T-model, make sure that the accessibility relation you’ve written
down is reflexive. (In our case, we were only trying to construct a K-model, and
so for us this step is trivial.) Secondly, make sure that the formula in question
really does come out false at one of the worlds in your model.
Simplifying models
Sometimes a model can be simplified. Consider the diagram of the final version
of the model above:
1 0 0
r 3P →2P
∗ ∗
??
??
??
??
??

a 1 0
b
P P
We needn’t have used three worlds in the model. When we discharged the first
asterisk, we needed to put in a world that r sees, in which P is true. But we
needn’t have made that a new world—we could have simply have made P true in
r. Of course we couldn’t haven’t done that for both asterisks, because that would
have made P both true and false at r. So, we could make one simplification:
1 1 0 0
r 3P →2P
∗ ∗
0

0
b
P
The official model would then look as follows:
W = {r, b }
R = {〈r, r 〉, 〈r, b 〉}
I (P, r) = 1, all others 0
Adapting models to different systems

We have showed that 3P →2P is not K-valid. Now, let’s show that this formula
isn’t D-valid, i.e. that it is false in some world of some model with a serial
accessibility relation (i.e., some “D-model”). Well, we haven’t quite done this,
since the model above does not have a serial accessibility relation. But we can
easily change this, as follows:
1 1 0 0
r 3P →2P
∗ ∗
0

0
b
0 P
Official model:
W = {r, b }
R = {〈r, r 〉, 〈r, b 〉, 〈b , b 〉}
That was easy—adding the fact that b sees itself didn’t require changing any-
thing else in the model.
Suppose we want now to show that 3P →2P isn’t T-valid. Well, we’ve
already done so! Why? Because we’ve already produced a T-model in which
this formula is false. Look back at the most recent model. Its accessibility
relation is reflexive. So it’s a T-model already. In fact, that accessibility relation
is also already transitive, so it’s already an S4-model.
So far we have established that 2K,D,T,S4 3P →2P . What about B and S5?
It’s easy to revise our model to make the accessibility relation symmetric:
1 1 0 0
r 3P →2P
∗ ∗
0 O

0
b
0 P
Official model:
W = {r, b }
R = {〈r, r〉, 〈r, b〉, 〈b, b〉, 〈b, r〉}
Now, we’ve got a B-model, too. What’s more, we’ve also got an S5-model:
notice that the accessibility relation is an equivalence relation. (In fact, it’s also
a total relation.)
So, we’ve succeeded in establishing that 3P →2P is not valid in any of our
systems. Notice that we could have done this more quickly, if we had given the
final model in the first place. After all, this model is an S5, S4, B, T, D, and
K-model. So one model establishes that the formula isn’t valid in any of the
systems.
In general, in order to establish that a formula is invalid in a number of
systems, try to produce a model for the strongest system (i.e., the system with
the most requirements on models). If you do, then you’ll automatically have a
model for the weaker systems. Keep in mind the diagram of systems:
S5 `@
|= @@
||| @@
|| @
S4 aB >B
BB
BB ~~~
B ~~~
TO
DO
K
An arrow from one system to another, recall, indicates that validity in the first
system implies validity in the second. The arrows also indicate facts about
invalidity, but in reverse: when an arrow points from one system to another,
then invalidity in the second system implies invalidity in the first. For example,
if a wff is invalid in T, then it is invalid in D. (That’s because every T-model is
a D-model; a countermodel in T is therefore a countermodel in D.)
When our task is to discover which systems a given formula is invalid in,
usually only one countermodel will be needed—a countermodel in the strongest
system in which the formula is invalid. But there is an exception involving B

and S4. Suppose a given formula is valid in S5, but we discover a model showing
that it isn’t valid in B. That model is automatically a T, D, and K-model, so we
know that the formula isn’t T, D, or K-valid. But we don’t yet know about that
formula’s S4-validity. If it is S4-invalid, then we will need to produce a second
countermodel, an S4 countermodel. (Notice that the B-model couldn’t already
be an S4-model. If it were, then its accessibility relation would be reflexive,
symmetric, and transitive, and so it would be an S5 model, contradicting the
fact that the formula was S5-valid.)
Additional steps in countermodelling

I gave a list of steps in constructing countermodels:
1. Place the formula in a box

2. Make the formula false in the world
3. Enter in forced truth values
4. Enter asterisks
5. Discharge bottom asterisks
6. The official model
We’ll need to adapt this list.
Above asterisks Let’s try to get a countermodel for 32P →23P in all the
systems in which it is invalid, and a semantic validity proof in all the systems in
which it is valid. We always start with countermodelling before doing semantic
validity proofs, and when doing countermodelling, we start by trying for a
K-model. After the first few steps, we have:
1 0 0
r 32P →23P
∗ ∗
CC
{{ CC
{{{ CC
{{ CC
{{ CC
{
} { C!
1 0
2P 3P
a b
At this point, we’ve got a true 2, and a false 3. Take the first: a true 2P . This
doesn’t commit us to adding a world in which P is true; rather, it commits us
to making P true in every world that a sees. Similarly, a zero over a 3, over
3P in world b in this case, commits us to making P false in every world that
b sees. We indicate such commitments, commitments in every world seen, by
putting asterisks above the relevant modal operators:
1 0 0
r 32P →23P
∗ ∗
~ @@
~~ @@
~ @@
~~ @@
~~~ @@
~ @@
∗ ~
~ ∗
a 1 b 0
2P 3P
Now, how can we discharge these asterisks? In this case, when trying to
construct a K-model, we don’t need to do anything. Since a, for example,
doesn’t see any world, then automatically P is true in every world it sees; the
statement “for every world, w, if Raw then V(P, w) = 1” is vacuously true.
Same goes for b—P is automatically false in all worlds it sees. So, we’ve got a
K-model in which 32P →23P is true.
Now let’s turn the model into a D-model. Every world must now see at
least one world. Let’s try:
1 0 0
r 32P →23P
∗ ∗
~ @@
~~ @@
~ @@
~~ @@
~~~ @@
~ @@
∗ ~
~ ∗
a 1 b 0
2P 3P

1 0
c d
P P
0 0
I added worlds c and d, so that a and b would each see at least one world.
(Further, worlds c and d each had to see a world, to keep the relation serial.
I could have added still more worlds that c and d saw, but then they would
themselves need to see some worlds…So I just let c and d see themselves.) But
once c and d were added, discharging the upper asterisks in worlds a and b
required making P true in c and false in d (since a sees c and b sees d).
Let’s now try for a T-model. This will involve, among other things, letting
a and b see themselves. But this gets rid of the need for worlds c and d, since
they were added just to make the relation serial. I’ll try:
1 0 0
r 32P →23P
∗ ∗
0 @@
~~ @@
~~~ @@
~~ @@
~~ @@
~ @@
∗ ~
~ ∗
a 1 1 b 0 0
2P 3P
0 0
When I added arrows, I needed to make sure that I correctly discharged the
asterisks. This required nothing of world r, since there were no top asterisks
there. There were top asterisks in worlds a and b; but it turned out to be easy
to discharge these asterisks—I just needed to let P be true in a, but false in b.
Notice that I could have moved straight to this T-model—which is itself a
D-model—rather than first going through the earlier mere-D-model. However,
this won’t always be possible—sometimes you’ll be able to get a D-model, but
no T-model.
At this point let’s verify that our model does indeed assign the value 0 to
our formula 32P →23P . First notice that 2P is true in a (since a only sees
one world—itself—and P is true there). But r sees a. So 32P is true at r. Now,
consider b. b only sees one world, itself, and P is false there. So 3P must also
be false there. But r sees b. So 23P is false at r. But now, the antecedent of
32P →23P is true, while its consequent is false, at r. So that conditional is
false at r. Which is what we wanted.
Onward. Our model is not a B-model, since a, for example, doesn’t see r,
despite the fact that r sees a. So let’s try to make this into a B-model. This
involves making the relation symmetric. Here’s how it looks before I try to
discharge the top asterisks in a and b:
1 0 0
r 32P →23P
∗ ∗
0 ? @_ @
~~~ @@
~~ @@
~~ @@
~ @@
~~ @@
∗ ~
~ ∗
a 1 1 b 0 0
2P 3P
0 0
Now I need to make sure that all top asterisks are discharged. For example,
since a now sees r, I’ll need to make sure that P is true at r. However, since
b sees r too, P needs to be false at r. But P can’t be both true and false at r.
So we’re stuck, in trying to get a B-model in which this formula is false. This
suggests that maybe it is impossible—that is, perhaps this formula is true in all
worlds in all B-models—that is, perhaps the formula is B-valid. So, the thing
to do is try to prove this: by supplying a semantic validity proof.
So, let 〈W , R, I 〉 be any model in which R is reflexive and symmetric, let
V be its valuation function, and let w be any member of W ; we must show that
V(32P →23P, w) = 1.
i) Suppose for reductio that V(32P →23P, w) = 0
ii) Then V(32, w) = 1 and …
iii) …V(23P, w) = 0
iv) By i), for some v, R wv and V(2P, v) = 1.
v) By symmetry, R v w.
vi) From iv), via the truth condition for 2, we know that P is true at every
world accessible from v; and so, by v), V(P, w) = 1.
vii) By iii), there is some world, call it u, such that R w u and V(3P, u) = 0.
viii) By symmetry, w is accessible from u.

ix) By vi), P is false in every world accessible from u; and so by viii), V(P, w) =
0, contradicting vi)
Just as we suspected: the formula is indeed B-valid. So we know that it is

S5-valid (the proof we just gave was itself a proof of its S5-validity). But what
about S4-validity? Remember the diagram—we don’t have the answer yet. The
thing to do here is to try to come up with an S4-model, or an S4 semantic
validity proof. Usually, the best thing to do is to try for a model. In fact, in the
present case this is quite easy: our T-model is already an S4-model.
So, we’re done. Our answer to what systems the formula is valid and invalid
in comes in two parts. First, validity. For the systems in which the formula
is valid, we gave a semantic proof of B-validity above. This was itself a proof
of S5-validity, as I noted. So B,S5 32P →23P . Second, invalidity. For the
systems in which the formula is invalid, we have an S4-model, which we display
officially as follows:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈r, b〉}
I (P, a) = 1, all others 0
This model is itself also a T, D, and K-model (since its accessibility relation is
reflexive and serial), so: 2K,D,T,S4 32P →23P .
Example 6.3: Determine in which systems 32P →3232P is valid and in

which systems it is invalid.
Well, we can get a T-model as follows:
∗
1 0 0 0
32P →3232P
r
∗ ∗ I discharged the second

0 bottom asterisk in
r by letting r see b

Notice how commitments ∗
to specific truth values for 1 1 0
2P 232P
different formulas are recorded a
by placing the formulas
side by side in the box ∗
0

∗
0 0 1
32P
b
P
∗
0

0
c
P
0
Official model:
W = {r, a, b , c}
R = {〈r, r 〉, 〈a, a〉, 〈b , b 〉, 〈c, c〉, 〈r, a〉, 〈r, b 〉, 〈a, b 〉, 〈b , c〉}
I (P, b) = 1, all others 0
Now consider what happens when we try to turn this model into a B-model.
World b must see back to world a. But then the false 32P in b conflicts with the
true 2P in a. So it’s time for a validity proof. In constructed this validity proof,
we can be guided by failed attempt to construct a countermodel (assuming all
of our choices in constructing that countermodel were forced). In the following
proof that the formula is B-valid, I chose variables for worlds that match up
with the countermodel above:
i) Suppose for reductio that V(32P →3232P, r ) = 0, in some world r in

some B-model 〈W , R, I 〉
ii) So V(32P, r ) = 1 and . . .
iii) V(3232P, r ) = 0
iv) From ii), there’s some world, call it a, such that V(2P, a) = 1 and R ra
v) From iii), since R ra, V(232P, a) = 0
vi) And so, there’s some world, call it b , such that V(32P, b ) = 0 and Rab
vii) By symmetry, R b a. And so, given vi), V(2P, a) = 0. This contradicts iv)
We now have a T-model for the formula, and a proof that it is B-valid. The
B-validity proof shows the formula to be S5-valid; the T-model shows it to
be K- and D-invalid. We still don’t yet know about S4. So let’s return to the
T-model above, and see what happens when we try to make its accessibility
relation transitive. World a must then see world c, which is impossible since
2P is true in a and P is false in c. So we’re ready for a S4-validity proof (the
proof looks like the B-validity proof at first, but then diverges):
i) Suppose for reductio that V(32P →3232P, r ) = 0, for some world r

in some S4-model 〈W , R, V 〉
ii) So V(32P, r ) = 1 and . . .
iii) V(3232P, r ) = 0
iv) From ii), there’s some world, call it a, such that V(2P, a) = 1 and R ra
v) From iii), since R ra, V(232P, a) = 0
vi) And so, there’s some world, call it b , such that V(32P, b ) = 0 and Rab
vii) By reflexivity, R b b , so given vi), V(2P, b ) = 0
viii) And so, there’s some world, call it c, such that V(P, c) = 0 and R b c.
ix) From vi) and viii), given transitivity, we have Rac. And so, given iv),
V(P, c) = 1, contradicting viii)
Daggers
There’s another kind of step in constructing models. When we make a condi-
tional false, we’re forced to enter certain truth values for its components: 1 for
the antecedent, 0 for the consequent. But consider making a disjunction true. A
disjunction can be true in more than one way. The first disjunct might be true,
or the second might be true, or both could be true. So we have a choice for
how to go about making a disjunction true. Similarly for making a conditional
true, a conjunction false, or a biconditional either true or false.
When one has a choice about which truth values to give the constituents
of a propositional compound, it’s best to delay making the choice as long as
possible. After all, some other part of the model might force you to make one
choice rather than the other. If you investigate the rest of the countermodel,
and nothing has forced your hand, you may need then to make a guess: try one
of the truth value combinations open to you, and see whether you can finish
the countermodel. If not, go back and try another combination.
To remind ourselves of these choice points, we will place a dagger (†) un-
derneath the major connective of the formula in question. Consider, as an
example, constructing a countermodel for the formula 3(3P ∨2Q)→(3P ∨Q).
Throwing caution to the wind and going straight for a T-model, we have after
a few steps:
∗
1 0 0 0 0 0
r
3(3P ∨2Q)→(3P ∨ Q)
∗
0

1 0
a 3P ∨2Q P
†
0
We still have to decide how to make 3P ∨2Q true in world a: which disjunct
to make true? Well, making 2P true won’t require adding another world to
the model, so let’s do that. We have, then, a T-model:
∗
1 0 0 0 0 0
r
3(3P ∨2Q)→(3P ∨ Q)
∗
0

∗
1 1 1 0
3P ∨2Q
a
P
†
0
W = {r, a}
R = {〈r, r〉, 〈a, a〉, 〈r, a〉}
I (Q, a) = 1, all else 0
OK, let’s try now to upgrade this to a B-model. We can’t simply leave
everything as-is while letting world a see back to world r, since 2Q is true
in a and Q is false in r. But there’s another possibility. We weren’t forced to
discharge the dagger in world a by making 2Q true. So let’s explore the other
possibility; let’s make 3P true:
∗
1 0 0 0 0 0
r
3(3P ∨2Q)→(3P ∨ Q)
∗
0 O

1 1 0
a 3P ∨2Q P
∗ †
0 O

1
b
P
0
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈a, r〉, 〈a, b〉, 〈b, a〉}
I (P, b) = 1, all else 0
What about an S4-model? We can’t just add the arrows demanded by

transitivity to our B-model, since 3P is false in world r and P is true in world
b. What we can do instead is revisit the choice of which disjunct of 3P ∨2Q to
make true. Instead of making 3P true, we can make 2Q true, as we did when
we constructed our T-model. In fact, that T-model is already an S4-model.
So, we have countermodels in both S4 and B. The first resulted from
one choice for discharging the dagger in world a, the second from the other
choice. An S5-model, though, looks impossible. When we made the first
choice—making the right disjunct of 3P ∨2Q true—we were able to make the
accessibility relation symmetric, and when we made the second choice—making
the left disjunct of 3P ∨2Q true—we were able to make the accessibility rela-
tion transitive. It would seem to be impossible, then, to make the accessibility
both transitive and symmetric. Here is an S5-validity proof, based on this
reasoning. Note the “separation of cases” reasoning:
i) Suppose for reductio that in some world r in some S5-model, V(3(3P ∨2Q)→(3P ∨Q), r ) =
0. Then V(3(3P ∨2Q), r ) = 1 and …
ii) …V(3P ∨Q, r ) = 0
iii) Given i), for some world a, R ra and V(3P ∨2Q, a) = 1. So, either
V(3P, a) = 1 or V(2Q, a) = 1
iv) The first possibility leads to a contradiction:
a) Suppose V(3P, a) = 1. Then for some world b , Rab and V(P, b ) =

1
b) R is transitive, so given a) and iii), R r b .
c) given ii), V(3P, r ) = 0, and so, given b), V(P, b ) = 0, which contra-
dicts a).
v) So does the second:

a) Suppose V(2Q, a) = 1.
b) R is symmetric. So, given iii), Ra r ; and so, given a), V(Q, r ) = 1
c) But given ii), V(Q, r ) = 0—contradiction.
vi) Either way we have a contradiction.
So we have demonstrated that S5 3(3P ∨2Q)→(3P ∨Q).
Summary of steps
Here, then, is a final list of the steps for constructing countermodels:
1. Place the formula in a box
2. Make the formula false in the world
3. Enter in forced truth values
4. Enter in daggers, and after all forced moves over…
5. Enter asterisks
6. Discharge asterisks (hint: do bottom asterisks first)
7. Back to step 3 if not finished
8. The official model

Exercise 6.2 For each of the following wffs, give a countermodel

for every system in which it is not valid, and give a semantic validity
proof for every system in which it is valid. When you use a single
countermodel or validity proof for multiple systems, indicate which
systems it is good for.
a) 2[P →3(Q→R)]→3[Q→(2P →3R)]
b) 2(P ∨3Q)→(2P ∨3Q)
c) 3(P ∧3Q)→(23P →32Q)
d) 2(P ↔Q)→2(2P ↔2Q)
e) 2(P ∧Q)→22(3P →3Q)
f) 2(2P →Q)→2(2P →2Q)
g) 332P ↔2P
h) 33P →23P
i) 2[2(P →2P )→2P ]→(32P →2P )
6.3.5 Schemas, validity, and invalidity

Let’s digress, for a moment, to clarify the notions of validity and invalidity, as
applied to formulas and schemas.
Formulas—official wffs, that is—are the strings of symbols that are sanc-
tioned by the official rules of grammar of our object language. Formulas of
MPL include P ∨(Q→R), 2P →3Q, and so on. Schemas are devices of our
metalanguage that are used to talk about infinitely many formulas at once. We
used schemas, for instance, in section 2.5 to state the axioms of propositional
logic; we said: “each instance of the schema φ→(ψ→φ) is an axiom of PL”.
And we have used schemas throughout the book to define the notion of a wff,
and to define valuation functions. For example, earlier in this chapter we said
that a valuation function must assign the value 1 to any instance of the schema
∼φ relative to any world iff it assigns 0 to the corresponding instance of φ
relative to that world. Schemas are not formulas, since they contain schematic
variables (“φ” and “ψ” here) that are not part of the primitive vocabulary of
the object language.
When we defined the notion of validity, what we defined was the notion of
a valid formula. (That’s because validity is defined in terms of truth in a model,
which itself was defined only for formulas.) So it’s not, strictly speaking, correct
to apply the notions of validity or invalidity to schemas.
However, it’s often interesting to show that every instance of a given schema
is valid. (Instances are schemas are formulas, and so the notion of validity can
be properly applied to them.) It’s easy, for example, to show that every instance
of the schema 2(φ→φ) is valid in each of our modal systems. (Let φ be any
MPL-wff, and take any world w in any model. Since the rules for evaluating
propositional compounds within possible worlds are the classical ones, φ→φ
must be true at w, no matter what truth value φ has at w. Hence 2(φ→φ) is
true in any world in any model, and so is valid in each system.)
There is, therefore, a kind of indirect notion of schema-validity: validity
of all instances. How about the invalidity of schemas? Here we must take
great care. In particular, the notion of a schema, all of whose instances are
invalid, is not a particularly interesting notion. Take, for instance, the schema
3φ→2φ. We showed earlier that a certain instance of this schema, namely
3P →2P is invalid in each of our systems. However, the schema 2φ→3φ also
has plenty of instances that are valid in various systems. The following formula,
for example, is an instance of 3φ→2ψ, and can easily be shown to be valid in
each of our systems:
3(P →P )→2(P →P )
Thus, even intuitively terrible schemas like 3φ→2φ have some valid instances.
(For an extreme example of this, consider the schema φ. Even this has some
valid instances: P →P , for one.) So it’s not interesting to inquire into whether
each instance of a schema is invalid. What is interesting is to inquire into
whether a given schema has some instances that are invalid. We can show, for
example, that the schema 3φ→2φ has some invalid instances (3P →2P , for
one), and hence is in this way unlike the schema 2(φ→φ).
So when dealing with schemas, it will often be of interest to ascertain
whether each instance of the schema is valid; it will rarely (if ever) be of interest
to ascertain whether each instance of the schema is invalid.
6.4 Axiomatic systems of MPL

We turn next to provability in modal logic. We’ll approach this axiomatically:
we’re going to write down axioms, which are sentences of propositional modal
logic that seem clearly to be logical truths, and we’re going to write down rules
of inference, which say which sentences can be logically inferred from which
other sentences.
We’re going to continue to follow C. I. Lewis in constructing multiple
modal systems, since it’s so unclear which sentences of MPL are logical truths.
Hence, we’ll need to formulate multiple axiomatic systems. These systems will
contain different axioms from one another. As a result, different theorems will
be provable in the different systems.
We will, in fact, give these systems the same names as the systems we
investigated semantically: K, D, T, B, S4, and S5. (Thus we will subscript
the symbol for theoremhood with the names of systems; `K φ, for example,
will mean that φ is a theorem of system K.) Our re-use of the system names
will be justified in sections ?? and ??, where we will establish soundness and
completeness for each system. Given soundness and completeness, for each
system, exactly the same formulas are provable as are valid.
6.4.1 System K
Our first system, K, is the weakest system—i.e., the system with the fewest
theorems.
Axiomatic system K:
· Rules: modus ponens and necessitation:
φ→ψ φ φ
MP NEC
ψ 2φ
· Axioms: for any MPL-wffs φ, ψ, and χ , the following are axioms:
φ→(ψ→φ) (A1)
(φ→(ψ→χ ))→((φ→ψ)→(φ→χ )) (A2)
(∼ψ→∼φ)→((∼ψ→φ)→ψ) (A3)
2(φ→ψ)→(2φ→2ψ) (K)
As before, a proof is defined as a series of wffs, each of which is either an

axiom or follows from earlier lines in proof by a rule, and a theorem is defined
as the last line of any proof.
This axiomatic system (like all the modal systems we will study) is an exten-
sion of propositional logic, in the sense that it includes all of the theorems of
propositional logic, but then adds more theorems. It includes all of proposi-
tional logic because one of its rules is the propositional logic rule MP, and each
propositional logic axiom is one of its axioms. It adds theorems by adding a
new rule of inference (NEC), and a new axiom schema (the K-schema) (as well
as adding new wffs—wffs containing the 2—to the stock of wffs that can occur
in the PL axioms.)
The rule of inference, NEC (for “necessitation”), says that if you have a
formula φ on a line, then you may infer the formula 2φ. This may seem
unintuitive. After all, can’t a sentence be true without being necessarily true?
Yes; but the rule of necessitation doesn’t contradict this. Remember that every
line in every axiomatic proof is a theorem. So whenever one uses necessitation in
a proof, one is applying it to a theorem. And necessitation does seem appropriate
when applied to theorems: if φ is a theorem, then 2φ ought also to be a theorem.
Think of it this way. The worry about the rule of necessitation is that it isn’t a
truth-preserving rule: its premise can be true when its conclusion is false. The
answer to the worry is that while necessitation doesn’t preserve truth, it does
preserve logical truth, which is all that matters in the present context. For in
the present context, we’re only using NEC in a definition of theoremhood.
We want our theorems to be, intuitively, logical truths; and provided that our
axioms are all logical truths and our rules preserve logical truth, the definition
will yield only logical truths as theorems. We will return to this issue.
Let’s investigate what one can prove in K. The simplest sort of distinctively
modal proof consists of first proving something from the PL axioms, and then
necessitating it, as in the following proof of 2((P →Q)→(P →P ))
1. P →(Q→P ) (A1)
2. P →(Q→P ))→((P →Q)→(P →P )) (A2)
3. (P →Q)→(P →P ) 1,2 MP
4. 2((P →Q)→(P →P )) 3, NEC
Using this technique, we can prove anything of the form 2φ, where φ
is provable in PL. And, since the PL axioms are complete (section 2.7), that
means that we can prove 2φ whenever φ is a tautology—i.e., a valid wff of
PL. But constructing proofs from the PL axioms is a pain in the neck!—and
anyway not what we want to focus on in this chapter. So let’s introduce the
following time-saving shortcut. Instead of writing out proofs of tautologies,
let’s instead allow ourselves to write any PL tautology at any point in a proof,
annotating simply “PL”.5 Thus, the previous proof could be shortened to:
1. (P →Q)→(P →P ) PL
2. 2((P →Q)→(P →P )) 1, NEC
Furthermore, consider the wff 2P →2P . Clearly, we can construct a proof

of this wff from the PL axioms: begin with any proof of the tautology Q→Q
from the PL axioms, and then construct a new proof by replacing each occur-
rence of Q in the first proof with 2P . (This is a legitimate proof, even though
2P isn’t a wff of propositional logic, because when we stated the system K, the
schematic letters φ, ψ, and χ in the PL axioms are allowed to be filled in with
any wffs of MPL, not just wffs of PL.) So let us also include lines like this in
our modal proofs:
2P →2P PL
Why am I making such a fuss about this? Didn’t I just say in the previous
paragraph that we can write down any tautology at any time, with the annotation
“PL”? Well, strictly speaking, 2P →2P isn’t a tautology. A tautology is a valid
wff of PL, and 2P →2P isn’t even a wff of PL (since it contains a 2). But it
is the result of beginning with some PL-tautology (Q→Q, in this case) and
uniformly changing sentence letters to chosen modal wffs (in this case, Qs to
2P s); hence any proof of the PL tautology may be converted into a proof of
it; hence the “PL” annotation is just as justified here as it is in the case of a
genuine tautology. So in general, MPL wffs that result from PL tautologies in
this way may be written down and annotated “PL”.
Back to investigating what we can prove in K. As we’ve seen, we can prove
that tautologies are necessary—we can prove 2φ whenever φ is a tautology.
One can also prove in K that contradictions are impossible. For instance,
∼3(P ∧∼P ) is a theorem of K:
5
How do you know whether something is a tautology? Figure it out any way you like: do a
truth table, or a natural deduction derivation—whatever.
1. ∼(P ∧∼P ) PL
2. 2∼(P ∧∼P ) 1, NEC
3. 2∼(P ∧∼P )→∼∼2∼(P ∧∼P ) PL
4. ∼∼2∼(P ∧∼P ) 2, 3, MP
But line 4 is a definitional abbreviation of ∼3(P ∧∼P ).

Let’s introduce another time-saving shortcut. Note that the move from 2 to
4 in the previous proof is just a move from a formula to a propositional logical
consequence of that formula. Let’s allow ourselves to move directly from any
lines in a proof, φ1 . . . φn , to any propositional logical consequence ψ of those
lines, by “PL”. Thus, the previous proof could be shorted to:
1. ∼(P ∧∼P ) PL
2. 2∼(P ∧∼P ) 1, NEC
3. ∼∼2∼(P ∧∼P ) 2, PL
Why is this legitimate? Suppose that ψ is a propositional logical semantic

consequence of φ1 . . . φn . Then the conditional φ1 →(φ2 → · · · (φn →ψ) . . . ) is
a PL-valid formula, and so, given the completeness of the PL axioms, is a
theorem of K. That means that if we have φ1 , . . . , φn in an axiomatic K-proof,
then we can always prove the conditional φ1 →(φ2 → · · · (φn →ψ) . . . ) using the
PL-axioms, and then use MP repeatedly to infer ψ. So inferring ψ directly,
and annotating “PL”, is justified. (As with the earlier “PL” shortcut, let’s use
this shortcut when the conditional φ1 →(φ2 → · · · (φn →ψ) . . .) results from some
tautology by uniform substitution, even if it contains modal operators and so
isn’t strictly a tautology.)
So far our modal proofs have only used necessitation and the PL axioms.
What about the K-axioms? The point of the K-schema is to enable “distribution
of the 2 over the →”. That is, if you ever have the formula 2(φ→ψ), then you
can always move to 2φ→2ψ as follows:
i . 2(φ→ψ)
i + 1. 2(φ→ψ)→(2φ→2ψ) K axiom
i + 2. 2φ→2ψ i, i + 1, MP
Distribution of the 2 over the →, plus the rule of necessitation, combine

to give us a powerful proof strategy. Whenever one can prove the conditional
φ→ψ, then one can prove the modal conditional 2φ→2ψ as well, as follows.
First prove φ→ψ, then necessitate it to get 2(φ→ψ), then distribute the 2
over the arrow to get 2φ→2ψ. This procedure is one of the core K-strategies,
and is featured in the following proof of 2(P ∧Q)→(2P ∧2Q):
1. (P ∧Q)→P PL
2. 2[(P ∧Q)→P ] NEC
3. 2[(P ∧Q)→P ]→[2(P ∧Q)→2P ] K axiom
4. 2(P ∧Q)→2P 3,4 MP
5. 2(P ∧Q)→2Q Insert steps similar to 1-4
6. 2(P ∧Q)→(2P ∧2Q) 4,5, PL
Notice that the preceding proof, like all of our proofs since we introduced
the time-saving shortcuts, is not a K-proof in the official defined sense. Lines 1,
5, and 6 are not axioms, nor do they follow from earlier lines by MP or NEC;
similarly for line 6.6 So what kind of “proof” is it? It’s a metalanguage proof:
an attempt to convince the reader, by acceptable standards of rigor, that some
real K-proof exists. A reader could use this metalanguage proof as a blueprint
for constructing a real proof. She would begin by replacing line 1 with a proof
from the PL axioms of the conditional (P ∧Q)→P . (As we know from chapter
??, this could be a real pain in the neck!—but the completeness of PL assures us
that it is possible.) She would then replace line 5 with lines parallel to lines 1-4,
but which begin with a proof of (P ∧Q)→Q rather than (P ∧Q)→P . Finally,
in place of line 6, she would insert a proof from the PL axioms of the sen-
tence (2(P ∧Q)→2P )→[(2(P ∧Q)→2Q)→(2(P ∧Q)→(2P ∧2Q))], and then
use modus ponens twice to infer 2(P ∧Q)→(2P ∧2Q).
Let’s introduce another time-saving shortcut, which we’ll use more and
more as we progress: doing two (or more) steps at once. This shortcut is
featured in the following proof of (2P ∨2Q)→2(P ∨Q):
1. P →(P ∨Q) PL
2. 2(P →(P ∨Q)) 1, NEC
3. 2(Q→(P ∨Q)) PL, NEC
4. 2P →2(P ∨Q) 2, K
5. 2Q→2(P ∨Q) 3, K
6. (2P ∨2Q)→2(P ∨Q) 4,5 PL
6
A further (even pickier) reason: the symbol ∧ isn’t allowed in wffs; the sentences in the
proof are mere abbreviations for official MPL-wffs.
Line 3 is really short for:
3a. Q→(P ∨Q) PL

3b. 2(Q→(P ∨Q)) 3a, NEC
And line 4 is short for:
4a. 2(P →(P ∨Q))→(2P →2(P ∨Q)) K axiom

4b. 2P →2(P ∨Q) 2, 4a, MP
One further comment about this last proof: it illustrates a strategy that is
common in modal proofs. We were trying to prove a conditional formula
whose antecedent is a disjunction of two modal formulas. But the modal
techniques we had developed didn’t deliver formulas of this form. They only
showed us how to put 2s in front of PL-tautologies, and how to distribute 2s
over →s. They only yield formulas of the form 2φ and 2φ→2ψ, whereas the
formula we were trying to prove looks different. To overcome this problem,
what we did was to use the modal techniques to prove two conditionals, namely
2P →2(P ∨Q) and 2Q→2(P ∨Q), from which the desired formula, namely
(2P ∨2Q)→2(P ∨Q), follows by propositional logic. The trick, in general, is
this: remember that you have PL at your disposal. Simply look for one or
more modal formulas you know how to prove which, by PL, imply the formula
you want. Assemble the desired formulas, and then write down your desired
formula, annotating “PL”. In doing so, it may be helpful to recall PL inferences
like the following:
φ→ψ ψ→φ φ→(ψ→χ ) φ→ψ φ→χ

φ↔ψ (φ∧ψ)→χ φ→(ψ∧χ )
φ→χ ψ→χ φ→ψ φ→∼ψ

(φ∨ψ)→χ ∼φ∨ψ ∼(φ∧ψ)
The next example illustrates our next major modal proof technique: com-
bining two 2 statements to get a single 2 statement. Let us construct a K-proof
of (2P ∧2Q)→2(P ∧Q):
1. P →(Q→(P ∧Q)) PL
2. 2[P →(Q→(P ∧Q))] NEC
3. 2P →2(Q→(P ∧Q)) 2, K
4. 2(Q→(P ∧Q))→[2Q→2(P ∧Q)] K axiom
5. 2P →[2Q→2(P ∧Q)] 3,4 PL
6. (2P ∧2Q)→2(P ∧Q) 5, PL
(If you wanted to, you could skip step 5, and just go straight to 6 by propositional
logic, since 6 is a propositional logical consequence of 3 and 4; I put it in for
perspicuity.)
The general technique illustrated by the last problem applies anytime you
want to move from several 2 statements to a further 2 statement, where the in-
side parts of the first 2 statements imply the inside part of the final 2 statement.
More carefully: it applies whenever you want to prove a formula of the form
2φ1 →(2φ2 → · · · (2φn →2ψ) . . . ), provided you are able to prove the formula
φ1 →(φ2 → · · · (φn →ψ) . . . ). (The previous proof was an instance of this because
it involved moving from 2P and 2Q to 2(P ∧Q); and this is a case where one
can move from the inside parts of the first two formulas (namely, P and Q), to
the inside part of the third formula (namely, P ∧Q)—by PL.) To do this, one
begins by proving the conditional φ1 →(φ2 → · · · (φn →ψ) . . . ), necessitating it to
get 2[φ1 →(φ2 → · · · (φn →ψ) . . . )], and then distributing the 2 over the arrows
repeatedly using K-axioms and PL to get 2φ1 →(2φ2 → · · · (2φn →2ψ) . . . ).
One cautionary note in connection with this last proof. One might think to
make it more intuitive by using conditional proof:
1. 2P ∧2Q assume for conditional proof

2. 2P 1, PL
3. 2Q 1, PL
4. P →(Q→(P ∧Q)) PL
5. 2[P →(Q→(P ∧Q))] NEC
6. 2P →2(Q→(P ∧Q)) 5, K
7. 2(Q→(P ∧Q)) 6,2, MP
8. 2Q→2(P ∧Q) 7,K
9. 2(P ∧Q) 3,8 MP
10. (2P ∧2Q)→2(P ∧Q) 1-9, conditional proof
But this is not a legal proof, since our axiomatic system allows neither assump-
tions nor conditional proof.
In fact, our decision to omit conditional proof was not at all arbitrary. Given
our rule of necessitation, we couldn’t add conditional proof to our system. If we
did, proofs like the following would become legal:
1. P assume for conditional proof

2. 2P 1, NEC
3. P →2P 1,2, conditional proof
Thus, P →2P would turn out to be a K-theorem. But we don’t want that: after
all, a statement P might be true without being necessarily true.
Once we have a soundness proof (section 6.5), we’ll be able to show that
P →2P isn’t a K-theorem. But as we just saw, one can construct a K-proof from
{P } of 2P (recall the notion of a proof from a set Γ, from section 2.5.) It follows
that the deduction theorem (section 2.7), which says that if there exists a proof
of ψ from {φ}, then there exists a proof of φ→ψ, fails for K (it likewise fails for
all the modal systems we will consider.) So there will be no conditional proof
in our axiomatic modal systems. (Of course, to convince yourself that a given
formula is really a tautology of propositional logic, you may sketch a proof of it
to yourself using conditional proof in some standard natural deduction system
for nonmodal propositional logic; and then you may write that formula down
in one of our axiomatic MPL proofs, annotating “PL”.)
Back to techniques for constructing proofs in K. The following proof of
22(P ∧Q)→22P illustrates a technique for proving formulas with “nested”
modal operators:
1. (P ∧Q)→P PL
2. 2(P ∧Q)→2P 1, NEC, K
3. 2[2(P ∧Q)→2P ] 2, NEC
4. 22(P ∧Q)→22P 3, K
Notice in line 3 that we necessitated something that was not a PL theorem.

That’s ok; we’re allowed to necessitate any K-theorems, even those whose proofs
were distinctly modal. Notice also how this proof contains two instances of our
basic K-strategy. This strategy involves obtaining a conditional, necessitating
it, then distributing the 2 over the →. We did this first using the conditional
(P ∧Q)→P ; that led us to a conditional, 2(P ∧Q)→2P . Then we started the

strategy over again, using this as our initial conditional.
So far we have no techniques dealing with the 3, other than eliminating it
by definition. It will be convenient to derive some shortcuts. For one, there
are the following theorem schemas, which may collectively be called “modal
negation”, or “MN” for short:
`K ∼2φ→3∼φ `K 3∼φ→∼2φ
`K ∼3φ→2∼φ `K 2∼φ→∼3φ
I’ll do one of these; the rest can be done as exercises.
Example 6.4: Prove ∼2φ→3∼φ (one of the MN theorems):
1. ∼∼φ→φ PL
2. 2∼∼φ→2φ 1, NEC, K
3. ∼2φ→∼2∼∼φ 2, PL
The final line, 3, is the definitional equivalent of ∼2φ→3∼φ.
Exercise 6.3 Prove the remaining MN theorems.
It will also be worthwhile to know that an analog of the K axiom for the 3
is a K-theorem:
2(φ→ψ)→(3φ→3ψ) (K3)
K3 is, by definition of the 3, the same formula as:
2(φ→ψ)→(∼2∼φ→∼2∼ψ)
How are we going to construct a K-proof of this theorem? In a natural de-

duction system we would use conditional proof and reductio ad absurdum,
but these strategies are not available to us here. What we must do instead is
look for a formula we know how to prove in K, which is PL-equivalent to the
formula we want to prove. Here is such a formula:
2(φ→ψ)→(2∼ψ→2∼φ)
This is equivalent, given PL, to what we want to show; and it looks like the
result of necessitating a tautology and then distributing the 2 over the → a
couple times—just the kind of thing we know how to do in K. Here, then, is
the desired proof of K3:
1. (φ→ψ)→ (∼ψ→∼φ) PL
2. 2(φ→ψ)→2(∼ψ→∼φ) 1, NEC, K
3. 2(∼ψ→∼φ)→(2∼ψ→2∼φ) K
4. 2(φ→ψ)→(2∼ψ→2∼φ) 2,3 PL
5. 2(φ→ψ)→(∼2∼φ→∼2∼ψ) 4,PL
In doing proofs, let’s also allow ourselves to refer to earlier theorems proved,
rather than repeating their proofs. The importance of K3 may be illustrated
by the following proof of 2P →(3Q→3(P ∧Q)):
1. P →[Q→(P ∧Q)] PL
2. 2P →2[Q→(P ∧Q)] 1, NEC, K
3. 2[Q→(P ∧Q)]→[3Q→3(P ∧Q)] K3
4. 2P →[3Q→3(P ∧Q)] 2,3, PL
In general, K3 lets us construct proofs of the following sort. Suppose we

wish to prove a formula of the form:
O1 φ1 →(O2 φ2 →(. . . →(On φn →3ψ) . . .)
where the Oi s are modal operators, all but one of which are 2s. (Thus, the
remaining Oi is the 3.) This can be done, provided that ψ is provable in K from
the φi s. The basic strategy is to prove a nested conditional, the antecedents
of which are the φi s, and the consequent of which is ψ; necessitate it; then
repeatedly distribute the 2 over the →s, once using K3, the rest of the times
using K. But there is one catch. We need to make the application of K3 last,
after all the applications of K. This in turn requires the conditional we use to
have the φi that is underneath the 3 as the last of the antecedents. For instance,
suppose that φ3 is the one underneath the 3. Thus, what we are trying to
prove is:
2φ1 →(2φ2 →(3φ3 →(2φ4 →(. . . →(2φn →3ψ) . . .)

In this case, the conditional to use would be:
φ1 →(φ2 →(φn →(φ4 →(. . . →(φn−1 →(φ3 →ψ) . . .)
In other words, one must swap one of the other φi s (I arbitrarily chose φn )
with φ3 . What one obtains at the end will therefore have the modal statements
out of order:
2φ1 →(2φ2 →(2φn →(2φ4 →(. . . →(2φn−1 →(3φ3 →3ψ) . . .)
But that problem is easily solved; this is equivalent in PL to what we’re trying
to get. (Recall that φ→(ψ→χ ) is logically equivalent in PL to ψ→(φ→χ ).)
Why do we need to save K3 for last? The strategy of successively distribut-
ing the box over all the nested conditionals comes to a halt as soon as the K3
theorem is used. Let me illustrate with an example. Suppose we wish to prove
`K 3P →(2Q→3(P ∧Q)). We might think to begin as follows:
1. P →(Q→(P ∧Q)) PL
2. 2[P →(Q→(P ∧Q))] 1, Nec
3. 3P →3(Q→(P ∧Q)) K3, 2, MP
4. ?
But now what? What we need to finish the proof is:
3(Q→(P ∧Q))→(2Q→3(P ∧Q)).
But neither K nor K3 gets us this. The remedy is to begin the proof with a
different conditional:
1. Q→(P →(P ∧Q)) PL
2. 2(Q→(P →(P ∧Q))) 1, Nec
3. 2Q→2(P →(P ∧Q)) K, 2, MP
4. 2(P →(P ∧Q))→(3P →3(P ∧Q)) K3
5. 2Q→(3P →3(P ∧Q)) 3, 4, PL
6. 3P →(2Q→3(P ∧Q)) 5, PL
One can, then, prove a number of theorems in K. Nevertheless, K is a

very weak system. You can’t prove the formula 2P →3P in K. (We’ll be able
to demonstrate this after section 6.5.) Relatedly, one can’t prove in K that
tautologies are possible or that contradictions aren’t necessary.
Exercise 6.4 Give axiomatic proofs in K of the following formulas:
a) 3(P ∧Q)→(3P ∧3Q)
b) 2∼P →2(P →Q)
c) ∼3(Q∧R)↔2(Q→∼R)
d) 2(P ↔Q)→(2P ↔2Q)
e) [2(P →Q)∧2(P →∼Q)]→∼3P
f) (2P ∧2Q)→2(P ↔Q)
g) 3(P →Q)↔(2P →3Q)
h) 3P →(2Q→3Q)
6.4.2 System D
System D results from adding a new axiom schema to system K:
Axiomatic system D:
· Rules: MP, NEC
· Axioms: the A1, A2, A3, and K schemas, plus the D-schema:
2φ→3φ (D)
Notice that since system D includes all the K axioms and rules, we retain all
the K-theorems. The addition of the D-schema just adds more theorems. In
fact, all of our systems will build on K in this way, by adding new axioms to K.
With the D-schema in place, we can now prove that tautologies are possible:
1. P ∨∼P PL
2. 2(P ∨∼P ) 1, NEC
3. 2(P ∨∼P )→3(P ∨∼P ) D
4. 3(P ∨∼P ) 2,3 MP
Example 6.5: Show that `D 22P →23P .
1. 2P →3P D
2. 2(2P →3P ) 1, NEC
3. 22P →23P 2, K
Like K, system D is very weak. As we will see later, we can’t prove 2φ→φ
in D. Therefore, D doesn’t seem to be a correct logic for metaphysical, or
nomic, or technological necessity, for surely, if something is metaphysically,
nomically, or technologically necessary, then it must be true. (If something is
true in all metaphysically possible worlds, or all nomically possible worlds, or
all technologically possible worlds, then surely it must be true in the actual
world, and so must be plain old true.) But perhaps there is some interest in
D anyway; perhaps D is a correct logic for moral necessity. Suppose we read
2φ as “One ought to make φ be the case”, and, correspondingly, read 3φ as
“One is permitted to make φ be the case”. Then the fact that 2φ→φ cannot
be proved in D would be a virtue, for from the fact that something ought be
done, it certainly doesn’t follow that it is done. The D-axiom, on the other
hand, would correspond to the principle that if something ought to be done
then it is permitted to be done, which does seem like a logical truth. But I won’t
go any further into the question of whether D in fact does give a correct logic
for moral necessity.
Exercise 6.5 Give axiomatic proofs in D of the following formulas:
a) ∼(2P ∧2∼P )
b) ∼2[2(P ∧Q)∧2(P →∼Q)]
6.4.3 System T
T is the first system we have considered that has any plausibility of being a
correct logic for a wide range of concepts of necessity (metaphysical necessity,
for example):
Axiomatic system T:
· Rules: MP, NEC
· Axioms: the A1, A2, A3, and K schemas, plus the T-schema:
2φ→φ (T)
Recall that in the case of K, we proved a theorem schema, K3, which was
the analog for the 3 of the K-axiom schema. Let’s do the same thing here; let’s
prove a theorem schema T3, which is the analog for the 3 of the T axiom
schema:
T3: φ→3φ
1. 2∼φ→∼φ T
2. φ→∼2∼φ 1, PL
2 is just the definition of φ→3φ. Thus, we have established that for every
wff φ, `T φ→3φ. So let’s allow ourselves to write down formulas of the form
φ→3φ, annotating simply “T3”.
Notice that instances of the D-axioms are now theorems: 2φ→φ is a T
axiom, we just proved that φ→3φ is a theorem; and from these two by PL we
can prove 2φ→3φ. Thus, T is an extension of D: every theorem of D remains
a theorem of T. (Since D was an extension of K, T too is an extension of K.)
Exercise 6.6 Give axiomatic proofs in T of the following formulas:
a) 32P →3(P ∨Q)
b) 2P ∧32(P →Q)]→3Q
c) 3(P →2Q)→(2P →3Q)
6.4.4 System B
Our systems so far don’t allow us to prove anything interesting about iterated
modalities, i.e., sentences with consecutive boxes or diamonds. Which such
sentences should be theorems? The B axiom schema decides some of these
questions for us; here is system B:
Axiomatic system B:
· Rules: MP, NEC

· Axioms: the A1, A2, A3, K, and T schemas, plus the B-schema:
32φ→φ (B)
Note that we retain the T axiom schema in B. Thus, B is an extension of T

(and hence of K and D as well.)
As with K and T, we can establish a theorem schema that is the analog for
the 3 of B’s characteristic axiom schema. (The proof illustrates techniques for
“moving” ∼s through strings of modal operators.)
B3: φ→23φ:
1. 32∼φ→∼φ B
2. φ→∼32∼φ 1, PL
3. ∼32∼φ↔2∼2∼φ MN
4. ∼2∼φ→3φ PL (since 3 abbreviates ∼2∼)
5. 2∼2∼φ→23φ 4, NEC, K, MP
6. φ→23φ 2, 3, 5, PL
Example 6.6: Show that `B [2P ∧232(P →Q)]→2Q.
1. 32(P →Q)→(P →Q) B

2. 232(P →Q)→2(P →Q) 1, Nec, K, MP
3. 2(P →Q)→(2P →2Q) K
4. 232(P →Q)→(2P →2Q) 2, 3 PL
5. [2P ∧232(P →Q)]→2Q 4, PL
Exercise 6.7 Give axiomatic proofs in B of the following formulas:
a) 32P ↔3232P
b) [2P ∧232(P →Q)]→2Q

6.4.5 System S4
The characteristic axiom of our next system, S4, is a different principle gov-
erning iterated modalities:
Axiomatic system S4:
· Rules: MP, NEC

· Axioms: the A1, A2, A3, K, and T schemas, plus the S4-schema:
2φ→22φ (S4)
S4 contains the S4-schema but does not contain the B-schema. Symmetri-
cally, B lacks the S4-schema, but of course contains the B-schema. As a result,
some instances of the B-schema are not provable in S4, and some instances of
the S4-schema are not provable in B (we’ll be able to show this after section
6.5). Hence, although S4 and B are each extensions of T, neither B nor S4 is
an extension of the other.
As before, we have a theorem schema that is the analog for the 3 of the S4
axiom schema:
S43: 33φ→3φ:
1. 2∼φ→22∼φ S4
2. 2∼φ→∼3φ MN
3. 22∼φ→2∼3φ 2, NEC, K, MP
4. 2∼3φ→∼33φ MN
5. ∼3φ→2∼φ MN
6. 33φ→3φ 5,1,3,4, PL
Example 6.7: Show that `S4 (3P ∧2Q)→3(P ∧2Q). This problem is rea-
sonably difficult. My approach is as follows. We saw in the K section above that
the following sort of thing may always be proved: 2φ→(3ψ→3χ ), whenever
the conditional φ→(ψ→χ ) can be proved. So we need to try to work the
problem into this form. As-is, the problem doesn’t quite have this form. But
something very related does have this form, namely: 22Q→(3P →3(P ∧2Q))
(since the conditional 2Q→(P →(P ∧2Q)) is a tautology). This thought in-
spires the following proof:
1. 2Q→(P →(P ∧2Q)) PL

2. 22Q→2(P →(P ∧2Q)) 1, Nec, K, MP
3. 2(P →(P ∧2Q))→(3P →3(P ∧2Q)) K3
4. 22Q→(3P →3(P ∧2Q)) 2, 3 PL
5. 2Q→22Q S4
6. (3P ∧2Q)→3(P ∧2Q) 4, 5 PL
Exercise 6.8 Give axiomatic proofs in S4 of the following formulas:
a) 2P →232P
b) 2323P →23P
c) 32P →3232P
6.4.6 System S5
Here, instead of the B or S4 schemas, we add the S5 schema to T:
Axiomatic system S4:
· Rules: MP, NEC
· Axioms: the A1, A2, A3, K, and T schemas, plus the S5-schema:
32φ→2φ (S5)
First let’s prove the analog of the S5-schema for the 3:
S53: 3φ→23φ
1. 32∼φ→2∼φ S5
2. ∼3φ→2∼φ MN
3. 3∼3φ→32∼φ 2, NEC, K3, MP
4. ∼23φ→3∼3φ MN
5. 2∼φ→∼3φ MN
6. 3φ→23φ 4,3,1,5, PL
Next, note that the B and S4 axioms are now derivable as theorems. The B
axiom, 32φ→φ, is trivial:
1. 32φ→2φ S5
2. 2φ→φ T
3. 32φ→φ 1,2 PL
And now the S4 axiom, 2φ→22φ. This is a little harder. I used the B3
theorem, which we can now appeal to since the theoremhood of the B-schema
has been established.
1. 2φ→232φ B3
2. 32φ→2φ S5
3. 2(32φ→2φ) 2, Nec
4. 232φ→22φ 3, K, MP
5. 2φ→22φ 4, 1, PL
Exercise 6.9 Give axiomatic proofs in S5 of the following formulas:
a) (2P ∨3Q)↔2(P ∨3Q)
b) 3(P ∧3Q)↔(3P ∧3Q)
c) 2(2P →2Q)∨2(2Q→2P )
d) 2[2(3P →Q)↔2(P →2Q)]
6.4.7 Substitution of equivalents and modal reduction

Let’s conclude our discussion of provability in modal logic by proving two
simple meta-theorems.
Substitution of equivalents: Where S is any of our modal systems, and wff

χ β results from wff χ by changing occurrences of wff α to occurrences of wff
β:
if `S α↔β, then `S χ ↔χ β
Proof. Suppose (*) `S α↔β. We’ll argue by induction that `S χ ↔χ β .

Base case: here χ is a sentence letter. Then either i) changing αs to βs has
no effect, in which case χ β is just χ , in which case obviously `S (χ ↔χ β ); or
ii) χ is α, in which case χ β is β, and we know `S (χ ↔χ β ) since we are given
(*).
Induction case: We now assume the result holds for some formulas χ1 and
β β
χ2 —that is, we assume that `S χ1 ↔χ1 and `S χ2 ↔χ2 —and we show the
result holds for ∼χ1 , 2χ1 , and χ1 →χ2 .
Take the first case. We must show that the result holds for ∼χ1 —i.e., we
β
must show that `S ∼χ1 ↔(∼χ1 )β . (∼χ1 )β is just ∼χ1 , so we must show `S
β β β
∼χ1 ↔∼χ1 . But ∼χ1 ↔∼χ1 follows by PL from χ1 ↔χ1 , and the inductive
β
hypothesis tells us that: `S χ1 ↔χ1 .
Take the second case. We must show `S (χ1 →χ2 )β . The inductive hypoth-
β
esis tells us that `S χ1 ↔χ1 , and so (since S includes PL):
`S (χ1 →χ2 ) ↔ (χ1 β →χ2 )
The inductive hypothesis also tells us that `S (χ2 ↔χ2 β ), from which, using
propositional logic in S, we obtain:
`S (χ1 β →χ2 ) ↔ (χ1 β →χ2 β )
Now from the two displayed equivalences, again using propositional logic in S,
we have:
`S (χ1 →χ2 ) ↔ (χ1 β →χ2 β )
But note that (χ1 β →χ2 β ) is just the same formula as (χ1 →χ2 )β . So we’ve shown
what we wanted to show.
β
Finally, take the third case. We must show that `S 2χ1 ↔2χ1 . This follows
β
from the inductive hypothesis `S χ1 ↔χ1 . For the inductive hypothesis implies
β β
`S χ1 →χ1 , by PL; and then, using NEC and a K-axiom, we have `S 2χ1 →2χ1 .
β
A parallel argument establishes `S 2χ1 →2χ1 ; and then the desired conclusion
follows by using PL. That completes the inductive proof.
The following examples illustrate the power of substitution of equivalents.
In our discussion of K we proved the following two theorems:
2(P ∧Q)→(2P ∧2Q)
(2P ∧2Q)→2(P ∧Q)
Hence (by PL), 2(P ∧Q)↔(2P ∧2Q) is a K-theorem. Given substitution

of equivalents, whenever we prove a theorem in which the formula 2(P ∧Q)
occurs as a subformula, we can infer that the result of changing 2(P ∧Q) to
2P ∧2Q is also a K-theorem—without having to do a separate proof.
Similarly, given the modal negation theorems, we know that all instances
of the following schemas are theorems of K (and hence of every other system):
2∼φ↔∼3φ
3∼φ↔∼2φ
Call these “the duals equivalences”.7 Given the duals equivalences, we can swap
∼3φ and 2∼φ, or ∼2φ and 3∼φ, within any theorem of any system, and
the result will also be a theorem of that system. So we can “move” ∼s through
series of modal operators at will. For example, it’s easy to show that each of the
following is a theorem of each system S:
332∼φ↔332∼φ (1)
33∼3φ↔332∼φ (2)
3∼23φ↔332∼φ (3)
∼223φ↔332∼φ (4)
(1) is a theorem of S, since it has the form ψ→ψ. (2) is the result of changing
2∼φ on the left of (1) to ∼3φ. Since (1) is a theorem of S, (2) is also a theorem
of S, by substitution of equivalents via a duals equivalence. We then obtain (3)
by changing 3∼3φ in (2) to ∼23φ; by substitution of equivalents via a duals
equivalence, this too is a theorem of S. Finally, (4) follows from (3) and a MN
theorem by PL, so it too is a theorem of S. (Note how this sort of technique
greatly simplifies the process of establishing the existence of theorems such as
K3, T3, B3, S43, and S53!)
Our second meta-theorem concerns only system S5:
Modal reduction theorem for S5: Where O1 . . . On are modal operators and
φ is a wff:
`S5 O1 . . . On φ↔On φ
7
Given the duals equivalences, the 2 is related to the 3 the way the ∀ is related to the
∃ (since ∀x∼φ↔∼∃xφ, and ∃x∼φ↔∼∀xφ are logical truths). This shared relationship,
which holds between the 2 and the 3, and between the ∀ and the ∃, is called “duality”; 2 and
3 are said to be duals, as are ∀ and ∃. This logical analogy would be neatly explained by a
metaphysics according to which necessity just is truth in all worlds and possibility just is truth
in some worlds!
That is, whenever a formula has a string of modal operators in front, it is always
equivalent to the result of deleting all the modal operators except the innermost
one. For example, 223232232323φ and 3φ are provably equivalent in
S5; i.e., 223232232323φ↔3φ is a theorem of S5). This follows from
the fact that the following equivalences are all theorems of S5:
32φ↔2φ (a)
22φ↔2φ (b)
23φ↔3φ (c)
33φ↔3φ (d)
The left-to-right direction of (a) is just S5; the right-to-left is T3; (b) is T and
S4; (c) is T and S53; and (d) is S43 and T3. Thus, by repeated applications
of these equivalences, using substitution of equivalents, we can reduce strings
of modal operators to the innermost operator. (It is straightforward to convert
this argument into a more rigorous inductive proof.)8
6.5 Soundness in MPL9

At this point, we have defined twelve logical systems: six semantic systems and
six axiomatic systems. But each semantic system was paired with an axiomatic
system to which we gave the same name. The time has come to justify this
pairing. In this section and the next, we show that for each semantic system,
exactly the same wffs are counted valid in that system as are counted theorems
by the axiomatic system of the same name. That is, for each of our systems, S
(for S = K, D, T, B, S4, and S5), we will prove soundness and completeness:
S-soundness: every S-theorem is S-valid
S-completeness: every S-valid formula is a S-theorem

8
The modal reduction formula, the duals equivalences, and substitution of equivalents
together let us “reduce” strings of operators that include ∼s as well as modal operators. Simply
use the duals equivalents to drive any ∼s in the string to the far right hand side, then use the
modal reduction theorem to eliminate all but the innermost modal operator.
9
The proofs of soundness and completeness in this and the next section are from Cresswell
and Hughes (1996).
Our study of modal logic has reversed that of history. We began with semantics,
because that is the more intuitive approach. Historically (as we noted earlier),
the axiomatic systems came first, in the work of C. I. Lewis. Given the uncer-
tainty over what formulas ought to be counted as axioms, modal logic was in
disarray. The discovery by the teenaged Saul Kripke in the late 1950s of the
possible-worlds semantics we studied in section 6.3, and of the correspondence
between simple constraints (reflexivity, transitivity, etc.) on the accessibility
relation in his models and Lewis’s axiomatic systems, was a major advance in
the history of modal logic.
The soundness and completeness theorems have practical as well as the-
oretical value. First, once we’ve proved soundness, we will for the first time
have a method for establishing that a given formula is not a theorem: construct
a countermodel for that formula, thus establishing that the formula is not valid,
and then conclude via soundness that the formula is not a theorem. Second,
given completeness, if we want to know that a given formula is a theorem, it
suffices to show that it is valid. Since semantic validity proofs are comparatively
easy to construct, it’s nice to be able to use them rather than axiomatic proofs.
Let’s begin with soundness. We’re going to prove a general theorem, which
we’ll use in several soundness proofs. First we’ll need a piece of terminology.
Where Γ is any set of modal wffs, let’s call “K+Γ” the axiomatic system that
consists of the same rules of inference as K (MP and NEC), and which has
as axioms the axioms of K (instances of the K- and PL- schemas), plus the
members of Γ. Here, then, is the theorem:
Theorem 6.1 If Γ is any set of modal wffs and M is an MPL-model in which

each wff in Γ is valid, then every theorem of K+Γ is valid in M
Modal systems of the form K+Γ are commonly called normal. Normal
modal systems contain all the K-theorems, plus possibly more. What Theorem
6.1 gives us is a method for constructing a soundness proof for any normal
system. Since all the systems we have studied here (K, D, etc.) are normal, this
method is sufficiently general for us. Here’s how the method works for system
T. System T has the same rules of inference as K, and its axioms are all the
axioms of K, plus the instances of the T-schema. In the “K+Γ” notation, T = K
+ {2φ→φ : φ is an MPL wff}. To establish soundness for T, all we need to do
is show that every instance of the T-schema is valid in all reflexive models; for
we may then conclude by Theorem 6.1 that every theorem of T is valid in all
reflexive models. This method can be applied to each of our systems: for any
system, S, to establish S’s soundness it will suffice to show that the S’s “extra-K”
axioms are valid in all of the S-models.
Theorem 6.1 follows from two lemmas we will need to prove:
Lemma 6.2 All PL and K-axioms are valid in all MPL-models
Lemma 6.3 For every MPL-model, M , MP and Necessitation preserve validity
in M
Proof of Theorem 6.1 from the lemmas. Assume that every wff in Γ is valid in a
given MPL-model M , and consider any theorem φ of K+Γ. That theorem is
a last line in a proof in which each line is either an axiom K+Γ, or follows from
earlier lines in the proof by MP or NEC. But axioms of K+Γ are either PL
axioms, K axioms, or members of Γ. The first two classes of axioms are valid in
all MPL-models, by Lemma 6.2, and so are valid in M ; and the final class of
axioms are valid in M by hypothesis. Thus, all axioms in the proof are valid
in M . Moreover, by Lemma 6.3, the rules of inference in the proof preserve
validity in M . Therefore, by induction, every line in the proof is valid in M .
Hence the last line in the proof, φ, is valid in M .
We now need to prove the lemmas.
Proof of Lemma 6.2. From our proof of soundness for PL (section 2.6), we know
that the PL truth tables generate the value 1 for each PL axiom, no matter
what truth value its immediate constituents have. But here in MPL, the truth
values of conditionals and negations are determined at a given world by the
truth values at that world of its immediate constituents via the PL truth tables.
So any PL axiom must have truth value 1 at any world, regardless of what truth
values its immediate constituents have. PL-axioms, therefore, are true at every
world in every model, and so are valid in every model. We need now to show
that any K axiom—i.e., any formula of the form 2(φ→ψ)→ (2φ→2ψ)—is
valid in any model:
i) Suppose for reductio that V(2(φ→ψ)→(2φ→2ψ), w) = 0, for some

model 〈W , R, I 〉, whose valuation is V, and some w ∈ W
ii) So V(2(φ→ψ), w) = 1 and…
iii) …V((2φ→2ψ), w) = 0
iv) Given iii), V(2φ, w) = 1 and …

v) …V(2ψ, w) = 0
vi) Given v), for some v, R wv and V(ψ, v) = 0
vii) Given iv), since R wv, V(φ, v) = 1
viii) Given ii), since R wv, V(φ→ψ, v) = 1
ix) Lines vi), vii), and viii) contradict, given the truth condition for the →
Proof of Lemma 6.3. We must show that the rules MP and NEC preserve va-
lidity in any given model. That is, we must show that if the inputs to one of
these rules is valid in some model, then that rule’s output must also be valid in
that model.
First MP. Let φ and φ→ψ be valid in model 〈W , R, I 〉; we must show that
ψ is also valid in that model. That is, where V is this model’s valuation, and
w is any member of W , we must show that V(ψ, w) = 1. Since φ and φ→ψ
are valid in this model, V(φ→ψ, w) = 1, and V(φ, w) = 1; but by the truth
condition for →, V(ψ, w) must also be 1.
Next NEC. Suppose φ is valid in model M . We must show that 2φ is
valid in M , i.e., that 2φ is true at each world in M , i.e., that for each world,
w, φ is true at every world accessible from w. But since φ is valid in M , φ
is true in every world in M , and hence is true at every world accessible from
w.
6.5.1 Soundness of K
We can now construct soundness proofs for the individual systems. I’ll do this
for some of the systems, and leave the verification of soundness for the other
systems as exercises.
First K. In the “K+Γ” notation, K is just K+∅, and so it follows immediately
from Theorem 6.1 that every theorem of K is valid in every MPL-model. So
K is sound.
6.5.2 Soundness of T
T is K+Γ, where Γ is the set of all instances of the T-schema. So, given Theorem
6.1, to show that every theorem of T is valid in all T-models, it suffices to show
that all instances of the T-schema are valid in all T-models:
i) Assume for reductio that V(2φ→φ, w) = 0 for some world w in some

T-model (i.e., some model with a reflexive accessibility relation)
ii) So V(2φ, w) = 1 and…
iii) …V(φ, w) = 0
iv) R ww, by reflexivity. So, from ii), V(φ, w) = 1, contradicting iii)
6.5.3 Soundness of B
B is K+ Γ, where Γ is the set of all instances of the T- and B- schemas. Given
Theorem 6.1, it suffices to show that every instance of the B-schema and every
instance of the T-schema is valid in every B-model. So, choose an arbitrary
model with a reflexive and symmetric accessibility relation, whose valuation is
V, and let w be any world in that model. We must show that V counts each
instance of the T-schema and the B-schema as being true at w. The proof
of the previous section shows that the T-axioms are true at w. Now for the
B-axioms:
i) Assume for reductio that V(32φ→φ, w) = 1.
ii) So V(32φ, w) = 1 and…
iii) …V(φ, w) = 0
iv) By ii), V(2φ, v) = 1, for some v such that R wv.
v) By symmetry, R v w. So, given iv), V(φ, w) = 1, contradicting iii)
6.6 Completeness of MPL

Next, completeness: for each system, we’ll show that every valid formula is a
theorem. As with soundness, most of the work will go into developing some
general-purpose machinery. At the end we’ll use the machinery to construct
completeness proofs for each system.
We’ll be constructing a kind of completeness proof known as a “Henkin-
proof”, after Leon Henkin, who used similar methods to demonstrate com-
pleteness for (nonmodal) predicate logic.
6.6.1 Canonical models

For each of our systems, we’re going to show how to construct a certain special
model, the canonical model for that system. The canonical model for a system,
S, will be shown to have the following feature:
If a formula is valid in the canonical model for S, then it is a

theorem of S
This sufficient condition for theoremhood can then be used to give complete-
ness proofs, as the following example brings out. Suppose we can demonstrate
that the accessibility relation in the canonical model for T is reflexive. Then,
since T-valid formulas are by definition true in every world in every model
with a reflexive accessibility relation, we know that every T-valid formula is
valid in the canonical model for T. But then the italicized statement tells us
that every T-valid formula is a theorem of T. So we would have established
completeness for T.
The trick for constructing canonical models will be to let the worlds in these
models be sets of formulas (remember, worlds are allowed to be anything we
like). And we’re going to construct the interpretation function of the canonical
model in such a way that a formula will be true at a world iff the formula is a
member of the set that is the world. Working out this idea will occupy us for
awhile.
6.6.2 Maximal consistent sets of wffs

To carry out this idea of constructing worlds as sets of formulas that are true at
those worlds, we’ll need to put some constraints on the nature of these sets of
wffs. It’s part of the definition of a valuation function that for any wff φ and
any world w, either φ or ∼φ is true at w. That means that any set of wffs that
we’re going to call a world had better contain either φ or ∼φ. Moreover, we’d
better not let such a set contain both φ and ∼φ, since a formula can’t be both
true and false at a world. Other constraints must be introduced as well.
Where S is any of our axiomatic systems, let’s define the following notions:
Definition of consistency and maximality:
· A set of MPL-wffs, Γ, is S-inconsistent iff for some φ1 . . . φn ∈ Γ, `S

∼(φ1 ∧ · · · ∧φn ). Γ is S-consistent iff it is not S-inconsistent
· A set of MPL-wffs, Γ, is maximal iff for every MPL-wff φ, either φ or

∼φ is a member of Γ
· A set is maximal S-consistent iff it is both maximal and S-consistent
A set is S-inconsistent if it contains some wffs (finite in number) that are
provably (in S) contradictory; it is S-consistent if it contains no such wffs. This
notion of consistency is proof-theoretic: it has to do with what can be proved in
axiomatic systems, not with truth in models. Furthermore, S-consistency has to
do with provability in system S. It therefore requires more than the mere absence
of contradictions. Thus consider a set of wffs that contains no contradictions
(i.e., for no φ does the set contain both φ and ∼φ), but which contains both
2P and ∼P . This set would be T-inconsistent, since ∼(2P ∧∼P ) is a theorem
of T.
A maximal S-consistent set of wffs contains, for each formula, either that
formula or its negation; and it contains no finite list of formulas that are
collectively disprovable in S. Maximal consistent sets are fit sets to be worlds
in our canonical models.
6.6.3 Definition of canonical models

We’re now ready to define canonical models. It may not be fully clear at this
point why the definition is phrased as it is; you’ll need to take it on faith, for
the moment, that the definition will get us where we want to go.
Definition of canonical model: The canonical model for system S is the
MPL-model 〈W , R, I 〉 where:
· W is the set of all maximal S-consistent sets of wffs
· R ww 0 iff 2− (w) ⊆ w 0
· I (α, w) = 1 iff α ∈ w, for each sentence letter α and each w ∈ W
(where 2− (∆) is defined as the set of wffs φ such that 2φ is a member of ∆)
Let’s think for a bit about this definition. As promised, we have defined the
members of W to be maximal S-consistent sets of wffs. And note that all
maximal S-consistent sets of wffs are included in W .
Accessibility is defined using the “2− ” notation. Think of this operation
as “stripping off the boxes”: to arrive at 2− (∆) (“the box-strip of set ∆”),
begin with set ∆, discard any formula that doesn’t begin with a 2, line up
the remaining formulas, and then strip one 2 off of the front of each. The
definition of accessibility, therefore, says that R ww 0 iff for each wff 2φ that is
a member of w, the wff φ is a member of w 0 .
The definition of accessibility in the canonical model says nothing about
formal properties like transitivity, reflexivity, and so on. As a result, it is not
true by definition that the canonical model for S is an S-model. T-models,
for example, must have reflexive accessibility relations, whereas the definition
of the accessibility relation in the canonical model for T says nothing about
reflexivity. As we will eventually see, the canonical model for each system S
turns out to be an S-model, but this fact must be proven; it’s not built into the
definition of a canonical model.
An atomic wff (sentence letter) is defined to be true at a world iff it is a
member of that world. Thus, for atomic wffs, truth and membership coincide.
What we really need to know, however, is that truth and membership coincide
for all wffs, including complex wffs. Proving this turns out to be a big task,
which will occupy us for several sections. We’ll need first to assemble some
firepower: a number of preliminary lemmas and theorems which will eventually
be used to prove that membership and truth coincide for all wffs in canonical
models. We’ll then finally be able to give completeness proofs.
6.6.4 Features of maximal consistent sets

We’ll begin with some lemmas governing the behavior of maximal consistent
sets:
Lemma 6.4 Where Γ is any maximal S-consistent set of wffs:
6.4a for any wff φ, exactly one of φ, ∼φ is a member of Γ
6.4b φ→ψ ∈ Γ iff either φ ∈

/ Γ or ψ ∈ Γ
6.4c if φ and φ→ψ are both members of Γ then so is ψ
6.4d if `S φ then φ ∈ Γ
Proof of Lemma 6.4a. We know from the definition of maximality that at least
one of φ or ∼φ is in Γ. But it cannot be that both are in Γ, for then Γ would be
S-inconsistent (it would contain the finite subset {φ, ∼φ}; but since all modal
systems incorporate propositional logic, it is a theorem of S that ∼(φ∧∼φ).)
Proof of Lemma 6.4b. Suppose first that φ→ψ is in Γ, and suppose for reductio
that φ is in Γ but ψ is not. Then, since Γ is maximal, ∼ψ is in Γ; but now
Γ is S-inconsistent by containing the subset {φ, φ→ψ, ∼ψ}. Suppose for the
other direction that either φ is not in Γ or ψ is in Γ, and suppose for reductio
that φ→ψ isn’t in Γ. Since Γ is maximal, ∼(φ→ψ) ∈ Γ. Now, if φ ∈ / Γ then
∼φ ∈ Γ, but then Γ would contain the S-inconsistent subset {∼(φ→ψ), ∼φ}.
And if on the other hand, ψ ∈ Γ then Γ again contains an S-inconsistent subset:
{∼(φ→ψ), ψ}. Either possibility contradicts Γ’s S-consistency.
Exercise 6.10 Prove lemmas 6.4c and 6.4d.
6.6.5 Maximal consistent extensions

Next let’s show that if we begin with an S-consistent set ∆, we can “expand” it
into a maximal S-consistent set Γ. (We prove this because we’ll need to know
that there exist enough maximal consistent sets in a canonical model’s W in
order for the model to do its thing.)
Theorem 6.5 If ∆ is an S-consistent set of wffs, then there exists some maximal
S-consistent set of wffs, Γ, such that ∆ ⊆ Γ
Proof of Theorem 6.5. In outline, we’re going to build up Γ as follows. We’re

going to start by dumping all the formulas in ∆ into Γ. Then we will go through
all the wffs in the language of MPL, φ1 , φ2 ,…, one at a time. For each of these
wffs, we’re going to dump either it or its negation into Γ, depending on which
choice would be S-consistent. After we’re done, our set Γ will obviously be
maximal; it will obviously contain ∆ as a subset; and, we’ll show, it will also be
S-consistent.
So, let φ1 , φ2 ,… be a list—an infinite list, of course—of all the wffs of
MPL.10 Our strategy, recall, is to construct Γ by starting with ∆, and then
10
We need to be sure that there is some way of arranging all the wffs of MPL into such a
list. Here is one method. Consider the following list of the primitive expressions of MPL:
( ) ∼ → 2 P1 P2 ...
1 2 3 4 5 6 7 ...
Since we’ll need to refer to what position an expression has in this list, the positions of the
going through this list one-by-one, at each point adding either φi or ∼φi .
Here’s how we do this more carefully. Let’s begin by defining an infinite
sequence of sets, Γ0 , Γ1 , . . . :
· Γ0 is ∆
· Γn+1 is Γn ∪ {φn+1 } if that is S-consistent; otherwise Γn+1 is Γn ∪ {∼φn+1 }
Note the recursive nature of the definition: the next member of the sequence
of sets, Γn+1 is defined as a function of the previous member of the sequence,
Γn .
Next let’s prove that each member in this sequence—that is, each Γi —is an
S-consistent set. We do this inductively, by first showing that Γ0 is S-consistent,
and then showing that if Γn is S-consistent, then so will be Γn+1 .
Obviously, Γ0 is S-consistent, since ∆ was stipulated to be S-consistent.
Next, suppose that Γn is S-consistent; we must show that Γn+1 is S-consistent.
Look at the definition of Γn+1 . What Γn+1 gets defined as depends on whether
Γn ∪{φn+1 } is S-consistent. If Γn ∪{φn+1 } is S-consistent, then Γn+1 gets defined
as that very set Γn ∪ {φn+1 }, and so of course is S-consistent. So we’re ok in
that case.
The remaining possibility is that Γn ∪ {φn+1 } is S-inconsistent. In that case,
Γn+1 gets defined as Γn ∪{∼φn+1 }. So must show that in this case, Γn ∪{∼φn+1 }
is S-consistent. Suppose for reductio that it isn’t. The conjunction of some
finite subset of its members must therefore be provably false in S. Since Γn was
S-consistent, the finite subset must contain ∼φn+1 , and so there exist ψ1 . . . ψ m ∈
expressions are listed underneath those expressions. (E.g., the position of the 2 is 5.) Now,
where φ is any wff, call the rating of φ the sum of the positions of the occurrences of its
primitive expressions. (The rating for the wff (P1 →P1 ), for example, is 1 + 6 + 4 + 6 + 2 = 19.)
We can now construct the listing of all the wffs of MPL by an infinite series of stages: stage 1,
stage 2, etc. In stage n, we append to our growing list all the wffs of rating n, in alphabetical
order. The notion of alphabetical order here is the usual one, given the ordering of the primitive
expressions laid out above. (E.g., just as ‘and’ comes before ‘nad’ in alphabetical order, since
‘a’ precedes ‘n’ in the usual ordering of the English alphabet, ∼2P2 comes before 2∼P2 in
alphabetical order since ∼ comes before the 2 in the ordering of the alphabet of MPL. Note
that each of these wffs are inserted into the list in stage 15, since each has rating 15.) In stages
1-5 no wffs are added at all, since every wff must have at least one sentence letter and P1 is
the sentence letter with the smallest position. In stage 6 there is one wff: P1 . Thus, the first
member of our list of wffs is P1 . In stage 7 there is one wff: P2 , so P2 is the second member of
the list. In every subsequent stage there are only finitely many wffs; so each stage adds finitely
many wffs to the list; each wff gets added at some stage; so each wff eventually gets added after
some finite amount of time to this list.
Γn such that `S ∼(ψ1 ∧ · · · ∧ψ m ∧∼φn+1 ). Furthermore, since Γn ∪ {φn+1 } is S-

inconsistent, it too contains a finite subset that is provably false in S. Since Γn
is S-consistent, the finite subset must contain φn+1 , so there exist χ1 . . . χ p ∈ Γn
such that `S ∼(χ1 ∧ · · · ∧χ p ∧φn+1 ). But notice that ∼(ψ1 ∧ · · · ∧ψ m ∧χ1 ∧ · · · ∧χ p )
is a PL-semantic-consequence of the formulas ∼(ψ1 ∧ · · · ∧ψ m ∧∼φn+1 ) and
∼(χ1 ∧ · · · ∧χ p ∧φn+1 ). It follows that `S ∼(ψ1 ∧ · · · ∧ψ m ∧χ1 ∧ · · · ∧χ p ) (each of
our modal systems “contains PL”: given the completeness of the PL axioms,
one can always move within an MPL axiomatic proof from some formulas to a
PL-semantic-consequence of those formulas.) Since ψ1 . . . ψ m and χ1 . . . χ p are
all members of Γn , this contradicts the fact that Γn is S-consistent.
We have shown that all the sets in our sequence Γi are S-consistent. Let
us now define Γ to be the union of all the sets in the infinite sequence—i.e.,
{φ : φ ∈ Γi for some i }. We must now show that Γ is the set we’re after: that i)
∆ ⊆ Γ, ii) Γ is maximal, and iii) Γ is S-consistent.
Any member of ∆ is a member of Γ0 (since Γ0 was defined as ∆), hence is a
member of one of the Γi s, and hence is a member of Γ. So ∆ ⊆ Γ.
Any wff of MPL is in the list somewhere—i.e., it is φi for some i. But by
definition of Γi , either φi or ∼φi is a member of Γi ; and so one of these is a
member of Γ. Γ is therefore maximal.
Suppose for reductio that Γ is S-inconsistent; there must then exist ψ1 . . . ψ m
∈ Γ such that `S ∼(ψ1 ∧ · · · ∧ψ m ). By definition of Γ, each of these ψi ’s are
members of Γ j , for some j . Let k be the largest such j . Note next that, given
the way the Γi ’s are constructed, each Γi is a subset of all subsequent ones.
Thus, all of the ψi ’s are members of Γk , and thus Γk is S-inconsistent. But that
can’t be—we showed that all the Γi ’s are S-consistent.
6.6.6 “Mesh”
Our ultimate goal is to show that in canonical models, a wff is true at a world
iff it is a member of that world. If we’re going to be able to show this, we’d
better be able to show things like this:
(2) If 2φ is a member of world w, then φ is a member of every world

accessible from w
(3) If 3φ is a member of world w, then φ is a member of some world

accessible from w
We’ll need to be able to show (2) and (3) because it’s part of the definition of
truth in any MPL-model (whether canonical or not) that 2φ is true at w iff φ
is true at each world accessible from w, and that 3φ is true at w iff φ is true at
some world accessible from w. Think of it this way: (2) and (3) say that the
modal statements that are members of a world w in a canonical model “mesh”
with the members of the other worlds in that canonical model. This sort of
mesh had better hold if truth and membership are going to coincide.
(2) we know to be true straightaway, since it follows from the definition of
the accessibility relation in canonical models. The definition of the canonical
model for S, recall, stipulated that w 0 is accessible from w iff for each wff 2φ
in w, the wff φ is a member of w 0 . (3), on the other hand, doesn’t follow
immediately from our definitions; we’ll need to prove it. Actually, it will be
convenient to prove something slightly different which involves only the 2:
Lemma 6.6 If ∆ is a maximal S-consistent set of wffs containing ∼2φ, then

there exists a maximal S-consistent set of wffs Γ such that 2− (∆) ⊆ Γ and
∼φ ∈ Γ
(Given the definition of accessibility in the canonical model and the definition
of the 3 in terms of the 2, Lemma 6.6 basically amounts to (3).)
Proof of Lemma 6.6. Let ∆ be as described. The first (and biggest) step is to
establish:
(*) 2− (∆) ∪ {∼φ} is S-consistent.
Suppose for reductio that (*) is false. By the definition of S-inconsistency, for
some χ1 . . . χ m ∈ 2− (∆) ∪ {∼φ} the following is a theorem of S:
∼(χ1 ∧ · · · ∧χ m )
The following is a PL-semantic-consequence of this formula, and hence, since

S includes PL, is also a theorem of S:
∼(χ1 ∧ · · · ∧χ m ∧∼φ)
Now go through the list χ1 . . . χ m , and if it contains any wffs that are not
members of 2− (∆), drop them from the list. Call the resulting list ψ1 . . . ψn .
Each of the ψi s, note, is a member of 2− (∆).11 The only wff that could have
been dropped, in moving from the χi s to the ψi s, is ∼φ (since each χi was
a member of 2− (∆) ∪ {∼φ}); the following wff is therefore a PL-semantic-
consequence of the previous wff, and so is itself a theorem of S:
∼(ψ1 ∧ · · · ∧ψn ∧∼φ)
Next, begin a proof in S with a proof of ∼(ψ1 ∧ · · · ∧ψn ∧∼φ), and then continue
as follows:
.
.
i. ∼(ψ1 ∧ · · · ∧ψn ∧∼φ)
i + 1. ψ1 →(ψ2 → · · · (ψn →φ)) . . . ) i, PL
i + 2. 2(ψ1 →(ψ2 → · · · (ψn →φ)) . . . ) i + 1, NEC
.
.
j. 2ψ1 →(2ψ2 → · · · (2ψn →2φ)) . . . ) i + 2…, K, PL (×n)
j + 1. ∼(2ψ1 ∧ · · · ∧2ψn ∧∼2φ) j , PL
This proof establishes that `S ∼(2ψ1 ∧ · · · ∧2ψn ∧∼2φ). But since 2ψ1 …2ψn ,
and ∼2φ are all in ∆, this contradicts ∆’s S-consistency (2ψ1 …2ψn are mem-
bers of ∆ because ψ1 …ψn are members of 2− (∆).)
We’ve established (*): 2− (∆) ∪ {∼φ} is S-consistent. It therefore has a
maximal S-consistent extension, Γ, by Theorem 6.5. Since 2− (∆) ∪ {∼φ} ⊆ Γ,
we know that 2− (∆) ⊆ Γ and that ∼φ ∈ Γ. Γ is therefore our desired set.
Exercise 6.11 Where S is any normal modal system, show that if

∆ is an S-consistent set of wffs containing the formula 3φ, then
2− (∆) ∪ φ is also S-consistent. You may appeal to lemmas and
theorems proved in this chapter so far.
11
If 2− (∆) is empty then there will be no ψi s. In that case, let’s regard “ψ1 ∧ · · · ∧ψn ∧∼φ”
as standing for ∼φ.
6.6.7 Truth and membership in canonical models

We’re now in a position to put all of our lemmas to work, and prove that
canonical models have the desired property that the wffs true at a world are
exactly the members of that world:
Theorem 6.7 Where M (= 〈W , R, I 〉) is the canonical model for any normal

modal system, S, for any wff φ and any w ∈ W , VM (φ, w) = 1 iff φ ∈ w
Proof of Theorem 6.7. We’ll use induction. The base case is when φ has zero
connectives—i.e., φ is a sentence letter. In that case, the result is immediate:
by the definition of the canonical model, I (φ, w) = 1 iff φ ∈ w; but by the
definition of the valuation function, VM (φ, w) = 1 iff I (φ, w) = 1.
Now the inductive step. We suppose (ih) that the result holds for φ, ψ, and
show that it holds for ∼φ, φ→ψ, and 2φ as well. First, ∼: we must show that
∼φ is true at w iff ∼φ ∈ w:
i) ∼φ ∈ w iff φ ∈
/ w (6.4a)
ii) φ ∈
/ w iff φ is not true at w (ih)
iii) φ is not true at w iff ∼φ is true at w (truth cond. for ∼)
iv) so, ∼φ is true at w iff ∼φ ∈ w (lines i), ii), iii))
Next, →: we must show that φ→ψ is true at w iff φ→ψ ∈ w:
i) φ→ψ is true at w iff either φ is not true at w or ψ is true at w (truth

cond for →)
ii) So, φ→ψ is true at w iff either φ ∈

/ w or ψ ∈ w (ih)
iii) So, φ→ψ is true at w iff φ→ψ ∈ w (6.4b)
Finally, 2: we must show that 2φ is true at w iff 2φ ∈ w. First the forwards

direction. Assume 2φ is true at w; then φ is true at every world w 0 such that
R ww 0 . By the ih, we have (+) φ is a member of every such w 0 . Now suppose
for reductio that 2φ ∈/ w; by 6.4a, ∼2φ ∈ w. Since w is maximal S-consistent,
by Lemma 6.6, we know that there exists some maximal S-consistent set Γ such
that 2− (w) ⊆ Γ and ∼φ ∈ Γ. By definition of W , Γ is a world; by definition of
R, R wΓ; and so by (+) Γ contains φ. But Γ also contains ∼φ, which contradicts
its S-consistency.
Now the backwards direction. Assume 2φ ∈ w. Then by definition of R,

for every w 0 such that R ww 0 , φ ∈ w 0 . By the ih, φ is true at every such world;
hence by the truth condition for 2, 2φ is true at w.
What was the point of proving theorem 6.7? The whole idea of a canonical
model was to be that a formula is valid in the canonical model for S iff it is a
theorem of S. This fact follows fairly immediately from Theorem 6.7:
Corollary 6.8 φ is valid in the canonical model for S iff `S φ
Proof of Corollary 6.8. Let 〈W , R, I 〉 be the canonical model for S. Suppose

`S φ. Then, by lemma 6.4d, φ is a member of every maximal S-consistent set,
and hence φ ∈ w, for every w ∈ W . By theorem 6.7, φ is true in every w ∈ W ,
and so is valid in this model. Now for the other direction: suppose 0S φ. Then
{∼φ} is S-consistent, and so by theorem 6.5, has a maximal consistent extension;
thus, ∼φ ∈ w for some w ∈ W ; by theorem 6.7, ∼φ is therefore true at w, and
so φ is not true at w, and hence φ is not valid in this model.
So, we’ve gotten where we wanted to go: we’ve shown that every system
has a canonical model, and that a wff is valid in the canonical model iff it is a
theorem of the system. We now use this fact to prove completeness for our
various systems:
6.6.8 Completeness of systems of MPL

Completeness of K
K’s completeness follows immediately. Any K-valid wff is valid in all MPL-
models, and so is valid in the canonical model for K, and so, by corollary 6.8, is
a theorem of K.
For any other system, S, all we need to do to prove S-completeness is to
show that the canonical model for S is an S-model. That is, we must show
that the accessibility relation in the canonical model for S satisfies the formal
constraint for system S (seriality for D, reflexivity for T and so on). This will
be made clear in the proof of completeness for D:
Completeness of D
Let us show that in the canonical model for D, the accessibility relation, R, is
serial. Let w be any world in that model. We showed above that 3(P →P ) is a
theorem of D, and so is a member of w by lemma 6.4d, and so is true at w by

theorem 6.7. Thus, by the truth condition for 3, there must be some world
accessible to w in which P →P is true; and hence there must be some world
accessible to w.
Now for D’s completeness. Let φ be D-valid. It is then valid in all D-
models, i.e., all models with a serial accessibility relation. But we just showed
that the canonical model for D has a serial accessibility relation. φ is therefore
valid in that model, and hence by corollary 6.8, `D φ.
Completeness of T
All we need to do is to prove that the accessibility relation in the canonical model
for T is reflexive; given that, every T-valid formula is valid in the canonical
model for T, and hence by corollary 6.8, every T-valid formula is a T-theorem.
Let φ be any wff. `T 2φ→φ, so, where w is any world in the canonical
model for T, by lemma 6.4d, 2φ→φ ∈ w. By lemma 6.4c, if 2φ ∈ w, then so
is φ. Formula φ was arbitrarily chosen, so we have: for any φ, if 2φ ∈ w then
φ ∈ w. But this is the definition of R ww. World w was arbitrarily chosen, so
R is reflexive.
Completeness of B
We must show that the accessibility relation in the canonical model for B is
reflexive and symmetric. Reflexivity can be demonstrated in the same way as it
was for T, since every T-theorem is a B-theorem.
Now for symmetry: in the canonical model for B, suppose that R wv. We
must show that R v w—that is, that for any 2ψ in v, ψ ∈ w. So, suppose that
2ψ ∈ v. By theorem 6.7, 2ψ is true at v; since R wv, by the definition of
3 it follows that 32ψ is true at w, and hence is a member of w by theorem
6.7. Since `B 32ψ→ψ, by lemma 6.4d, 32ψ→ψ ∈ w, and so, by lemma 6.4c,
ψ ∈ w.
Completeness of S4
We must show that the accessibility relation in the canonical model for S4 is
reflexive and transitive. Again, reflexivity can be demonstrated as it was for
T; transitivity remains. Suppose R wv and R v u. We must show R w u—that
is, for any 2ψ ∈ w, ψ ∈ u. If 2ψ ∈ w, since `S4 2ψ→22ψ, by lemma 6.4d,
2ψ→22ψ ∈ w, and so by lemma 6.4c, 22ψ ∈ w. By theorem 6.7, 22ψ is

true at w; hence by the truth condition for 2, 2ψ is true at v; again by the
truth condition for 2, ψ is true at u; by theorem 6.7, ψ ∈ u.
Completeness of S5
We must show that the accessibility relation in the canonical model for S5 is
reflexive, symmetric, and transitive. But since each T, B, and S4 theorem is
an S5 theorem, the proofs of reflexivity, symmetry, and transitivity from the
previous three sections apply here.
Exercise 6.12 Consider the system that results from adding to K

every axiom of the form 3φ→2φ. Let the frames for this system
be defined as those whose accessibility relation meets the following
condition: every world can see at most one world. Prove completeness
for this (strange) system.
Chapter 7
Variations on Propositional Modal

Logic
s we have seen, possible worlds are useful for giving a semantics for propo-
A sitional modal logic. Possible worlds are useful in other areas of logic as
well. In this chapter we will briefly examine two other uses for possible worlds:
semantics for tense logic, and semantics for intuitionist propositional logic.
7.1 Propositional tense logic1

7.1.1 The metaphysics of time
Propositional modal logic concerned the logic of the non-truth-functional
sentential operators “it is necessary that” and “it is possible that”. Another
set of sentential operators that can be similarly treated are propositional tense
operators, such as “it will be the case that”, “it has always been the case that”,
etc.
A full logical treatment of natural language obviously requires that we
pay attention to temporal notions. Some philosophers, however, think that it
requires nothing beyond standard predicate logic. This was the view of many
early logicians, most notably Quine.2 Here are some examples of how Quine
would regiment temporal sentences in predicate logic:
1
See Gamut (1991b, section 2.4); Cresswell and Hughes (1996, pp. 127-134).
2
See, for example, Quine (1953b).
191
CHAPTER 7. VARIATIONS ON MPL 192
Everyone who is now an adult was once a child

∀x(Axn → ∃t [E t n ∧ C x t ])
A dinosaur once trampled a mammal
∃x∃y∃t (E t n ∧ D x ∧ M y ∧ T xy t )
Comments:
· ‘n’ is to be a name for the present time
· The predicate “E” is to be a predicate for the earlier-than relation over

moments of time. Thus, “E t n” means that t is a time that is before the
present moment; and so, “∃t (E t n∧ . . . ” means that there exists some
time, t , before the present moment, such that …”.
· we add in a new place for all predicates, for the time at which the object
satisfies the predicate. Thus, instead of saying “C x”—“x is a child”—we
say “C x t ”: “x is a child at t ”
· the quantifier ∃x is atemporal, ranging over all objects at all times. That’s
how we can say that there is a thing, x, that is a dinosaur, and which, at
some previous time, trampled a mammal.
So: we can use Quine’s strategy to represent temporal notions using standard
predicate logic. But some philosophers reject the conception of time that is
presupposed by Quine’s strategy. First, Quine presupposes that the past, present,
and future are equally real. After all, his symbolization of “A dinosaur once
trampled a mammal” says that there is such a thing as a dinosaur. Quine’s view is
that time is “space-like”. Other times are as real as the present, just temporally
distant, just as other places are equally real but spatially distant. Second, Quine
presupposes a distinctive metaphysics of change. Quine accounts for change
by adding argument places to temporary predicates like ‘is a child’ and ‘is an
adult’. For him, the statement ‘Ted is an adult’ is incomplete in something
like the way ‘Philadelphia is north of’ is complete: its predicate has an unfilled
argument place. When all of a sentence’s argument places are filled, it can no
longer change its truth value; as a result (according to some), Quine’s approach
leaves no room for genuine change.
Arthur Prior (1967; 1968) and others reject Quine’s picture of time. Ac-
cording to Prior, rather than reducing notions of past, present, and future to
notions about what is true at times, we must instead include certain special
temporal expressions—sentential tense operators—in our most basic languages,
and develop an account of their logic. Thus he initiated the study of tense logic.
One of Prior’s tense operators was P, symbolizing “it was the case that”.
Grammatically, P behaves like the ∼ and the 2: it attaches to a complete
sentence and forms another complete sentence. Thus, if R symbolizes “it is
raining”, then P R symbolizes “it was raining”. If a sentence letter occurs by
itself, outside of the scope of all temporal operators, then for Prior it is to
be read as present-tensed. Thus, it was appropriate to let R symbolize “It is
raining”—i.e., it is now raining.
Suppose we symbolize “there exists a dinosaur” as ∃xD x. Prior would then
symbolize “There once existed a dinosaur” as:
P ∃xD x
And according to Prior, P ∃xD x is not to be analyzed as saying that there exist
dinosaurs located in the past. For him, there is no further analysis of P ∃xD x.
Prior’s attitude toward P is like everyone else’s attitude toward the ∼: no one
thinks that ∼∃xU x, “there are no unicorns”, is to be analyzed as saying that
there exist unreal unicorns. Further, Prior can represent the fact that I am
now, but have not always been, an adult, without adding argument places for
times to predicates. Symbolizing ‘is an adult’ with ‘A’, and ‘Ted’ with ‘t ’, Prior
would write: At ∧ P∼At (“Ted is an adult, but it was the case that: Ted isn’t an
adult”). For Prior, the sentence At (“Ted is an adult”) is a complete statement,
but nevertheless can alter its truth value.
7.1.2 Tense operators

One can study various tense operators. Here is one group:
Gφ: “it is, and is always going to be the case that φ”
Hφ: “it is, and always has been the case that φ”
Fφ: “it either is, or will at some point in the future be the case that, φ”
Pφ: “it either is, or was at some point in the past the case that φ”
Notice how these tense operators come in interdefinable pairs (related to

each other as the 2 and the 3):
Gφ iff ∼F∼φ
Hφ iff ∼P∼φ
Thus one could start with just two of them, G and H say, and define the others.
And one could use these two to define further tense operators, for example A
and S, for “always” and “sometimes”
Aφ iff Hφ∧Gφ
Sφ iff ∼H∼φ∨∼G∼φ (i.e., iff Pφ∨Fφ)
There are further tense operators that are not definable in terms of those
we’ve been considering so far. There are, for example, metrical tense operators,
which concern what happened or will happen at specific temporal distances in
the past or future:
P x φ: “it was the case x minutes ago that φ”

F x φ: “it will be the case in x minutes that φ”
We will not consider metrical tense operators further.

The (nonmetrical) tense operators, as interpreted above, “include the
present moment”. For example, if Gφ is now true, then φ must now be true.
One could specify an alternate interpretation on which they do not include the
present moment:
Gφ: “it is always going to be the case that φ”

Hφ: “it always has been the case that φ”
Fφ: “it will at some point in the future be the case that φ”
Pφ: “it was at some point in the past the case that φ”
Whether we take the tense operators as including the present moment will
affect what kind of logic we develop for them. For example, the “2-like”
operators G and H will obey the T-principle (Gφ and Hφ will imply φ) if they
are interpreted as including the present moment, but not otherwise.
7.1.3 Syntax of tense logic

In this chapter we will study only propositional tense logic. Its syntax is straight-
forward: each tense operator has the grammar of the ∼ and the 2. For example,
if we take the tense operators G and H as basic, then we could begin with the
definition of a wff from propositional logic (section 2.1) and add the following
clause:
· If φ is a wff then so are Gφ and Hφ
7.1.4 Possible worlds semantics for tense logic

Let’s turn now to semantics. The most natural semantics for tense logic is
a possible worlds-style semantics, in which we think of the members of W
as times rather than possible worlds, we think of the accessibility relation as
the temporal ordering relation, and we think of the interpretation function as
assigning truth values to sentence letters at times.
(A Priorean faces hard philosophical questions about the use of such a
semantics, since according to him, the semantics doesn’t accurately model the
metaphysics of time. The questions are like those questions that confront
someone who uses possible worlds semantics for modal logic, but doesn’t think
that possible worlds are part of the metaphysics of modality.)
This change in how we think about possible-worlds models doesn’t require
any change to the definition of a model from section 6.3. A model, M , is still
an ordered triple, whose first member is a nonempty set, whose second member
is a binary relation over that set, and whose third member is a function that
assigns to each sentence letter a truth value relative to each member of the
first member of M . To signify the change in how we’re thinking about these
models, however, let’s change our notation. Let’s call a model’s first member T ,
rather than W , and let’s use variables like t , t 0 , etc., for its members. And since
we’re thinking of a model’s second member as a relation of temporal ordering—
the at-least-as-early-as relation over times—let’s rename it too: “≤”. (If we were
interpreting the tense operators as not including the present moment, then
we would think of the temporal ordering relation as the strictly-earlier-than
relation, and would write it “<”.) Thus, instead of writing “R ww 0 ”, we write:
t ≤ t 0.
We’ll need to update the definition of the valuation function. The clauses
for the propositional connectives remain the same; what we need to add is
clauses for the tense operators. Let’s take just G and H as primitive; here are
the clauses:
VM (Gφ, t ) = 1 iff for every t 0 such that t ≤ t 0 , VM (φ, t 0 ) = 1
VM (Hφ, t ) = 1 iff for every t 0 such that t 0 ≤ t , VM (φ, t 0 ) = 1
If we define F and P as ∼G∼ and ∼H∼, respectively, then we get the following
derived clauses:
VM (Fφ, t ) = 1 iff for some t 0 such that t ≤ t 0 , VM (φ, t 0 ) = 1
VM (Pφ, t ) = 1 iff for some t 0 such that t 0 ≤ t , VM (φ, t 0 ) = 1
Call an MPL-model, thought of in this way, a “PTL-model” (for “Priorean
Tense Logic”). And say that a wff is PTL-valid iff it is true in every time in every
PTL-model. Given our discussion of system K from chapter 6, we already
know a lot about PTL-validity. The truth condition for the G is the same as
the truth condition for the 2 in MPL. Thus, for each K-valid formula φ of
MPL, there is a PTL-valid formula of tense logic: simply replace each 2 in
φ with G. Replacing 2s with Gs in the K-valid formula 2(P ∧Q)→2P results
in the PTL-valid formula G(P ∧Q)→GP , for example. Similarly, the result of
replacing 2s with Hs in a K-valid formula also results in a PTL-valid formula.
But there are further cases of PTL-validity that depend on the interaction
between different tense operators, and hence have no direct analog in MPL.
For example, we can demonstrate that PTL φ→GPφ:
i) Suppose for reductio that in some PTL-model M (= 〈T , ≤, I 〉) and
some t ∈ T , VM (φ→GPφ, t ) = 0. (I henceforth drop the subscript M .)
ii) So V(φ, t ) = 1 and …
iii) …V(GPφ, t ) = 0.
iv) Given iii), by the truth condition for G: for some t 0 ∈ T , t ≤ t 0 and
V(Pφ, t ) = 0
v) Given iv), by the (derived) truth condition for P: for every t 00 ∈ T , if
t 00 ≤ t 0 then V(φ, t 00 ) = 0
vi) letting t 00 in v) be t , given that t ≤ t 0 (from iv)), we have: V(φ, t ) = 0,
contradicting ii).
Similarly, one can show that PTL φ→HFφ.
7.1.5 Formal constraints on ≤

PTL-validity is not a good model for logical truth in tense logic. We have
so far placed no constraints on the formal properties of the relation ≤ in a
PTL-model. That means that there are PTL models in which the ≤ looks
nothing like a temporal ordering. We don’t normally think that time could
consist of a number of wholly temporally disconnected points, for example,
or of many points each of which is at-least-as-early-as all of the rest, and so
on, but there are PTL-models answering to these strange descriptions. As we
have defined them, PTL-valid formulas must be true at every world in every
PTL-model, even these strange models. This means that many tense-logical
statements that ought, intuitively, to count as logical truths, are in fact not
PTL-valid.
The formula GP →GGP is an example. It is PTL-invalid, for consider a
model with three times, t1 , t2 , and t3 , where t1 ≤ t2 , t2 ≤ t3 , and t1 6≤ t3 , and in
which P is true at t1 and t2 , but not at t3 :
t1 t2 t3
( (
• • •
P P ∼P
In this model, GP →GGP is false at time t1 . But GP →GGP is, intuitively, a
logical truth. If it is and will always be raining, then surely it must also be
true that: it is and always will be the case that: it is and always will be raining.
The problem, of course, is that the ≤ relation in the model we considered is
intransitive, whereas, one normally assumes, the at-least-as-early-as relation
must be transitive.
So: a more interesting notion of validity for PTL formulas results from
considering only PTL-models with transitive ≤ relations. Doing this validates
every instance of the “S4” schemas:
Gφ→GGφ
Hφ→HHφ
There are other interesting constraints on ≤ that one might impose. One might
impose reflexivity, for example. This is natural to impose if we are construing
the tense operators as including the present moment; not otherwise. Imposing
reflexivity validates the “T-schemas” Gφ→φ and Hφ→φ.
One might also impose “connectivity” of some sort.
Definition of kinds of connectivity: Let R be any binary relation over A.
· R is strongly connected in A iff for every u, v ∈ A, either Ruv or Rv u

· R is weakly connected iff for every u, v, v 0 , IF: either Ruv and uv 0 , or
Rv u and Rv 0 u, THEN: either Rv v 0 or Rv 0 v
So, we might require that the ≤ relation be strongly connected (in T ), or,
alternatively, merely weakly connected. This would be to disallow “incom-
parable” pairs of times—pairs of times neither of which bears the ≤ relation
to the other. The stronger requirement disallows all incomparable pairs; the
weaker requirement merely disallows incomparable pairs when each member
of the pair is after or before some one time. Thus, the weaker requirement
disallows “branches” in the temporal order but allows distinct timelines wholly
disconnected from one another, whereas the stronger requirement insures that
all times are part of a single non-branching structure. Each sort validates every
instance of the following schemas:
G(Gφ→ψ) ∨ G(Gψ→φ)
H(Hφ→ψ) ∨ H(Hψ→φ)
Here’s a sketch of a validity proof for the first schema:
i) Assume for reductio that V(G(Gφ→ψ)∨G(Gψ→φ), t ) = 0 in some PTL-

model whose accessibility relation is weakly connected.
ii) So V(G(Gφ→ψ), t ) = 0 and V(G(Gψ→φ), t ) = 0
iii) so there exist times, t 0 and t 00 , such that t ≤ t 0 and t ≤ t 00 , and V(Gφ→ψ, t 0 ) =
0 and V(Gψ→φ, t 00 ) = 0
iv) thus, Gφ is true at t 0 , ψ is false at t 0 , Gψ is true at t 00 , and φ is false at t 00
v) but by weak connectivity, t 0 ≤ t 00 or t 00 ≤ t 0 . Either way iv) leads to a

contradiction.
There are other constraints one might impose, for example anti-symmetry
(no distinct times bear ≤ to each other), density (between any two times there is
another time), or eternality (there exists neither a first nor a last time). In some
cases, imposing a constraint validates an interesting schema being validated.
Further, some constraints are more philosophically controversial than others.
Notice that one should not impose symmetry on ≤. Obviously if one time
is at least as early as another, then the second time needn’t be at least as early
as the first. Moreover, imposing symmetry would validate the “B” schemas
FGφ→φ and PHφ→φ; but these clearly ought not to be validated. Take the
first, for example: it doesn’t follow from it will be the case that it is always going to
be the case that I’m dead that I’m (now) dead.
So far we have been interpreting the tense operators as including the present
moment. That led us to call the temporal ordering relation in our models “≤”,
and require that it be reflexive. What if we instead interpreted the tense
operators as not including the present moment? We would then call the
temporal ordering relation “<”, and think of it as the earlier-than relation; and
we would no longer require that it be reflexive. Indeed, it would be natural to
require that it be irreflexive: that it never be the case that t < t .
We have considered only the semantic approach to tense logic. What of a
proof-theoretic approach? Given the similarity between tense logic and modal
logic, it should be no surprise that axiom systems similar to those of section
6.4 can be developed for tense logic. Moreover, the techniques developed in
sections 6.5-6.6 can be used to give soundness and completeness proofs for
tense-logical axiom systems, relative to the possible-worlds semantics that we
have developed in this section.
7.2 Intuitionist propositional logic

7.2.1 Kripke semantics for intuitionist propositional logic3
As we saw in section 3.4, intuitionists think of meaning in proof-theoretic terms,
rather than truth-theoretic terms, and as a result reject classical propositional
logic in favor of intuitionist propositional logic. We have already developed a
proof-theory for intuitionist logic: we began with the original sequent calculus
and then dropped double-negation elimination while adding ex falso. But we
still need a semantics. What should such a semantics look like?
In this book we have been thinking of logical truth, on the semantic
conception—i.e., validity—as “truth no matter what”. It is natural for in-
tuitionists to think, rather, in terms of “provability no matter what”. We will
lay out a semantics for intuitionist propositional logic—due to Saul Kripke—
that is based on this idea. The semantics will be like that of possible-worlds
3
See (Priest, 2001, chapter 6)
semantics for propositional modal logic, and so it will include valuation func-
tions that assign the values 1 and 0 to formulas relative to the members of a
set W . But the idea is to now think of the members of W as stages in the
construction of proofs, rather than as possible worlds, and to think of 1 and 0 as
“proof statuses”, rather than truth values. That is, we are to think of V(φ, w) = 1
as meaning that formula φ has been proved at stage w.
Let us treat the ∧ and the ∨ as primitive connectives. Here is Kripke’s
semantics for intuitionist propositional logic. (To emphasize the different way
we are regarding the “worlds”, we rename W “S ”, for stages in the construction
of proofs, and we will use the variables s, s 0 , etc., for its members.)
Definition of I-model: An I-model is a triple 〈S , R, I 〉, such that:
· S is a non-empty set (“proof stages”)

· R is a binary relation over S (“accessibility”) that is reflexive, transitive,
and obeys the heredity condition: for any sentence letter α, if I (α, s) = 1 and
R s s 0 then I (α, s 0 ) = 1
· I is a function from sentence letters and stages to truth values (“inter-
pretation function”).
Definition of valuation: Where M (= 〈S , R, I 〉) is any I-model, the I-

valuation for M , IVM , is defined as the two-place function that assigns either
0 or 1 to each wff relative to each member of S , subject to the following
constraints, where α is any sentence letter, φ and ψ are any wffs, and s is any
member of S :
VM (α, s) = I (α, s)
VM (φ∧ψ, s) = 1 iff VM (φ, s) = 1 and VM (ψ, s) = 1
VM (φ∨ψ, s) = 1 iff VM (φ, s) = 1 or VM (ψ, s ) = 1
VM (∼φ, s) = 1 iff for every s 0 such that R s s 0 , VM (φ, s 0 ) = 0
VM (φ→ψ, s) = 1 iff for every s 0 such that R s s 0 , either VM (φ, s 0 ) = 0
or VM (ψ, s 0 ) = 1
Note that the truth conditions for the → and the ∼ at stage s no longer
depend exclusively on what s is like; they are sensitive to what happens at
stages accessible from s. Unlike the ∧ and the ∨, → and ∼ are not “truth
functional” (relative to a stage); they behave like modal operators.
Let us think intuitively about these models. We are to think of each member
of S as a stage in the construction of mathematical proofs. At any stage, one
has come up with proofs of some things but not others. When V assigns 1 to a
formula at a stage, that means intuitively that as of that state of information,
the formula has been proven. The assignment of 0 means that the formula has
not been proven thus far (though it might nevertheless in the future.)
The holding of the accessibility relation R represents which future stages
are possible, given one’s current stage. If s 0 is accessible from s , that means that
s 0 contains all the proofs in s, plus perhaps more. Given this understanding,
reflexivity and transitivity are obviously correct to impose, as is the heredity
condition, since (on the somewhat idealized conception of proof we are oper-
ating with) one does not lose proved information when constructing further
proofs. But the accessibility relation will not in general be symmetric: for
sometimes one will come across a new proof that one did not formerly have.
Let’s also think through why the truth conditions for →, ∧, ∨ and ∼ are
intuitively correct. Intuitionists, recall, associate with each propositional con-
nective, a conception of what proofs of formulas built using that connective
must be like:
· a proof of ∼φ is a proof that φ leads to a contradiction
· a proof of φ∧ψ is a proof of φ and a proof of ψ
· a proof of φ∨ψ is a proof of φ or a proof of ψ
· a proof of φ→ψ is a construction that can be used to turn any proof of φ

into a proof of ψ
This is what inspires the definition of a valuation function. As of a time, one

has proved φ∧ψ iff one has proved both φ and ψ then. As of a time, one has
proved φ∨ψ iff one has proved one of the disjuncts. As for ∼, a proof of ∼φ,
according to an intuitionist, is a proof that φ leads to a contradiction. But i) if
one has proved that φleads to a contradiction, then in no future stage could
one prove φ (at least if one’s methods of proof are consistent); and ii) if one
has not proved that φ leads to a contradiction, this leaves open the possibility
of a future stage at which one proves φ. Thus the valuation condition for ∼ is
justified.4 As for →: if one has a method of converting proofs of φ into proofs
4
I’m fudging here a bit. Are the stages idealized so that one has already proven everyone
of ψ, then there could never be a possible future in which one has a proof of
φ but not one of ψ. Conversely, if one lacks such a method, then it should be
possible one day to have a proof of φ without being able to convert it into a
proof of ψ, and thus without then having a proof of ψ.
We can now define intuitionist validity and semantic consequence in the
obvious way:
Definitions of validity and semantic consequence:
· φ is I-valid (I φ) iff VM (φ, s) = 1 for each stage s in each intuitionist

model M
· φ is an I-semantic-consequence of Γ (Γ I φ) iff for every intuitionist
model M and every stage s in M , if VM (γ , s ) = 1 for each γ ∈ Γ, then
VM (φ, s) = 1
Exercise 7.1 Show that φ I ψ iff I φ→ψ.
Exercise 7.2 Show that intuitionist consequence implies classical

consequence. That is, show that if Γ I φ then Γ PL φ.
7.2.2 Examples
Given the semantics just introduced, it’s straightforward to demonstrate facts
about validity and semantic consequence.
Example 7.1: Show that Q I P →Q. (I’ll omit the qualifier “I” from now
on.) Take any model and any stage s ; assume that V(Q, s) = 1 and V(P →Q, s ) =
0. Thus, for some s 0 , R s s 0 and V(P, s 0 ) = 1 and V(Q, s 0 ) = 0. But this violates
heredity.
one can in principle prove? Clearly not, for then any formula assigned 1 at any accessible
stage should already be assigned 1 at that stage. But if stages are not idealized in this way,
then why suppose that the assignment of 0 at a stage to ∼φ (failure to prove that φ leads to a
contradiction) insures that there is some future stage at which φ is proved? A similar worry
confronts the valuation condition for →.
Example 7.2: Show that P →Q ∼Q→∼P (contraposition). Suppose

V(P →Q, s ) = 1 and V(∼Q→∼P, s ) = 0. Given the latter, there’s some stage s 0
such that R s s 0 and V(∼Q, s 0 ) = 1 and V(∼P, s 0 ) = 0. Given the latter, for some
s 00 , R s 0 s 00 and V(P, s 00 ) = 1. Given the former, V(Q, s 00 ) = 0. Given transitiv-
ity, R s s 00 . Given the truth of P →Q at s, either V(P, s 00 ) = 0 or V(Q, s 00 ) = 1.
Contradiction.
It’s also straightforward to use the techniques of section 6.3.4 to construct

countermodels.
Example 7.3: Show that 2 P ∨∼P . Here’s a model in which P ∨∼P is valu-
ated as 0 in stage r:
0 0 0
r P ∨∼P
∗
0

∗
a 1
P
0
As in section 6.3, we use asterisks to remind ourselves of commitments that

concern other worlds/stages. The asterisk is under ∼P in stage r because a
negation with value 0 carries a commitment to including some stage at which
the negated formula is 1. The asterisk is over the P in stage a because of the
heredity condition: a sentence letter valuated 1 carries a commitment to make
that letter 1 in every accessible stage. (Likewise, negations and conditionals
valuated as 1 generate top-asterisks, and conditionals valuated as 0 generate
bottom-asterisks). The official model:
S : {r, a}
R : {〈r, r〉, 〈a, a〉, 〈r, a〉}
I (P, a) = 1, all other atomics 0 everywhere
(I’ll skip the official models from now on.)

Example 7.4: Show that ∼∼P 2 P . Here is a countermodel:
∗
1 0 0
r
∼∼P P
∗
0

∗
1 0
a
P ∼P
∗
0
Note: since ∼∼P is 1 at r, that means that ∼P must be 0 at every stage at

which r sees. Now, Rrr, so ∼P must be 0 at r. So r must see some stage in
which P is 1. World a takes care of that.
Exercise 7.3 Establish the following facts.
a) ∼(P ∧Q) 2 ∼P ∨∼Q
b) ∼P ∨∼Q ∼(P ∧Q)
c) P →(Q∨R) 2 (P →Q)∨(P →R)
7.2.3 Soundness
Recall our proof system for intuitionism from section 3.4. What I’d like to
do next is show that that proof system is sound, relative to our semantics for
intuitionism. But first we’ll need to prove an intermediate theorem:
Generalized heredity: The heredity condition holds for all formulas. That
is, for any wff φ, whether atomic or no, and any stage, s , in any intuitionist
model, if V(φ, s ) = 1 and R s s 0 then V(φ, s 0 ) = 1.
Proof. The proof is by induction. The base case is just the official heredity
condition. Next we make the inductive hypothesis (ih): heredity is true for
formulas φ and ψ; we must now show that heredity also holds for ∼φ, φ→ψ,
φ∧ψ, and φ∨ψ. I’ll do this for φ∧ψ, and leave the rest as exercises.
∧: Suppose for reductio that V(φ∧ψ, s ) = 1, R s s 0 , and V(φ∧ψ, s 0 ) = 0.
Given the former, V(φ, s) = 1 and V(ψ, s ) = 1. By (ih), V(φ, s 0 ) = 1 and
V(ψ, s 0 ) = 1—contradiction.
Exercise 7.4 Complete the proof of generalized heredity.
Now for soundness. What does soundness mean in the present context?
The proof system in section 3.4 is a proof system for sequents, not individual
formulas. So first, we need a notion of intuitionist validity for sequents.
Definition of sequent I-validity: Sequent Γ ` φ is intuitionistically valid

(“I-valid”) iff Γ I φ
We can now formulate soundness:
Soundness for intuitionism: Every intuitionistically provable sequent is I-

valid
Proof. This will be an inductive proof. Since a provable sequent is the last
sequent in any proof, all we need to show is that every sequent in any proof
is I-valid. And to do that, all we need to show is that the rule of assumptions
generates I-valid sequents (base case), and all the other rules preserve I-validity
(induction step). For any set, Γ, valuation function V, and stage s, let’s write
“V(Γ, s) = 1” to mean that V(γ , s) = 1 for each γ ∈ Γ.
Base case: the rule of assumptions generates sequents of the form φ ` φ,
which are clearly I-valid.
Induction step: we show that the other sequent rules from section 3.4
preserve I-validity.
∧I: Here we assume that the inputs to ∧I are I-valid, and show that its
output is I-valid. That is, we assume that Γ ` φ and ∆ ` ψ are I-valid sequents,
and we must show that it follows that Γ, ∆ ` φ∧ψ is also I-valid. So, consider
any model with valuation V and any stage s such that V(Γ ∪ ∆)=1, and suppose
for reductio that V(φ∧ψ, s) = 0. Since Γ ` φ is I-valid, V(φ, s) = 1; since ∆ ` ψ
is I-valid, V(ψ, s) = 1; contradiction.
∨E: Assume that Γ ` φ∨ψ, ∆1 ,φ ` Π, and ∆2 ,ψ ` Π are all I-valid, and

suppose for reductio that V(Γ ∪ ∆1 ∪ ∆2 , s) = 1 but V(Π, s ) = 0. The first
assumption tells us that V(φ∨ψ, s) = 1, so either φ or ψ is 1 at s. If the former,
then the second assumption tells us that V(Π, s ) = 1; if the second, then the
third assumption tells us that V(Π, s) = 1. Either way, we have a contradiction.
I leave the demonstration that the remaining rules preserve I-validity as an
exercise.
Exercise 7.5 Show that ∧E, ∨I, DNI, RAA, →I, →E, and EF
preserve I-validity.
I can now justify an assertion I made, but did not prove, in section 3.4. I
asserted there that the sequent ∅ ` P ∨∼P is not intuitionistically provable.
Given the soundness proof, to demonstrate that a sequent is not intuitionisti-
cally provable, it suffices to show that its premises do not I-semantically-imply
its conclusion. But in example 7.3 we showed that 2 P ∨∼P , which is equivalent
to saying that ∅ 2 P ∨∼P .
Similarly, we showed in example 7.4 that ∼∼P 2 P . Thus, by the soundness
theorem, the sequent ∼∼P ` P isn’t provable. (Recall how, in constructing our
proof system for intuitionism in section 3.4, we dropped the rule of double-
negation elimination.)
Chapter 8
Counterfactuals1
here are certain conditionals in natural language that are not well-
T represented either by propositional logic’s material conditional or by
modal logic’s strict conditional. In this chapter we consider “counterfactual”
conditionals—conditionals that (loosely speaking) have the form:
If it had been that φ, then it would have been that ψ

For instance:
If I had struck this match, it would have lit

The counterfactuals that we typically utter have false antecedents (hence
the name), and are phrased in the subjunctive mood. They must therefore
be distinguished from English conditionals phrased in the indicative mood.
Counterfactuals are generally thought to semantically differ from indicative
conditionals. A famous example: the counterfactual conditional ‘If Oswald
hadn’t shot Kennedy, someone else would have’ is false (assuming that certain
conspiracy theories are false and Oswald was acting alone); but the indicative
conditional ‘If Oswald didn’t shoot Kennedy then someone else did’ is true
(we know that someone shot Kennedy, so if it wasn’t Oswald, it must have been
someone else.) The semantics of indicative conditionals is an important topic
in its own right, but we won’t take up that topic here.
We represent the counterfactual with antecedent φ and consequent ψ thus:
φ2→ψ
1
This section is adapted from my notes from Ed Gettier’s fall 1988 modal logic class.
207
CHAPTER 8. COUNTERFACTUALS 208
What should the logic of this new connective be, if it is to accurately represent
natural language counterfactuals?
8.1 Natural language counterfactuals

Well, let’s have a look at how natural language counterfactuals behave. Our
survey will provide guidance for our main task: developing a semantics for 2→.
As we’ll see, counterfactuals behave very differently from both material and
strict conditionals.
8.1.1 Not truth-functional

Our system for counterfactuals should have the following features:
∼P 2 P 2→Q
Q 2 P 2→Q
For consider: I did not strike the match; but it doesn’t logically follow that
if I had struck the match, it would have turned into a feather. So if 2→ is to
represent ‘if it had been that…, it would have been that…’, ∼P should not
semantically imply P 2→Q. Similarly, George W. Bush (somehow) won the last
United States presidential election, but it doesn’t follow that if the newspapers
had discovered beforehand that Bush had an affair with Al Gore, he would
still have won. So our semantics had better not count P 2→Q as a semantic
consequence of Q either.
These implications hold for the material conditional, however (for any φ
and ψ):
∼φ φ→ψ
ψ φ→ψ
We have our first difference in logical behavior between counterfactuals and

the material conditional →.
8.1.2 Can be contingent

It’s not true, presumably, that if Oswald hadn’t shot Kennedy, then someone else
would have (assuming that the conspiracy theory is false). But the conspiracy
theory might have been true; in a possible world in which there is a conspiracy,
it would be true that if Oswald hadn’t shot Kennedy, someone else would
have. Thus, our logic should allow counterfactuals to be contingent statements.
Just because a counterfactual is true, it should not follow logically that it is
necessarily true; and just because a counterfactual is false, it should not follow
logically that it is necessarily false. Our semantics for 2→, that is, should have
the following features:
P 2→Q 2 2(P 2→Q)

∼(P 2→Q) 2 →2∼(P 2→Q)
One reason this is important is that it shows an obstacle to using the strict
conditional ⇒ to represent natural language counterfactuals. For remember
that φ⇒ψ is defined as 2(φ→ψ). As a result:
φ⇒ψ S4,S5 2(φ⇒ψ)

∼(φ⇒ψ) S5 2∼(φ⇒ψ)
So if, as is commonly supposed, the logic of the 2 is at least as strong as S4, we

have a logical mismatch between counterfactuals and the ⇒.
8.1.3 No augmentation
The → and the ⇒ obey the argument form augmentation
φ→ψ φ⇒ψ
(φ∧χ )→ψ (φ∧χ )⇒ψ
That is, φ→ψ PL (φ∧χ )→ψ and φ⇒ψ K,… (φ∧χ )⇒ψ. However, natural
language counterfactuals famously do not obey augmentation. Consider:
If I were to strike the match, it would light.
Therefore, if I were to strike the match and I was in outer

space, it would light.
So, our next desideratum is that the corresponding argument should not hold
good for 2→ (that is, P 2→Q 2 (P ∧R)2→Q.)
8.1.4 No contraposition
→ and ⇒ obey contraposition:
φ→ψ φ⇒ψ
∼ψ→∼φ ∼ψ⇒∼φ
But counterfactuals do not. Suppose I’m on the firing squad, and we shoot
someone dead. My gun was loaded, but so were those of the others. Then the
premise of the following argument is true, while its consequent is false:
If my gun hadn’t been loaded, he would still be dead.
Therefore, if he weren’t dead, my gun would have been

loaded.
8.1.5 Some implications

Here is an argument form that intuitively should hold for the 2→:
φ2→ψ
φ→ψ
The counterfactual conditional should imply the material conditional. 2→ will

then obey modus ponens and modus tollens:
φ φ2→ψ ∼ψ φ2→ψ
ψ ∼φ
The reason is, of course, that modus ponens and modus tollens are valid for
the →. (Note that it’s not inconsistent to say that modus tollens holds for the
2→ and also that contraposition fails.)
Another implication: the strict conditional should imply the counterfactual:
φ⇒ψ
φ2→ψ
To see that these implications should hold, consider first the argument from
the strict conditional to the counterfactual conditional. Surely, if φ entails—
necessitates—ψ, then if φ were indeed true, ψ would be as well. As for the
counterfactual implying the material, suppose that you think that if φ were
true, ψ would also be true. Now suppose that someone tells you that φ is true,
but that ψ is false. Wouldn’t you then need to give up your original claim that
if φ were to be true, then ψ would be true? It seems so. So, the statement
φ2→ψ isn’t consistent with φ∧∼ψ—that is, it isn’t consistent with the denial
of φ→ψ.
8.1.6 Context dependence

Years ago, a few of us were at a restaurant in NY—Red Smith, Frank
Graham, Allie Reynolds, Yogi [Berra] and me. At about 11.30 p.m., Ted
[Williams] walked in helped by a cane. Graham asked us what we thought
Ted would hit if he were playing today. Allie said, “due to the better
equipment probably about .350.” Red Smith said. “About .385.” I said,
“due to the lack of really great pitching about .390.” Yogi said, “.220.” We
all jumped up and I said, “You’re nuts, Yogi! Ted’s lifetime average is .344.”
“Yeah” said Yogi “but he is 74 years old.”
–Buzzie Bavasi, baseball executive.
Who was right? If Ted Williams had played at the time the story was told,
would he or wouldn’t he have hit over .300?
Clearly, there’s no single correct answer. The first respondents were imag-
ining Williams playing as a young man. Understood that way, the answer is, no
doubt: yes, he would have hit over .300. But Berra took the question a different
way: he was imagining Williams hitting as he was then: a 74 year old man. Berra
took the others off guard, by deliberately (?—this is Yogi Berra we’re talking
about) shifting how the question was construed, but he didn’t make a semantic
mistake in so doing. It’s perfectly legitimate, in other circumstances anyway,
to take the question in Berra’s way. (Imagine Williams talking to himself five
years after he retired: “These punks today! If I were playing today, I’d still
hit over .300!”) Counterfactual sentences can be interpreted in different ways
depending on the conversational context in which they are uttered.
Another example:
If Syracuse were in Louisiana, Syracuse winters would

be warm.
True or false? It might seem true: Louisiana is in the south. But wait—perhaps
Louisiana would include Syracuse by extending its borders north to Syracuse’s
actual latitude.
Would Syracuse be warm in the winter? Would Williams hit over .300?
No one answer is correct, once and for all. Which answer is correct depends
on the linguistic context. Whether a counterfactual is true or whether it is false
depends in part on what the speaker means to be saying, and what her audience
takes her to be saying, when she utters the counterfactual. Would Syracuse be
warm?—in some contexts, it would be correct to say yes, and in others, to say
no. When we imagine Syracuse being warm, we imagine reality being different
in certain respects from actuality. In particular, we imagine Syracuse as being
in Louisiana. In other respects, we imagine a situation that is a lot like reality—
we don’t imagine a situation, for example, in which Syracuse and Louisiana
are both located in China. Now, when considering counterfactuals, there is
a question of what parts of reality we hold constant. In the Syracuse-Louisiana
case, we seem to have at least two choices. Do we hold constant the location of
Syracuse, or do we hold constant the borders of Louisiana? The truth value of
the counterfactual depends on which we hold constant.
What determines which things are to be held constant, when we evaluate
the truth value of a counterfactual? It large part: the context of utterance of
the counterfactual. Suppose I am in the middle of the following conversation:
“Syracuse restaurants struggle to survive because the climate there is so bad:
no one wants to go out to eat in the winter. If Syracuse were in Louisiana,
its restaurants would do much better.” In such a context, an utterance of the
counterfactual “If Syracuse were in Louisiana, Syracuse winters would be warm”
would be regarded as true. But if this counterfactual were uttered in the midst of
the following conversation, it would be regarded as false: “You know, Louisiana
is statistically the warmest state in the country. Good thing Syracuse isn’t in
Louisiana, because that would ruin the statistic.”
Does just saying a sentence, intending it to be true, make it true? Well,
sort of! When a certain sentence has a meaning that is partly determined by
context, then when a person utters that sentence with the intention of saying
something true, that tends to create a context in which the sentence is true.
Compare ‘flat’—we’ll say “the table is flat”, and thereby utter a truth. But when
a scientist looks at the same table and says “you know, macroscopic objects are
far from being flat. Take that table, for instance. It isn’t flat at all—when viewed
under a microscope, it can be seen to have a very irregular surface”. The term
‘flat’ has a certain amount of vagueness—how flat does a thing have to be to
count as being “flat”? Well, the amount required is determined by context.2
2
See Lewis (1979).
8.2 The Lewis/Stalnaker approach

Here is the core idea of David Lewis (1973) and Robert Stalnaker (1968) of how
to interpret counterfactual conditionals. Consider a counterfactual conditional
P 2→Q. To determine its truth, Lewis and Stalnaker instruct us to consider
the possible world that is as similar to reality as possible, in which P is true.
Then, the counterfactual is true in the actual world if and only if Q is true in
that possible world. Consider Lewis’s example:
If kangaroos had no tails, they would topple over.
When we consider the possible world that would be actual if kangaroos had
no tails, we do not depart gratuitously from actuality. For example, we do
not consider a world in which kangaroos have wings, or crutches. We do not
consider a world with different laws of nature, in which there is no gravity. We
keep the kangaroos as they actually are, but remove the tails, and we keep the
laws of nature as they actually are. It seems that the kangaroos would then fall
over.
Take the examples of the previous section, in which I got you to give
differing answers to certain sentences. Consider:
If Syracuse were in Louisiana, Syracuse winters would

be warm.
How does the contextual dependence of this sentence work, on the Lewis-
Stalnaker view? By supplying different standards of comparison of similarity.
Think about similarity, for a moment: things can be similar in certain respects,
while not being similar in other respect. A blue square is similar to a blue circle
in respect of color, not in respect of shape. Now, when we answer affirmative
to this counterfactual, according to Lewis and Stalnaker, when we consider
the possible world most similar to the actual world in which Syracuse is in
Louisiana, we are using a kind of similarity that weights heavily Louisiana’s
actual borders. When we count the counterfactual false, we are using a kind of
similarity that weights very heavily Syracuse’s actual location.
8.3 Stalnaker’s system3

I now lay out Stalnaker’s system, SC (for “Stalnaker-Counterfactuals”). The
idea is to add the 2→ to propositional modal logic.
8.3.1 Syntax of SC
The primitive vocabulary of SC is that of propositional modal logic, plus the
connective 2→. Here’s the grammar:
Definition of SC-wff:
· Sentence letters are wffs

· if φ, ψ are wffs then (φ→ψ), ∼φ, 2φ, and (φ2→ψ) are wffs
· nothing else is a wff
8.3.2 Semantics of SC
Where R is a three-place relation, let’s abbreviate “Rxy z” as “R z xy”. And,
where u is any object, let “R u ” be the two-place relation that holds between
objects x and y iff R u xy. (Think of R u as the two-place relation that results
from “plugging” up one place of the three-place relation R with object u.)
We can now define SC-models:
Definition of SC-model: An SC-model, M , is an ordered triple 〈W , , I 〉,

where:
· W is a nonempty set (“worlds”)

· I is a two-place function that assigns either 0 or 1 to each sentence letter
relative to each w ∈ W (“interpretation function”)
· is a three-place relation over W (“nearness relation”)
· The valuation function VM for M (see below) and satisfy the following
conditions:
· for any w, w is strongly connected in W

3
See Stalnaker (1968). The version of the theory I present here is slightly different from
Stalnaker’s original version; see Lewis (1973, p. 79).
· for any w, w is transitive

· for any w, w is anti-symmetric
· for any x, y, x x y (“Base”)
· for any SC-wff, φ, provided φ is true in at least one world, then for
every z, there’s some w such that VM (φ, w) = 1, and such that for
any x, if VM (φ, x) = 1 then w z x (“Limit”)
(Recall that a binary relation R is “strongly connected” in set A iff for each
u, v ∈ A, either Ruv or Rv u, and “anti-symmetric” iff u = v whenever both
Ru v and Rv u.)
Definition of SC-valuation: Where M (= 〈W , , I 〉) is any SC-model, the

SC-valuation for M , VM , is defined as the two-place function that assigns
either 0 or 1 to each SC-wff relative to each member of W , subject to the
following constraints, where α is any sentence letter, φ and ψ are any wffs, and
w is any member of W :
i) VM (α, w) = I (α, w)
ii) VM (∼φ, w) = 1 iff VM (φ, w) = 0
iii) VM (φ→ψ, w) = 1 iff either VM (φ, w) = 0 or VM (ψ, w) = 1
iv) VM (2φ, w) = 1 iff for any v, VM (φ, v) = 1
v) VM (φ2→ψ, w) = 1 iff for any x, IF [V(φ, x) = 1 and for any y such that
VM (φ, y) = 1, x w y] THEN VM (ψ, x) = 1
Phew! Let’s look into what this means.

First, notice that much of this is exactly the same as for our MPL models—
we still have the set of worlds, and formulas being given truth values at worlds.
We’ll still say that φ is true “at w” iff V(φ, w) = 1.
What happened to the accessibility relation? It has simply been dropped,
in favor of a simplified truth-clause for the 2. 2φ is now true at a world w iff
φ is true at all worlds in the model, not just all worlds accessible from w. It
turns out that this in effect just gives us an S5 logic for the 2, for you get the
same valid formulas for MPL, whether you make the accessibility relation an
equivalence relation, or a total relation. Clearly, if φ is valid in all equivalence
relation models, then it is valid in all total models, since every total relation is
an equivalence relation. What’s more, the converse is true—if φ is valid in all
total models then it’s also valid in all equivalence relation models:
Rough proof. Let φ be any formula that’s valid in all total models, and let M be
any equivalence relation model. We need to show that φ is true in an arbitrary
world r ∈ W (M ’s set of worlds). Now, any equivalence relation partitions
its domain into non-overlapping subsets in which each world sees every other
world. So W is divided up into one or more non-overlapping subsets. One of
these, W r , contains r . Now, consider a model, M 0 , just like M , but whose
set of worlds is just W r . M 0 is a total model, so φ is valid in it by hypothesis.
Thus, in this model, φ is true at r . But then φ is true at r in M , as well.
Why? Roughly: the truth value of φ at r in M isn’t affected by what goes
on outside r ’s partition, since chains of modal operators just take us to worlds
seen by r , and worlds seen by worlds seen by r , and… Such chains will never
have us “look at” anything out of r ’s partition, since these worlds are utterly
unconnected to r via the accessibility relation. So φ’s truth value at r in M is
determined by what goes on in W r , and so is the same as its truth value at r in
M 0.
So, we get the same class of valid formulas whether we require the accessibility
relation to be total, or an equivalence relation. Things are easier if we make
it a total relation, because then we can simply drop talk of the accessibility
relation and define necessity as truth at all worlds. The corresponding clause
for possibility is:
· VM (3φ, w) = 1 iff for some v, VM (φ, v) = 1
The derived clauses for the other connectives remain the same:
· VM (φ∧ψ, w) = 1 iff VM (φ, w) = 1 and V(ψ, w) = 1

· VM (φ∨ψ, w) = 1 iff VM (φ, w) = 1 or VM (ψ, w) = 1
· VM (φ↔ψ, w) = 1 iff VM (φ, w) = VM (ψ, w)
Next, what about this nearness relation? Read “x z y” as “x is at least as

near to/similar to z as is y”; thus, think of as the similarity relation between
possible worlds that we talked about before. In order to evaluate whether
x w y, we place ourselves in possible world w, and we ask whether x is more
similar to our world than y is. (Recall the bit about counterfactual conditionals
being highly context dependent. In a full treatment of counterfactuals, we
would complicate our semantics by introducing contexts of utterance, and evaluate
sentences relative to these contexts of utterance. The point of this would be to
allow different nearness relations in the different contexts.)
I say “we can think of” as a similarity relation, but take this with a grain
of salt—just as our definitions allow the members of W to be any old things, so,
is allowed to be any old relation over W . Just as the members of W could be
fish, so the relation could be any old relation over fish. (But as before, if the
truth conditions for natural language counterfactuals have nothing in common
with the truth conditions for 2→ statements in our models, the interest in our
semantics is diminished, since our models wouldn’t be modeling the behavior of
natural language counterfactuals.)
The constraints on the formal properties of the nearness relation—certain
of them, at least, seem plausible constraints on if it is to be thought of as
a similarity relation. C1 simply says that it makes sense to compare any two
worlds in respect of similarity to a given world. C2 has a transparent meaning.
C3 means “no ties”—it says that, relative to a given base world w, it is never
the case that there are two separate worlds x and y such that each is at least as
close to w as the other. C4 is the “base” axiom—it says that every world is at
least as close to itself as every other. Given C3, it has the further implication
that every world is closer to itself than every other. (We define “x is closer to w
than y is” (x ≺w y) to mean x w y and not: y w x.) C5 is called the “limit”
assumption: according to it, for any formula φ and any base world w, there
is some world that is a closest world to w in which φ is true (that is, unless φ
isn’t true at any worlds at all). This rules out the following possibility: there
are no closest φ worlds, only an infinite chain of φ worlds, each of which is
closer than the previous. Certain of these assumptions have been challenged,
especially C3 and C5. We will consider those issues below.
Note how condition C5 in the definition of an SC-model made reference
to the valuation function that we went on to define. This is in contrast to
our earlier definitions of models, in which the definition of a model made
no reference to the valuation function. The reason for this difference is that
constraint C5 (the limit assumption) is a constraint that relates the nearness
relation to the truth values of all formulas, complex or otherwise: it says that
any formula φ that is true somewhere is true in some closest-to-w world.
Given our definitions, we can define validity and semantic consequence:
· φ is SC-valid (SC φ) iff φ is true at every world in every SC-model
· Γ SC-semantically implies φ (Γ SC φ) iff for every SC-model and every
world w in that model, if every member of Γ is true at w then φ is also
true at w
8.4 Validity proofs in SC

Given this semantic system, we can give semantic validity proofs just as we did
for the various modal systems.
Example 8.1: Let’s show that the formula (P ∧Q)→(P 2→Q) is SC-valid.
We pick an arbitrary SC-model, 〈W , , I 〉, pick an arbitrary world r ∈ W , and
show that this formula is true at r :
i) Suppose for reductio that V((P ∧Q)→(P 2→Q), r ) = 0
ii) then V(P, r ) = 1, V(Q, r ) = 1, and V(P 2→Q, r ) = 0
iii) the truth condition for 2→ says that P 2→Q is true at r iff for every
closest P -world (to r ), Q is true as well. So since P 2→Q is false at r ,
there must be a closest-to-r P -world at which Q is false—that is, there
is some world a such that:
a) V(P, a) = 1
b) for any x, if V(P, x) = 1 then a r x
c) V(Q, a) = 0
iv) Since V(P, r ) = 1 (line ii)), given b), a r r
v) by “base”, r r a. So, by anti-symmetry, r = a. But now, from lines c)

and ii), Q is both true and false at r
Example 8.2: Show that SC [(P 2→Q)∧((P ∧Q)2→R)] → [P 2→R]. (This
formula is worth taking note of, because it is valid despite its similarity to the
invalid formula [(P 2→Q)∧(Q2→R)] → [P 2→R]):
i) Suppose for reductio that P 2→Q and (P ∧Q)2→R are true at r , but
P 2→R is false there.
ii) Then there’s a world, a, that is a nearest-to-r P world, in which R is

false.
iii) Since P 2→Q is true at r , Q is true in all the nearest-to-r P worlds, and
so V(Q, a) = 1.
iv) Note now that a is a nearest-to-r P ∧Q world:

a) P and Q are both true there, so P ∧Q is true there.

b) let x be any P ∧Q world. x is then a P world. But since a is a nearest-
to-r P world, we know that a r x. (Remember: “a is a nearest-to-r
P world” means: “V(P, a) = 1, and for any x, if V(P, x) = 1 then
a r x”.)
v) Since (P ∧Q)2→R is true at r , it follows that R is true at a. This contra-

dicts ii).
Exercise 8.1 Show that in the SC semantics, the counterfactual

conditional is intermediate in strength between the strict and mate-
rial conditionals. That is, show that:
a) φ⇒ψ φ2→ψ
b) φ2→ψ φ→ψ
8.5 Countermodels in SC
In this section we’ll learn how to construct countermodels in SC. Along the
way we’ll also look at how to decide whether a given formula is SC-valid or
SC-invalid. As with plain old modal logic, the best strategy is to attempt to
come up with a countermodel. If you fail, then you can use your failed attempt
to guide the construction of a validity proofs.
We can use diagrams like those from section 6.3.4 to represent SC-counter-
models. The diagrams will be a little different though. They will still contain
boxes (rounded now, to distinguish them from the old countermodels) in which
we put formulas; and we again indicate truth values of formulas with small
numbers above the formulas. But since there is no accessibility relation, we
don’t need the arrows between the boxes. And since we need to represent the
nearness relation, we will arrange the boxes vertically. At the bottom goes a box
for the world, r , of our model in which we’re trying to make a given formula
false. We string the other worlds in the diagram above this bottom world r :
the further away a world is from r in the r ordering, the further above r we
place it in the diagram. Thus, a countermodel for the formula ∼P →(P 2→Q)
might look as follows:
/.1 -,
()P Q*+
1
b
/.1 -,
()P Q*+
0
a
O
no P
/.1 0 0 -,
()∼P →(P 2→Q)*+

0
r
In this diagram, the world we’re primarily focusing on is the bottom world,
world r. The nearest world to r is world r itself. The next nearest world to r is
the next world moving up from the bottom: world a. The furthest world from
r is world b. Notice that P is false at world r, and true at worlds a and b. Thus,
a is the nearest world to r in which P is true. Since Q is false at world a, that
makes the counterfactual P 2→Q false at world r . Since ∼P is true at r, the
entire material conditional ∼P →(P 2→Q) is false at world r, as desired. (World
b isn’t needed in this countermodel; I included it merely for illustration.) The
“no P ” sign to the left of worlds a and r is a reminder to ourselves in case we
want to add further worlds to the diagram: don’t include any worlds between a
and r in which P is true. Otherwise world a would no longer be the nearest P
world.
What strategy should one use for constructing SC-countermodels? As we
saw in section 6.3.4, a good policy is to make “forced” moves first. For example,
if you are committed to making a material conditional false at a world, go
ahead and make its antecedent true and consequent false in that world, right
away. In fact, a false counterfactual also forces certain moves. It follows from
the truth condition for the 2→ that if φ2→ψ is false at world w, then there
exists a nearest-to-w φ world at which ψ is false. So if you put a 0 overtop of
a counterfactual φ2→ψ in some world w, it’s good to do the following two
things right away. First, add a nearest-to-w world in which φ is true (if such a
world isn’t already present in your diagram). And second, make ψ false there.
True counterfactuals don’t force your hand quite so much, since there are
two ways for a counterfactual to be true. If φ2→ψ is true at w, then ψ must be
true at every nearest-tow φ world. This could happen, not only if there exists
a nearest-to-w φ world in which ψ is true, but also if there are no nearest-to-w
φ worlds. In the latter case we say that φ2→ψ is “vacuously true” at w. A

counterfactual can be vacuously true only when its antecedent is necessarily
false, since the limit assumption guarantees that if there is at least one φ world,
then there is a nearest φ world. So: if you want to make a counterfactual true at
a world, it’s a good idea to wait until you’ve been forced to make its antecedent
true at at least one world. Only when this has happened, thus closing off
the possibility of making the counterfactual vacuously true, should you add a
nearest world in which its antecedent is true, and make its consequent true at
that nearest antecedent-world.
Suppose, for example, that we want to show that the following formula is
SC-invalid: [(P 2→Q)∧(Q2→R)] → (P 2→R). We begin as follows:
/. 1 1 1 0 0
-,
()[(P 2→Q)∧(Q2→ R)]→(P 2→ R)*+

r
In keeping with the advice I gave a moment ago, let’s deal with the false
counterfactual first: let’s make P 2→R false in r. This means that we need to
add a nearest-to-r P world in which R is false. At this point, nothing prevents
us from making this world r itself, but that might collide with other things we
might want to do later, so I’ll make this nearest-to-r P world a distinct world
from r:
/.1 0 1
-,
()P Q*+
a
O R
no P
/. 0 1 1 1 0 0
-,
()[(P 2→Q)∧(Q2→ R)]→(P 2→ R)*+

r
“No P ” reminds me not to add any P -worlds between a and r. Since world r is
in the “no P zone”, I made P false there.
Notice that I made Q true in a. This is because P 2→Q is true in r . This
formula says that Q is true in the nearest-to-r P world; and a is the nearest-to-r
P world. In general, whenever you add a new world to one of these diagrams,
you should go back to all the counterfactuals in the bottom world and see
whether they require their consequents to have certain truth values in the new
world.
Now for the final counterfactual Q2→R. This can be true in two ways—
either there is no Q world at all (the vacuous case), or there is a nearest-to-r
Q world in which R is true. Q is already true in world a, so the vacuous
case is ruled out. So we must include a nearest-to-r Q world, call it “b”, and
make R true there. Where will we put this new world b? There are three
possibilities. World b could be farther away from, identical to, or closer to r
than a. (These are the only three possibilities, given anti-symmetry.) Let’s try
the first possibility:
/.1 -,
()Q R*+
1
b
O
/.1 0 1
-,
()P Q*+
O
a
R no Q
no P
/. 0 1 0 1 1 0 0
-,
()[(P 2→Q)∧(Q2→ R)]→(P 2→ R)*+
r
This doesn’t work, because world a is in the no-Q zone, but Q is true at world
a. Put another way: in this diagram, b isn’t the nearest-to-r Q world; world a
is. And so, since R is false at world a, the counterfactual Q2→R would come
out false at world r, whereas we want it to be true. we’ve got Q true at a nearer
world—namely, a.
Likewise, we can’t make world b be identical to world a, since we need to
make R true in b and R is already false in a.
But the final possibility works out just fine—let world b be closer to r than
a:
/.1 0 1
-,
()P Q*+
a
O R
/.1 -,
()Q P*+
1 0
no P b
O
R
no Q
/. 0 1 0 1 1 0 0
-,
()[(P 2→Q)∧(Q2→ R)]→(P 2→ R)*+
r
Notice that I made P false in b, since b is in the no P zone. Here’s the official
model:
W = {r, a, b}
r = {〈b, a〉 . . . }
I (P, a) = I (Q, a) = I (Q, b) = I (R, b) = 1, all others 0
In this official model I left out a lot in giving the similarity relation for this
model. First, I left out some of the elements of r . Fully written out, it would
be:
r = {〈r, b〉, 〈b, a〉, 〈r, a〉, 〈r, r〉, 〈a, a〉, 〈b, b〉}
I left out 〈r, b〉 because it gets included automatically given the “base” assump-
tion (C4). Also, the element 〈r, a〉 is required to make r transitive. The
elements 〈r, r〉, 〈a, a〉, and 〈b, b〉 were entered to make that relation reflexive.
(Why must it be reflexive? Because reflexivity comes from strong connectivity.
Let w and x be any members of W ; we get (x w x or x w x) from Strong
connectivity of w , and hence x w x.) My policy will be to write out enough
of r so that the rest can be inferred, given the definition of an SC-model. Sec-
ondly, this isn’t a complete writing out of itself; it is just r . To be complete,
we’d need to write out a , and b . But in this case, these latter two parts of
don’t matter, so I omitted them. (Later we’ll consider cases where we need to
consider more of than simply r .)
Example 8.3: Is the formula (P 2→R)→((P ∧Q)2→R) valid or invalid? (This
is the formula corresponding to the inference pattern of augmentation.) The
best approach to such problems is to first attempt to find a countermodel. In

this case we succeed:
/.1 1 1 0
-,
()P ∧Q R*+
a
O
/.1 -,
()P Q*+
1 0
no P ∧Q b
O
R
/. 0 1 -,
no P
0 0
()(P 2→ R)→[(P ∧Q)2→ R)*+
r
I began with the false: (P ∧Q)2→R. This forced the existence of a nearest
P ∧Q world, in which R was false. But since P ∧Q was true there, P was true
there; this ruled out the true P 2→R in r being vacuously true. So I was forced
to consider the nearest P world. It couldn’t be farther out than a, since P is
true in a. It couldn’t be a, since R was already false there. So I had to put it
nearer than a. Notice that I had to make Q false at b. Why? Well, it was in the
“no P ∧Q zone”, and I had made P true in it. Here’s the official model:
W = {r, a, b}
r = {〈b, a〉 . . . }
I (P, a) = I (Q, a) = I (P, b) = I (R, b) = 1, all else 0
Example 8.4: Determine whether SC 3P →[(P 2→Q)→∼(P 2→∼Q)]. An
attempt to find a countermodel fails at the following point:
/. 1
-,
()P ∼Q*+
a 1 1 0
O
no P
/.1 0 0 -,
()3P →[(P 2→Q)→∼(P 2→∼Q)]*+

1 0 0 1
r
At world a, I’ve got Q being both true and false. A word about how I got
to that point. I noticed that I had to make two counterfactuals true: P 2→Q
and P 2→∼Q. Now, this isn’t a contradiction all by itself. Remember that
counterfactuals are vacuously true if their antecedents are impossible. So if P
were impossible, then both of these would indeed be true, without any problem.
But 3P has to be true at r. This rules out those counterfactual’s being vacuously
true. Since P is possible, the limit assumption has the result that there is a closest
P world. This then with the two true counterfactuals created the contradiction.
This reasoning is embodied in the following semantic validity proof:
i) Suppose for reductio that (P 2→Q)→∼(P 2→∼Q) is false at some world

r in which 3P is true.
ii) Then P 2→Q and P 2→∼Q are both true at r as well.
iii) Since 3P is true at r , P is true at some world. So, by the limit assumption,
we have: there exists a world, a, such that V(P, a) = 1 and for any x, if
V(P, x) = 1 then a r x. For short, a is a closest-to-r P world.
iv) The truth condition for 2→, applied to P 2→Q, gives us that Q is true
at all the closest-to-r P worlds.
v) Similarly, applied to P 2→∼Q, we know that ∼Q is true at all the closest-

to-r P worlds.
vi) Thus, both Q and ∼Q would be true at a. Impossible.
Note the use of the limit assumption. It is the limit assumption we must use
when we need to know that there is a nearest φ-world, in cases where we can’t
get this knowledge from other things in the proof.
Cases where one counterfactual is nested within another call for something
new. Let’s consider how to show that [P 2→(Q2→R)]→[(P ∧Q)2→R] is SC-
invalid (this is the formula corresponding to “importation”). We begin by
making the formula false in r, the actual world of the model. This means making
the antecedent true and the consequent false. Now, since the consequent is a
false counterfactual, we are forced to make there be a nearest P ∧Q world in
which R is false:
/.1 1 1 -,
()P ∧Q R*+
0
a
O
no P ∧Q
/. -,
()[P 2→(Q2→ R)]→[(P ∧Q)2→ R]*+

1 0 0
r
Next we must make P 2→(Q2→R) true. We can’t make it vacuously true,

because we’ve already got a P -world in the model: a. So, we’ve got to put in
the nearest-to-r P world. Could it be farther away than a? No, because a would
be a closer P world. Could it be a? No, because we’ve got to make Q2→R true
in the closest P world, and since Q is true but R is false in a, Q2→R is already
false in a. So, we do it as follows:
/.1 1 1 -,
()P ∧Q R*+
0
a
O
/.1 -,
()P Q2→R*+
0 1
no P ∧Q b
O
no P
/. 0 1 -,
()[P 2→(Q2→ R)]→[(P ∧Q)2→ R]*+
0 0
r
(Why did I make Q false at b? Well, because b is in the no P ∧Q zone, and P is

true at b, so Q had to be false there.)
The remaining thing to do is to make Q2→R true at b. This requires some
thought. The diagram right now represents “the view from r”—it represents
how near the worlds in the model are to r. That is, it represents the r relation.
But the truth value of Q2→R at b depends on “the view from b”—that is, the
b relation. So we need to consider a new diagram, in which b is the bottom
world:
/.1 -,
()Q R*+
1
c
O
no Q
/.1 -,
()P Q2→R*+
0 1
b
I made there be a nearest-to-b Q world, and made R true there. Notice that
I kept the old truth values of b from the other diagram. This is because this
new diagram is a diagram of the same worlds as the old diagram; the difference
is that the new diagram represents the b nearness relation, whereas the old
one represented a different relation: r . Now, this diagram isn’t finished. The
diagram is that of the b relation, and that relation relates all the worlds in the
model. So, worlds r and a have to show up somewhere here. The safest practice
is to put them far away from b, so that there isn’t any possibility of conflict
with the no Q zone that has been established. Thus, the final appearance of
this part of the diagram is as follows:
/. -,
r () *+
/. -,
a () *+
/.1 -,
()Q R*+
1
c
O
no Q
/.1 -,
()P Q2→R*+
0 1
b
The old truth values from worlds r and a are still in effect (remember that this
is another diagram of the same model, but representing a different nearness
relation), but I left them out because of the fact that they’ve already been
written on the other part of the diagram.
Notice that the order of the worlds in the r-diagram does not in any way
affect the order of the worlds on the b diagram. The nearness relations in
the two diagrams are completely independent, because in our definition of

‘SC-model’, we entered in no conditions constraining the relations between i
and j when i 6= j . This sometimes seems unintuitive. For example, we could
have two halves of a model looking as follows:
The view from r The view from a

c r
b b
a c
r a
It might, for example, seem odd that b is physically closer to a than to c in the
view from r, but not in the view from a. But remember that in any diagram,
only some of the features are intended to be genuinely representative. These
diagrams are in ink, but this is not intended to convey the idea that the worlds in
the model are made of ink. This feature of the diagram isn’t intended to convey
information about the model. Analogously, the fact that in b is physically closer
to a than to c in the view from r is not intended to convey the information that,
in the model, ba c. In fact, the diagram of the view from r is only intended to
convey information about r ; it doesn’t carry any information about a , b , or
c .
Back to the countermodel. That other part of the diagram, the view from r,
must be updated to include world c. The safest procedure is to put c far away
on the model to minimize possibility of conflict. Thus, the final picture of the
view from r is:
/. -,
c () *+
/.1 1 1 -,
()P ∧Q R*+
0
a
O
/.1 -,
()P Q2→R*+
0 1
no P ∧Q b
O
no P
/. 0 1 -,
()[P 2→(Q2→ R)]→[(P ∧Q)2→ R]*+
0 0
r
Again, I haven’t re-written the truth values in world c, because they’re already
in the other diagram, but they are to be understood as carrying over. Now for
the official model:
W = {r, a, b, c}
r = {〈b, a〉, 〈a, c〉 . . . }
b = {〈c, a〉, 〈a, r〉 . . . }
I (P, a) = I (Q, a) = I (P, b) = I (Q, c) = I (R, c) = 1, all else 0
Notice that we needed to express two of ’s subrelations: r and b . Remember
that any model has got to contain i for every world i in the model. For
example, if we were to write out this model completely officially, we’d have to
specify a and c . But we don’t bother with those parts of that don’t matter.
Exercise 8.2 Determine whether the following wffs are SC-valid

or invalid. Give a falsifying model for every invalid wff, and a
semantic validity proof for every valid wff.
a) 3P →[(P 2→Q)↔∼(P 2→∼Q)]
b) [P 2→(Q→R)]→[(P ∧Q)2→R]
c) [P 2→(Q2→R)]→[Q2→(P 2→R)]
8.6 Logical Features of SC

Here we’ll discuss various features of SC, which appear to confirm that it’s a
good logic for counterfactual conditionals. In part, we will be showing that our
semantics for 2→ matches the logical features of natural language that were
discussed in section 8.1.
8.6.1 Not truth-functional

We wanted our system for counterfactuals to have the following features:
∼P 2 P 2→Q
Q 2 P 2→Q
Clearly, it does. The first fact is demonstrated by a model in which P is false

at some world, r, and in which there’s a nearest-to-r P world in which Q is
false; the second, by a model in which Q is true at r, P is false at r, and in which
there’s a nearest-to-r P world in which Q is false.
8.6.2 Can be contingent

We wanted it to turn out that:
P 2→Q 2 2(P 2→Q)

∼(P 2→Q) 2 →2∼(P 2→Q)
Our semantics does indeed have this result, because the similarity metrics based
on different worlds can be very different. For example: consider a model with
worlds r and a, in which Q is true in the nearest-to-r P world, but in which Q
is false at the nearest-to-a P world. P 2→Q is true at r and false at a, whence
2(P 2→Q) is false at r.
8.6.3 No augmentation
In example 8.3 we produced a model containing a world in P 2→Q was true
but (P ∧R)2→Q was false. Thus, P 2→Q 2 (P ∧R)2→Q.
8.6.4 No contraposition
Let’s show that P 2→Q 2 ∼Q2→∼P :
/.1 0 -,
()∼Q ∼P*+
0 1
a
O
/.1 -,
()P Q*+
1
no ∼Q b
O
no P
/. -,
()(P 2→Q) (∼Q2→∼P )*+
1 0
r
I won’t bother with the official model.
8.6.5 Some implications

Exercises 8.1a and 8.1b show that the SC semantics vindicates the inference
from the strict to the counterfactual conditional, and from the counterfactual
conditional to the material conditional.
8.6.6 No exportation
We have shown that the SC-semantics reproduces the logical features of natural
language counterfactuals discussed in section 8.1. In the next few sections we
discuss some further logical features of the SC-semantics, and compare them
with the logical features of the →, the ⇒, and natural language counterfactuals.
The → obeys exportation:
(φ∧ψ)→χ
φ→(ψ→χ )
But the ⇒ doesn’t in any system; (P ∧Q)⇒R 2S5 P ⇒(Q⇒R). Nor does
the 2→; it can be easily shown with a countermodel that (P ∧Q)2→R 2SC
P 2→(Q2→R).
Does the natural language counterfactual obey exportation? Here is an
argument that it does not. The following is true:
If Bill had married Laura and Hillary, he would have

been a bigamist.
But one can argue that the following is false:
If Bill had married Laura, then it would have been the

case that if he had married Hillary, he would have been
a bigamist.
Suppose Bill had married Laura. Would it then have been true that: if he had
married Hillary, he would have been a bigamist? Well, let’s ask for comparison:
what would the world have been like, had George W. Bush married Hillary
Rodham Clinton? Would Bush have been a bigamist? Here the natural answer
is no. George W. Bush is in fact married to Laura Bush; but when imagining him
married to Hillary Rodham Clinton, we don’t hold constant his actual marriage.
We imagine him being married to Hillary instead. If this is true for Bush, then
one might think it’s also true for Bill in the counterfactual circumstance in
which he’s married to Laura: it would then have been true of him that, if he
had married Hillary, he wouldn’t have still been married to Laura, and hence
would not have been a bigamist.
It’s unclear whether this is a good argument, though, since it assumes that
ordinary standards for evaluating unembedded counterfactuals (“If George had
married Hillary, he would have been a bigamist”) apply to counterfactuals
embedded within other counterfactuals (“If Bill had married Hillary, he would
have been a bigamist” as embedded within “If Bill had married Laura then…”.)
Contrary to the assumption, it seems most natural to evaluate the consequent of
an embedded counterfactual by holding its antecedent constant. But a defender
of the SC semantics might argue that the second displayed counterfactual
above has a reading on which it is false (recall the context-dependence of
counterfactuals), and hence that we need a semantics that allows for the failure
of exportation.
8.6.7 No importation
Importation holds for →, and for ⇒ in T and stronger systems:
φ→(ψ→χ )φ⇒(ψ⇒χ )
(φ∧ψ)→χ (φ∧ψ)⇒χ
but not for the 2→: above we produced an SC-model with a world in which
the conditional [P 2→(Q2→R)]→[(P ∧Q)2→R] was false.
The status of importation for natural language counterfactuals is similar to
the status of exportation. One can argue that the following is true, at least on
one reading:
If Bill had married Laura, then it would have been the

case that if he had married Hillary, he would have been
happy.
without the result of importing being true:
If Bill had married Laura and Hillary, he would have

been happy
(if he had married both he would have become a public spectacle, which would
have made him most unhappy.)
8.6.8 No hypothetical syllogism (transitivity)

The following inference pattern is valid for → and ⇒ (in all systems):
ψ→χ φ→ψ ψ⇒χ φ⇒ψ

φ→χ φ⇒χ
But the model produced above invalidating [(P 2→Q)∧(Q2→R)]→(P 2→R)

contained a world at which P 2→Q and Q2→R were both false but in which
P 2→R was true.
Natural language counterfactuals also seem not to obey hypothetical syllo-
gism. I am the oldest child in my family; my brother Mike is the second-oldest.
So the following counterfactuals seem true:4
4
Note that if these two statements are written in the reverse order, it seems far less clear
that they’re both true: “If my parents had never met, I wouldn’t have been born; If I hadn’t
been born, Mike would have been my parent’s oldest child.” It’s natural in this case to interpret
the second conditional by holding constant the antecedent of the first conditional. This fact,
together with what we observed about embedded counterfactuals in section 8.6.6, suggests a
systematic dependence of the interpretation of counterfactuals on their immediate linguistic
context. See von Fintel (2001) for a “dynamic” semantics for counterfactuals, which more
accurately models this feature of their use, and also makes sense of how hard it is to hear the
readings argued for in sections 8.6.6, 8.6.7, and 8.6.8.
If I hadn’t been born, Mike would have been my parent’s

oldest child.
If my parents had never met, I wouldn’t have been born.
But the result of applying hypothetical syllogism seems false:
If my parents had never met, Mike would have been their

oldest child.
8.6.9 No transposition
Transposition governs the →:
φ→(ψ→χ )
ψ→(φ→χ )
but not the ⇒ (in any of our modal systems); P ⇒(Q⇒R) 2S5 Q⇒(P ⇒R). Nor
does it govern the 2→; it’s easy to show that P 2→(Q2→R) 2SC Q2→(P 2→R).
The status of transposition for natural language counterfactuals is sim-
ilar to that of importation and exportation. If we can ignore the effects of
embedding on the evaluation of counterfactuals, then we have the following
counterexample to transposition. It is true that:
If Bill Clinton had married Laura Bush, then if he had

married Hillary Rodham, he’d have been married to a
Democrat.
But it is not true that:
If Bill Clinton had married Hillary Rodham, then if he

had married Laura Bush, he’d have been married to a
Democrat.
8.7 Lewis’s criticisms of Stalnaker’s theory

David Lewis has a rival theory of counterfactuals. Like Stalnaker’s theory, it is
based on similarity, but it differs from Stalnaker’s in certain respects. So far, we
have only discussed features of Stalnaker’s system that are shared by Lewis’s.
Let’s turn, now, to the differences.
Lewis challenges Stalnaker’s assumption that w is always anti-symmetric.
Real similarity relations permit ties. So it seems implausible to rule out the
possibility of two worlds being exactly similar to a given world.
The challenge to Stalnaker here is most straightforward if Stalnaker intends
to be giving truth conditions for natural language counterfactuals, rather than
merely doing model theory. In that case, the set W in an SC-model must be the
set of genuine possible worlds, and must be a relation of genuine similarity,
in which case it ought to admit ties. But even if Stalnaker is not doing this,
the objection may yet have bite, to the extent that the semantics of natural
language conditionals is like similarity-theoretic semantics.5
The validity of certain formulas depends on the “no ties” assumption; the
following two wffs are SC-valid, but are challenged by Lewis:
(P 2→Q) ∨ (P 2→∼Q) (“Conditional excluded middle”)

[P 2→(Q∨R)] → [(P 2→Q)∨(P 2→R)] (“distribution”)
Take the first one, for example. Suppose you gave up anti-symmetry, thereby
allowing ties. Then the following would be a countermodel for the law of
conditional excluded middle:
/.1 -, /.1 -,
()P ∼Q*+
0
()P Q*+
0 1
a
O b
no Q
/. 0 0 0
-,
()(P 2→Q)∨(P 2→∼Q)*+

r
Remember that P 2→Q is true only if Q is true in all the nearest P worlds.
In this model, Q is true in one of the nearest P worlds, but not all, so that
counterfactual is false at r. Similarly for P 2→∼Q.
A similar model shows that distribution fails if the “no ties” assumption is
given up.
So, should we give up conditional excluded middle? As Lewis concedes,
the principle is initially plausible. An equivalent formulation of conditional
5
For an interesting response to Lewis, see Stalnaker (1981).
excluded middle is this:
∼(P 2→Q)→(P 2→∼Q)
But everyone agrees that (P 2→∼Q)→∼(P 2→Q) is always true, at least, when
P is possibly true. So, in cases where P is possibly true anyway, the question
of whether conditional excluded middle is valid is the question of whether
∼(P 2→Q) and P 2→∼Q are equivalent to each other. And it does indeed
seem that in ordinary usage, one expresses the negation of a counterfactual
by negating its consequent. To deny the counterfactual “if she had played,
she would have won", one says "no, she wouldn’t have!”, meaning “if she had
played, she would not have won”.
And take the other formula validated by Stalnaker’s theory, distribution. In
reply to: “if the coin had been flipped, it would have come out either heads or
tails”, one might ask: “which would it have been, heads or tails?”. The thinking
behind the reply is that “if the coin had been flipped, it would have come up
heads”, or “if the coin had been flipped, it would have come up tails” must be
true.
So there’s some plausibility to both these formulas. But Lewis says two
things. The first is metaphysical: if we’re going to accept the similarity analysis,
we’ve got to give them up, because ties just are possible. The second is purely
semantic: the intuitions aren’t completely compelling. About the coin-flipping
case, Lewis denies that if the coin had been flipped, it would have come up
heads, and he also denies that if the coin had been flipped, it would have come
up tails. Rather, he says, if it had been flipped, it might have come up heads.
And if it had been flipped, it might have come up tails. But neither outcome is
such that it would have resulted, had the coin been flipped.
Concerning excluded middle, Lewis says:
It is not the case that if Bizet and Verdi were compatriots, Bizet would be
Italian; and it is not the case that if Bizet and Verdi were compatriots, Bizet
would not be Italian; nevertheless, if Bizet and Verdi were compatriots,
Bizet either would or would not be Italian. (Counterfactuals, p. 80.)
Lewis can follow this up by noting that if Bizet and Verdi were compatriots,
Bizet might be Italian, but it’s not the case that if they were compatriots, he
would be Italian.
Here is a related complaint of Lewis’s about Stalnaker’s semantics. In the
last little bit, I’ve used English phrases of the form “if it were the case that φ,
then it might have been the case that ψ”. This conditional Lewis calls “the
‘might’ counterfactual”; he symbolizes it as φ3→ψ, and defines it thus:
· “φ3→ψ” is short for “∼(φ2→∼ψ)”
Lewis criticizes Stalnaker’s system for the fact that this definition of 3→ doesn’t
work in Stalnaker’s system. Why not? Well, since internal negation is valid in
Stalnaker’s system, φ3→ψ would always imply φ2→ψ—not good, since the
might-conditional in English seems weaker than the would-conditional. So,
Lewis’s definition of 3→ doesn’t work in Stalnaker’s system. Moreover, there
doesn’t seem to be any other plausible definition. So, Stalnaker can’t define
3→.6
Lewis also objects to Stalnaker’s limit assumption. The following line is
less than one inch long:
Now, consider the counterfactual:
If the line were more than one inch long, it would be

over one hundred miles long.
Seems false. But if we use Stalnaker’s truth conditions as truth conditions for
natural language counterfactuals, and take our intuitive judgments of similarity
seriously, we seem to get the result that it is true! The reason is that there
doesn’t seem to be a closest world in which the line is more than one inch long.
For every world in which the line is, say, 1 + k inches long, there’s another
world in which the line has a length closer to its actual length but still more
than one inch long: say, 1 + k2 inches. So there doesn’t seem to be any closest
world in which the line is over one inch long.
In light of these criticisms, Lewis proposes a new similarity-based seman-
tics for counterfactuals, which assumes neither anti-symmetry nor the limit
assumption. Let’s look at that system.
6
Lewis (1973, p. 80).
8.8 Lewis’s system7

To move from Stalaker’s system to Lewis’s, we can start by just dropping the
anti-symmetry assumption. We also want to drop the limit assumption. But
after dropping the limit assumption, if we made no further adjustments to the
system, we would get unwanted vacuous truths, as we did in the example of the
one-inch long line above.8 The truth definition for 2→ needs to be changed.
Instead of saying that φ2→ψ is true iff ψ is true in all the nearest φ worlds,
we will instead say that φ2→ψ is true iff either i) φ is true in no worlds (the
vacuous case), or ii) there is some φ world such that for every φ world at least
as close, φ→ψ is true there.
Here is the new system, LC (Lewis-counterfactuals). It is exactly the same
as the Stalnaker system except that limit and anti-symmetry are dropped, and
the parts indicated in boldface are changed:
Definition of LC-model: An LC-model, M , is an ordered triple 〈W , , I 〉,

where:
· W is a nonempty set
· I is a function that assigns either 0 or 1 to each sentence letter relative
to each member of W
· is a three-place relation over W
· The valuation function, VM , for M (see below) and satisfy the following
conditions:
· for any w, w is strongly connected

· for any w, w is transitive
· for any x, y, if y x x then x = y (Lewis’s “base”)
Definition of LC-valuation: Where M (= 〈W , , I 〉) is any LC-model, the

valuation for M , VM , is defined as the two-place function that assigns either
0 or 1 to each wff relative to each member of W , subject to the following
7
See Lewis (1973, pp. 48-49). My formulation does away with the accessibility relation (in
Lewis’s terminology, Si , the set of worlds accessible from world i , is always W , the set of all
worlds in the model), so it is a bit simpler.
8
Actually, dropping the limit assumption doesn’t affect the class of valid formulas, which is
the same with or without the limit assumption (Lewis, 1973, p. 121).
constraints, where α is any sentence letter, φ and ψ are any wffs, and w is any
member of W :
· VM (α, w) = I (α, w)
· VM (∼φ, w) = 1 iff VM (φ, w) = 0
· VM (φ→ψ, w) = 1 iff either VM (φ, w) = 0 or VM (ψ, w) = 1
· VM (2φ, w) = 1 iff for any v, VM (φ, v) = 1
· VM (φ2→ψ, w) = 1 iff EITHER φ is true at no worlds, OR: there is
some world, x, such that VM (φ, x) = 1 and for all y, if y w x then
VM (φ→ψ, y) = 1
It may be verified that every LC-valid wff is SC-valid.9 The converse is not
true, as the discussion of conditional excluded middle in the previous section
shows.
Comments on all this: First, notice that the limit and anti-symmetry con-
ditions are simply dropped. Second, the Base condition is modified; now it
says that no world is as close to a world as itself. Before, it said that each world
is at least as close to itself as any other. Stalnaker’s Base condition, plus anti-
symmetry, entails the present Base condition. But Lewis’s system doesn’t have
anti-symmetry, so the Base condition must be stated in the stronger form.10
Third, let’s think about what the truth condition for the 2→ says. First,
there’s the vacuous case: if φ is necessarily false then φ2→ψ comes out true.
But if φ is possibly true, then what the clause says is this: φ2→ψ is true at
w iff there’s some φ world where ψ is true, such that no matter how much
closer to w you go, you’ll never get a φ world where ψ is false. If there is a
nearest-to-w φ world, then this implies that φ2→ψ is true at w iff ψ is true in
all the nearest-to-w φ worlds.
So, thinking of these as truth-conditions for natural-language counterfac-
tuals for a moment, recall the sentence:
9
Let’s say that an LC model is “Stalnaker-acceptable” iff it obeys the limit and anti-symmetry
assumptions. Suppose that φ is LC-valid. Then it’s true in all Stalnaker acceptable LC-models.
Now, notice that in Stalnaker-acceptable models, Lewis’s truth-conditions for formulas yield
the same results as Stalnaker’s (exercise 8.3). So, φ must be true in all SC-models.
10
Why do we want to prohibit worlds being just as close to w as w is to itself? So that P ∧Q
semantically implies P 2→Q. Otherwise P ∧Q could be true at w while P ∧∼Q was true at
some world as close to w as w is to itself, in which case P 2→Q would turn out false at w.
If the line were over one inch long, it would be over ten
inches long.
There’s no nearest world in which the line is over one inch long, only an infinite
series of worlds where the line has lengths getting closer and closer to one
inch long. But this doesn’t make the counterfactual true. A counterfactual is
true if its antecedent is impossible, but that’s not true in this case. So the only
way the counterfactual could be true is if the second part of the definition is
satisfied—if, that is, there is some world, x, such that the antecedent is true
at x, and the material conditional (antecedent→consequent) is true at every
world at least as similar to the actual world as is x. Since the “at least as similar
as” relation is reflexive, this can be rewritten thus:
· for some world, x, the antecedent and consequent are both true at x, and
the material conditional (antecedent→consequent) is true at every world
at least as similar to the actual world as is x
So, is there any such world, x? No. For let x be any world at which the
antecedent and consequent are both true—i.e., any world in which the line is
over ten inches long. We can always find a world that is more similar to the
actual world than x in which the material conditional (antecedent→consequent)
is false: just choose a world just like x but in which the line is only, say, two
inches long.
Let’s see how Lewis’s theory works in the case of a true counterfactual, for
instance:
If I were more than six feet tall, then I would be less than
nine feet tall
(I am, in fact, less than six feet tall.) The situation here is similar to the previous
example in that there is no nearest world in which the antecedent is true. But
now, we can find a world x, in which the antecedent and consequent are both
true, and such that the material conditional (antecedent→ consequent) is true
in every world at least as similar to the actual world as is x. Simply take x to
be a world just like the actual world but in which I am, say, six-feet-one. Any
world that is at least as similar to the actual world as this world must be one in
which I’m less than nine feet tall; so in any such world the material conditional
(I’m more than six feet tall→I’m less than nine feet tall) is true.
Notice that the formulas representing Conditional Excluded Middle and

Distribution come out invalid now, because of the possibility of ties.
Another thing: Lewis gives the following definition for the ‘might’-counter-
factual:
· “φ3→ψ” is short for “∼(φ2→∼ψ)”
From this we may obtain a derived clause for the truth conditions of φ3→ψ:
· VM (φ3→ψ, w) = 1 iff for some x, VM (φ, x) = 1, and for any x, if

VM (φ, x) = 1 then there’s some y such that y w x and VM (φ∧ψ, y) = 1)
That is, φ3→ψ is true at w iff φ is possible, and for any φ world, there’s a
world as close or closer to w in which φ and ψ are both true. In cases where
there is a nearest φ world, this means that ψ must be true in at least one of the
nearest φ worlds.
Exercise 8.3 Show that in any Lewis model in which the limit
and anti-symmetry conditions hold, Lewis’s truth conditions reduce
to Stalnaker’s. That is, in any such model, a wff counts as being
true at a given world given Lewis’s definition of truth in a model if
and only if it counts as being true at that world given Stalnaker’s
definition.
8.9 The problem of disjunctive antecedents

Before we leave counterfactual conditionals, I want to talk about one criticism
that has been raised against both Lewis’s and Stalnaker’s systems.11 In neither
system does the formula (P ∨Q)2→R semantically imply P 2→R. (Take a model
where there is a unique nearest P ∨Q world to r, in which Q is true but not P ;
and make there be a unique nearest P world in which R is false.) But shouldn’t
this implication hold? Imagine a conversation between Butch Cassidy and
the Sundance Kid in heaven, after having been surrounded and killed by the
Bolivian army. They say:
11
For references, see the bibliography of Lewis (1977).
If we had surrendered or tried to run away, we would

have been shot.
Intuitively, if this is true, so is this:
If we had surrendered, we would have gotten shot.
In general, one is entitled to conclude from “If P or Q had been the case,
then R would have been the case” that “if P had been the case, R would have
been the case”. If Butch Cassidy and the Sundance Kid could have survived by
surrendering, they certainly would not say to each other “If we had surrendered
or tried to run away, we would have been shot”.
Is this a problem for Lewis and Stalnaker? Some have argued this, but
others respond as follows. One must take great care in translating from natural
language into logic. For example,12 no one would want to criticize the law
∼∼P →P on the grounds that “There ain’t no way I’m doing that” doesn’t
imply that I might do that. And there are notorious peculiar things about the
behavior of ‘or’ in similar contexts. Consider:
You are permitted to stay or go.
One can argue that this does not have the form:
You are permitted to do the action: (Stay ∨ Go)
After all, suppose that you are permitted to stay, but not to go. If you stay, you
can’t help doing the following act: staying ∨ going. So, surely, you’re permitted
to do that. So, the second sentence is true. But the first isn’t; if someone uttered
it to you when you were in jail, they’d be lying to you! It really means:
You are permitted to stay, AND you are permitted to go.
Similarly, “If either P or Q were true then R would be true” seems usually to
mean “If P were true then R would be true, and if Q were true then R would be
true”. We can’t just expect natural language to translate directly into our logical
language—sometimes the surface structure of natural language is misleading.
12
The example is adapted from Loewer (1976).
Chapter 9
Quantified Modal Logic
e’re going to look at possible-worlds semantics for quantified modal

W logic—QML. The language is what you get by adding the 2 and 3 to
the language of predicate logic. There are many interesting issues concerning
the interaction of modal operators with quantifiers.
9.1 Grammar of QML

The grammar of the language of QML is exactly what you’d expect: that of
plain old predicate logic, but with the 2 added. Thus, the one new clause to
the definition of a predicate-logic wff is the clause that if φ is a wff, then so
is 2φ. (3φ continues to be defined as meaning ∼2∼φ.) You get a different
grammar for QML depending on what version of predicate logic grammar you
begin with. To keep things simple, let’s consider a stripped-down version of
predicate logic: no function symbols, and no definite description operator. But
let’s include the identity sign =.
9.2 Symbolizations in QML

Like any logical extension, moving to QML gives us more power to analyze the
logical structure of natural language sentences. We began with propositional
logic, which let us analyze a certain level of structure, structure in terms of ‘and’,
‘or’, ‘not’, and so on. The move to predicate logic let us analyze quantificational
243
CHAPTER 9. QUANTIFIED MODAL LOGIC 244
structure; the move to modal propositional logic let us analyze modal structure.
Moving to QML lets us do all three at once, as with:
It’s not possible for something to create itself

whose tripartite propositional, predicate, and modal structure is revealed in its
QML symbolization:
∼3∃xC x x
This deeper level of analysis reveals some new logical features. One ex-
ample is the famous distinction between de re and de dicto modal statements.
Consider:
Some rich person might have been poor.

∃x(Rx∧3P x)
It might have been the case that some rich person is poor.
3∃x(Rx∧P x)
The first sentence asserts the existence of someone who is in fact rich, but
who might have been poor. This seems true, in contrast to the absurd second
sentence, which says that the following state of affairs is possible: someone
is both rich and poor. The second sentence is called “de dicto” because the
modality is attributed to a sentence (dictum): the modal operator 3 attaches to
the closed sentence ∃x(Rx∧P x). The first sentence is called “de re” because
the modality is attributed to an object (res): the 3 attaches to a sentence with a
free variable, P x, and thus can be thought of as attributing a modal property,
the property of possibly being poor, to an object u when x is assigned the value u.
Modal propositional logic alone does not reveal this distinction. Given only
a Q to stand for “some rich person is poor”, we can write only 3Q, which
represents only the absurd second sentence. To represent the first sentence
we need to insert the 3 inside the Q, as we can when we further analyze Q as
∃x(Rx∧P x) using predicate logic.
A further example of the de re/de dicto distinction:
Every bachelor is such that he is necessarily male

∀x(B x→2M x)
It is necessary that all bachelors are male
2∀x(B x→M x)
The second, de dicto, sentence makes the true claim that in any possible world,
anyone that is in that world a bachelor is, in that world, male. The first, de
re, sentence makes the false claim that if any object, u, is a bachelor in the
actual world, then that object u is necessarily a bachelor—i.e., the object u is a
bachelor in all possible worlds.
What do the following English sentences mean?
All bachelors are necessarily male

Bachelors must necessarily be male
Surface grammar suggests that they would mean the de re claim that each
bachelor is such that he is necessarily male. But in fact, it’s very natural to
hear these sentences as making the de dicto claim that it’s necessary that all
bachelors are male.
The de re/de dicto distinction also emerges with definite descriptions. This
may be illustrated by using Russell’s theory of descriptions (section 5.3.3). Re-
call how Russell’s method generated two possible symbolizations for sentences
containing definite descriptions and negations. “The striped bear is not dan-
gerous”, for example, can be symbolized as either of the following, depending
on whether the definite description is given wide or narrow scope relative to
the negation operator:
∃x(S x∧B x∧∀y([S y∧B y]→x=y)∧∼D x)

∼∃x(S x∧B x∧∀y([S y∧B y]→x=y)∧D x)
(The second denies the existence of something that is both i) the one and only
striped bear, and ii) dangerous; the first says that there exists something that
is the one and only striped bear, and adds that this bear is non-dangerous.) A
similar phenomenon arises with sentences containing definite descriptions and
modal operators. There are two symbolizations of “The number of the planets
is necessarily odd” (letting “N x” mean that x numbers the planets):
∃x(N x∧∀y(N y→x=y)∧2O x)

2∃x(N x∧∀y(N y→x=y)∧O x)
(Let’s count Pluto as a planet.) The second is de dicto; it says that it’s necessary
that: there is one and only one number of the planets, and that number is odd.
This claim is false, since there could have been eight planets. The second is de
re; it says that (in fact) there is one and only one number of the planets, and
that that number is necessarily odd. That’s true, I suppose: the number nine
(the number that in fact numbers the planets) is necessarily odd.
Natural language sentences containing both definite descriptions and modal
operators are perhaps ambiguous. “The number of the planets is necessarily
odd” is naturally heard as expressing a de re claim; but “The American president
is necessarily an American citizen” can be heard as expressing a de dicto claim.
9.3 A simple semantics for QML

Let’s begin with a very simple semantics, SQML (for “simple QML”). It’s simple
in two ways. First, there is no accessibility relation. 2φ will be said to be true
iff φ is true in all worlds in the model. In effect, each world is accessible from
every other (and hence the underlying propositional modal logic is S5). Second,
it will be a “constant domain” semantics. (We’ll discuss what this means, and
more complex semantical treatments of QML, below.)
Definition of SQML-model: An SQML-model is an ordered triple 〈W , D, I 〉

such that:
· W is a nonempty set (“possible worlds”)

· D is a nonempty set (“domain”)
· I is a function such that: (“interpretation function”)

· if Πn is an n-place predicate then I (Πn ) is a set of n + 1-tuples
〈u1 , . . . , un , w〉, where u1 , . . . , un are members of D, and w ∈ W
Recall that our semantics for moda propositional logic assigned truth values to
sentence letters relative to possible worlds. We have something similar here: we
relativize the interpretation of predicates to possible worlds. The interpretation
of a two-place predicate, R, for example is a set of ordered triples, two members
of which are in the domain, and one member of which is a possible world.
When 〈u1 , u2 , w〉 is in the interpretation of R, that represents R’s applying to
u1 and u2 in possible world w. In a possible worlds setting, this relativization
makes intuitive sense: a predicate can apply to some objects in one possible
world but fail to apply to those same objects in some other possible world.
Notice that the interpretations of constants are not relativized in any way to
possible worlds. The interpretation I assigns simply a member of the domain
to a name. This reflects the common belief that natural language proper
names—which constants are intended to represent—are rigid designators, i.e.,
terms that have the same denotation relative to every possible world (see Kripke
(1972).) We’ll discuss the significance of this feature of our semantics below.
On to the definition of the valuation function for an SQML-model. First,
we keep the definition of a variable assignment from nonmodal predicate logic
(section 4). Our variable assignments therefore assign members of the domain
to variables absolutely, rather than relative to worlds. (This is an appropriate
choice given our choice to assign constants absolute semantic values.) But the
valuation function will now relativize truth values to possible worlds (as well
as to variable assignments). After all, the sentence ‘F a’, if it represents “Ted is
tall”, should vary in truth value from world to world.
Definition of valuation: The valuation function VM ,g , for SQML-model

M (= 〈W , D, I 〉) and variable assignment g , is defined as the function that
assigns either 0 or 1 to each wff relative to each member of W , subject to the
following constraints:
· for any terms α, β, VM , g (α=β, w) = 1 iff [α]M ,g = [β]M ,g

· for any n-place predicate, Π, and any terms α1 , . . . , αn ,
VM ,g (Πα1 . . . αn , w) = 1 iff 〈[α1 ]M ,g , . . . , [αn ]M ,g , w〉 ∈ I (Π)
· for any wffs φ, ψ, and variable, α,
VM ,g (∼φ, w) = 1 iff VM ,g (φ, w) = 0

VM ,g (φ→ψ, w) = 1 iff either VM ,g (φ, w) = 0 or VM ,g (ψ, w) = 1
VM ,g (∀αφ, w) = 1 iff for every u ∈ D, VM ,g α (φ, w) = 1
u
VM ,g (2φ, w) = 1 iff for every v ∈ W , VM ,g (φ, v) = 1
The derived clauses are what you’d expect, including the following one for
3:
VM , g (3φ, w) = 1 iff for some v ∈ W , VM , g (φ, v) = 1
Finally, we have:
· φ is valid in M (= 〈W , D, I 〉) iff for every variable assignment, g , and

every w ∈ W , VM ,g (φ, w) = 1
· φ is SQML-valid (“QML φ”) iff φ is valid in all SQML models.
· Γ SQML-semantically-implies φ (“Γ SQML φ”) iff for every world w in
every SQML model, if every member of Γ is true at w, then so is φ
9.4 Countermodels and validity proofs in SQML

As before, we want to come up with countermodels for invalid formulas, and
validity proofs for valid ones. Validity proofs introduce nothing new.
Example 9.1: Show that SQML 3∃x(x = a∧2F x)→F a:
i) suppose for reductio that (for some model, world r , and variable assign-
ment g ,) Vg (3∃x(x=a∧2F x)→F a, r ) = 0. Thus Vg (3∃x(x=a∧2F x), r ) =
1 and …
ii) …Vg (F a, r ) = 0
iii) From i), for some w ∈ W , Vg (∃x(x=a∧2F x), w) = 1
iv) so for some u ∈ D, Vg x (x=a∧2F x, w) = 1)

u
v) Thus, Vg x (x=a, w) = 1 and …

u
vi) …Vg x (2F x, w) = 1

u
vii) from vi), Vg x (F x, r ) = 1

u
viii) Thus, 〈[x] g x , r 〉 ∈ I (F )—that is, 〈u, r 〉 ∈ I (F )

u
ix) from v, [x] g x = [a] g x

u u
x) By the definition of denotation plus facts about variable assignments,

u = I (a)
xi) By viii) and x), 〈I (a), r 〉 ∈ I (F )

xii) Thus, Vg (F a, r ) = 1. Contradicts line ii)
Notice that in line xii) I inferred that Vg assigned “F a” truth at r . I could have
subscripted ‘V’ with any variable assignment, since the truth condition for the
formula “F a” is the same, regardless of the variable assignment; I picked g
because that’s what I needed to get the contradiction.
As for countermodels, we can use the pictorial method of section 6.3.4,

asterisks and all, with a few changes. First, there’s no need for the arrows
between worlds, since we’ve dropped the accessibility relation, thereby making
every world accessible to every other. Secondly, we have predicates and names
for atomics instead of sentence letters, so how to account for this? Let’s look at
an example: finding a countermodel for the formula (3F a∧3Ga)→3(F a∧Ga).
We begin as follows:
∗
1 1 1 0 0
r
(3F a∧3Ga)→3(F a∧Ga)
∗ ∗
The understars make us create two new worlds:
∗
1 1 1 0 0
r
∗ ∗
1 1
a b
Fa Ga
We must then discharge the overstar from the false diamond in each world
(since every world is accessible to every other world in our models):
∗
1 1 1 0 0 0 0
r
∗ ∗ †
1 0 0 1 0 0
a b
Fa F a∧Ga Ga F a∧Ga
(I had to make either F a or Ga false in r—I chose F a arbitrarily.) Now, we’ve
indicated the truth-values that we want the atomics to have. How do we make
the atomics have the TVs we want in the picture?
We do this by introducing a domain for the model, and stipulating what the
names refer to and what objects are in the extensions of the predicates. Let’s
use letters like ‘u’ and ‘v’ as the members of the domain in our models. Now, if
we let the name ‘a’ refer to (the letter) u, and let the extension of F in world r
be {} (the empty set), then the truth value of ‘F a’ in world r will be 0 (false),
since the denotation of a isn’t in the extension of F at world r. Likewise, we
need to put u in the extension of F (but not in the extension of G) in world
a, and put u in the extension of G ((but not in the extension of F ) in world b.
This all may be indicated on the diagram as follows:
a: u
∗
1 1 1 0 0 0 0
r
∗ ∗ †
F :{}
1 0 0 1 0 0
Fa F a∧Ga Ga F a∧Ga
a b
F : {u} G : {} F : {} G : {u}
Within each world I’ve included a specification of the extension of each predi-
cate. But the specification of the referent of the name ‘a’ does not go within
any world; it was rather indicated (in boldface) at the top of model. This is
because names, unlike predicates, get assigned semantic values absolutely in a
model, not relative to worlds.
Time for the official model:
W = {r, a, b}
D = {u}
I (a) = u
I (F ) = {〈u, a〉}
I (G) = {〈u, b〉}
What about formulas with quantifiers? A countermodel for 2∃xF x→∃x2F x

begins as follows:
∗ +
1 1 0 0
r
2 ∃ xF x→ ∃ x2F x
+
The overstar above the 2 in the antecedent must be discharged in r itself, since,
remember, every world sees every world in these models. That gives us a true
existential. Now, a true existential is a bit like a true 3—the true ∃xF x means
that there must be some object u from the domain that’s in the extension of F
in r. I’ll put a + under true ∃s and false ∀s, to indicate a commitment to some
instance of some sort or other. Analogously, I’ll indicate a commitment to all
instances of a given type (which would arise from a true ∀ or a false ∃) with a +
above the connective in question.
OK, how do we make ∃xF x true in r? By making “F x” true for some
value of x. Let’s put the letter u in the domain, and make “F x” true when u is
assigned to x. We’ll indicate this by putting a 1 overtop of “F ux ” in the diagram.
Now, “F ux ” isn’t a formula of our language—what it indicates is that “F x” is to
be true when u is assigned to x. And to make this come true, we treat it as an
atomic—we put u in the extension of F at r:
∗ +
1 1 0 0 1
2 ∃ xF x→ ∃ x2F x F ux
+
r
F : {u}
Good. Now we’ve got to attend to the overplus, the + sign overtop the false
∃x2F x. Since it’s a false ∃, we’ve got to make 2F x false for every object in the
domain (otherwise—if there were something in the domain for which 2F x was
true—∃x2F x would be true after all). So far, we’ve got only one object in our
domain, u, so we’ve got to make 2F x false, when u is assigned to the variable
‘x’. We’ll indicate this on the diagram by putting a 0 overtop of “2F ux ”:
∗ +
1 1 0 0 1 0
2 ∃ xF x→ ∃ x2F x F u
x
2F ux
+
r
∗
F : {u}
Ok, now we have an understar, which means we should add a new world to
our model. When doing so, we’ll need to discharge the overstar from the
antecedent. We get:
∗ +
1 1 0 0 1 0
2 ∃ xF x→ ∃ x2F x F ux 2F ux
+
r
∗
F : {u}
0 1 1
F u
x
∃ xF x F v
x
a +
F : {v}
This move requires some explanation. Why the v? Well, I was required to
make F x false, with u assigned to x. Well, that means keeping u out of the
extension of F at a. Easy enough, right? Just make F ’s extension {}? Well,
no—because of the true 2 in r, I’ve got to make ∃xF x true in a. But that means
that something’s got to be in F ’s extension in a! It can’t be u, so I’ll add a new
object, v, to the domain, and put it in F ’s extension in a.
But adding v to the domain of the model adds a complication. We had
an overplus in r—over the false ∃. That meant that, in r, for every member of
the domain, 2F x is false. So, 2F x is false in r when v is assigned to x. That
creates another understar, requiring the creation of a new world. The model
then looks as follows:
∗ +
1 1 0 0 1 0 0
2 ∃ xF x→ ∃ x2F x F u
x
2F u
x
2F v
x
+
r
∗ ∗
F : {u}
0 1 1 0 1 1
F u
x
∃ xF x F v
x
F v
x
∃ xF x F ux
a + b +
F : {v} F : {u}
(Notice that we needn’t have made another world b—we could simply have
discharged the understar on r.)
Ok, here’s the official model:
W = {r, a, b}
D = {u, v}
I (F ) = {〈u, r〉, 〈u, b〉, 〈v, a〉}
Exercise 9.1 For each formula, give a validity proof if the wff is
SQML-valid, and a countermodel if it is invalid.
a) 3∀xF x→∃x3F x
b) ∃x3Rax→32∃x∃yRxy
c) ∃x(N x∧∀y(N y→y=x)∧2O x) →

2∃x(N x∧∀y(N y→y=x)∧O x)
9.5 Philosophical questions about SQML

Our semantics for quantified modal logic faces philosophical challenges. In
each case we will be able to locate a particular feature of our semantics that
gives rise to the alleged problem. In response, one can stick with the simple
semantics and give it a philosophical defense, or one can revise the semantics.
9.5.1 The necessity of identity

Let’s try to come up with a countermodel for the following formula:
∀x∀y(x=y→2(x=y))
When we try to make the formula false by putting a 0 over the initial ∀, we get
an under-plus. So we’ve got to make the inside part, ∀y(x=y→2x=y), false
for some value of x. We do this by putting some object u in the domain, and
letting that be the value of x for which ∀y(x=y→2x=y) is false. We get:
0 0
r ∀ x∀y(x=y→2x=y) ∀y( ux =y→2( ux =y))
+ +
Now we need to do the same thing for our new false universal: ∀y(x=y→2x=y).
For some value of y, the inside conditional has to be false. But that means that
the antecedent must be true. So the value for y has to be u again. We get:
0 0 1 0 0
r ∀ x∀y(x=y→2x=y) ∀y( ux =y→2( ux =y)) = y →2( ux = uy )

u u
x
+ + ∗
The understar now requires creation of a world in which x=y is false, when
both x and y are assigned u. But there cannot be any such world! An identity
sentence is true (at any world) if the denotations of the terms are identical. Our
attempt to find a countermodel has failed; we must do a validity proof. Consider
any SQML model 〈W , D, I 〉, any r ∈ W , and any variable assignment g ; we’ll
show that Vg (∀x∀y(x=y→2x=y), r ) = 1:
i) suppose for reductio that Vg (∀x∀y(x=y→2x=y), r ) = 0.
ii) Then, for some u ∈ D, Vg x (∀y(x=y→2x=y), r ) = 0

u
iii) So, for some v ∈ D, Vg xy (x=y→2(x=y), r ) = 0.

uv
iv) Thus, Vg xy (x=y, r ) = 1, and …

uv
v) …Vg xy (2(x=y), r ) = 0
uv
vi) from iv) [x] g xy = [y] g xy

uv uv
vii) From v), at some world, w, Vg xy (x=y, w) = 0

uv
viii) And so, [x] g xy 6= [y] g xy . Contradicts vi).

uv uv
Notice at the end how the particular world at which the identity sentence was
false didn’t matter. The truth condition for an identity sentence is simply that
the terms denote the same thing; it doesn’t matter what world this is evaluated
relative to.1
1
A note about variables. In validity proofs, I’m using italicized ‘u’ and ‘v’ as variables to
range over objects in the domain of the model I’m considering. So, a sentence like ‘u = v’
might be true, just as the sentence ‘x=y’ of our object language can be true. But when I’m
doing countermodels, I’m using the roman letters ‘u’ and ‘v’ as themselves being members of
the domain, not as variables ranging over members of the domain. Since the letters ‘u’ and
‘v’ are different letters, they are different members of the domain. Thus, in a countermodel
with letters in the domain, if the denotation of a name ‘a’ is the letter ‘u’, and the denotation
We can think of ∀x∀y(x=y→2(x=y)) as expressing “the necessity of iden-

tity”: it says that whenever identity holds between objects, it necessarily holds.
The necessity of identity is philosophically controversial. On the one hand
it can seem obviously correct. “x=y” says that x and y are one and the same
thing. Now, if there were a world in which x was different from y, since x and
y are the same thing, this would have to be a world in which x was different
from x. How could that be? On the other hand, it was a great discovery that
Hesperus = Phosphorus. Surely, it could have turned out the other way—surely,
Hesperus might not have turned out identical to Phosphorus! But isn’t this
a counterexample to this formula? For a discussion of this example, see Saul
Kripke’s book Naming and Necessity.
It’s worth noting why the necessity of identity turns out valid, given our
semantics. It turns out valid because of the way we defined variable assignments:
our variable assignments assign members of the domain to variables absolutely,
rather than relative to worlds. (Similarly: since the interpretation function I ,
according to our definition above, assigns referents to names absolutely, rather
than relative to worlds, the formula a=b →2 a=b turns out valid.) One could,
instead, define variable assignments as functions that assign members of the
domain to variables relative to worlds. Given appropriate adjustments to the
definition of the valuation function, this would have the effect of invalidating
the necessity of identity.2 (Similarly, one could make I assign denotations to
names relative to worlds, thus invaliding a=b →2 a=b .)
9.5.2 The necessity of existence

Another (in)famous valid formula of SQML is the “Barcan Formula” (named
after Ruth Barcan Marcus)
∀x2F x→2∀xF x
(Call the schema ∀α2φ→2∀αφ the “Barcan schema”.) If we try to find a

countermodel for this formula we get to the following stage:
of the name ‘b ’ is the letter ‘v’, then the sentence ‘a=b ’ has got to be false, since ‘u’6=‘v’. If
I were using ‘u’ and ‘v’ as variables ranging over members of the domain, then the sentence
‘u = v’ might be true! This just goes to show that it’s important to distinguish between the
sentence u = v and the sentence ‘u’ = ‘v’. The first could be true, depending on what ‘u’ and
‘v’ currently refer to, but the second one is just plain false, since ‘u’ and ‘v’ are different letters.
2
See Gibbard (1975).
0 0
+
1 0 0
∀ xF x F ux
r a +
∀ x2F x→2∀xF x
∗
F : {}
When you have a choice between discharging over-things and under-things,
whether plusses or stars, always do the under things first. In this case, this
means discharging the understar and ignoring the over-plus for the moment.
So, discharging the understar gave us world a, in which we made a universal
false. This gave an underplus, and forced us to make an instance false. So I put
object u in our domain, and keep it out of the extension of F in a. This makes
F x false in a, when x is assigned u.
But now, I need to discharge the overplus in r. I must make 2F x true for
every member of the domain, including u, which is now in the domain. But
then this requires F x to be true, when u is assigned to x, in a:
+ ∗ 0 0 1
1 0 0 1 1 ∀ xF x F ux F ux
∀ x2F x→2∀xF x 2F ux +
r a
∗
F : {u} F : {?}
So, we fail to get a model. Time for a validity proof; let’s show that every
instance of the Barcan Schema is valid:
i) suppose for reductio that Vg (∀α2φ→2∀αφ, r ) = 0. Then Vg (∀α2φ, r ) =

1 and …
ii) …Vg (2∀αφ, r ) = 0.
iii) from ii), for some w, Vg (∀αφ, w) = 0
iv) so, for some u in the domain, Vgu/α (φ, w) = 0
v) from i), for every member of the domain, and so for u in particular,
Vg α (2φ, r ) = 1.
u
vi) thus, for every world, and so for w in particular, Vg α (φ, w) = 1. Contra-
u
dicts iv).
The validity of the Barcan formula in our semantics is infamous because the
Barcan formula seems, intuitively, to be invalid. To see why, we need to think a
bit about the intuitive significance of the relative order of quantifiers and modal
operators. Consider the difference between the following two sentences:
3∃xF x
∃x3F x
In general, a sentence of the form 3φ says that it’s possible for the component
sentence, φ, to be true. So the first of our two sentences, 3∃xF x, says that
it’s possible for “∃xF x” to be true. That is: it’s possible for there to exist an F .
What about the second sentence? In general, a sentence that begins without
a modal operator in front makes a statement about the actual world. Thus,
a statement that begins with “∃x . . . ” is saying that there exists, in the actual
world, an object x, such that…. Our second statement, then, says that there
actually exists an object, x, that is possibly F . It matters, therefore, whether
the ∃ comes after or before the 3. If the ∃ comes first, then the statement is
saying that there actually exists a certain sort of object (namely, an object that
could have been a certain way.) But if it comes second, after the 3, then the
statement is merely saying that there could have existed a certain sort of object.
There is a similar contrast with the following two statements:
2∀xF x
∀x2F x
The first says that it’s necessary that: everything is F . That is, in every possible
world, every object that exists in that world is F in that world. The objects
ranged over by the ∀, so to speak, are drawn from the worlds the 2 introduces,
because the ∀ occurs inside the scope of the 2. The second statement, by
contrast, says that: every actual object is necessarily F . That is, every object that
exists in the actual world is F in every possible world. The second statement
concerns just actually existing objects because the ∀ occurs in the front of the
formula, not inside the scope of the 2.
With all this in mind, return to the Barcan formula, ∀x2F x→2∀xF x. It
says:
“If every actually existing thing is F in every possible

world, then in every world, every object in that world is
F in that world”
Now we can see why this claim is questionable. Even if every actual thing is
necessarily F , there could still be worlds containing non-F things, so long as
those non-F things don’t exist in the actual world. Suppose, for instance, that
every object in the actual world is necessarily a material object. Then, letting
F stand for “is a material object”, ∀x2F x is true. Nevertheless, 2∀xF x seems
false—it would presumably be possible for there to exist an immaterial object:
a ghost, say. Possible worlds containing ghosts would simply need to contain
objects that do not exist in the actual world (since all the objects in the actual
world are necessarily material.)
This objection to the validity of the Barcan formula is obviously based on
the idea that what objects exist can vary from possible world to possible world.
But this sort of variation is not represented in the SQML definition of a model.
Each such model contains a single domain, D, rather than different domains
for different possible worlds. The truth condition we specified for a quantified
sentence ∀αφ, at a world w, was simply that φ is true at w of every member of
D—the quantifier ranges over the same domain, regardless of which possible
world is being described. That is why the Barcan formula turns out valid under
our definition.
This feature of SQML models is problematic for an even more direct reason:
the sentence ∀x2∃y y = x, i.e., “everything necessarily exists”, turns out valid!:
i) Suppose for reductio that Vg (∀x2∃y y=x, w) = 0.
ii) Then Vg x (2∃y y=x, w) = 0, for some u ∈ D

u
iii) So Vg x (∃y y=x, w 0 ) = 0, for some w 0 ∈ W

u
iv) So Vg xy (y=x, w 0 ) = 0, for every u 0 ∈ D

u u0
v) So, since u ∈ D, we have Vg xy (y=x, w 0 ) = 0. But that can’t be, given the
uu
clause for ‘=’ in the definition of the valuation function.
It’s clear that this formula turns out valid for the same reason that the Barcan
formula turns out valid: SQML models have a single domain common to each
possible world.
The Barcan schema is just one of a number of interesting schemas concern-

ing how quantifiers and modal operators interact (for each schema I also list an
equivalent schema with 3 in place of 2):
∀α2φ→2∀αφ 3∃αφ→∃α3φ (Barcan)

2∀αφ→∀α2φ ∃α3φ→3∃αφ (converse Barcan)
∃α2φ→2∃αφ 3∀αφ→∀α3φ
2∃αφ→∃α2φ ∀α3φ→3∀αφ
We have already discussed the Barcan schema. The third schema raises no philo-
sophical problems for SQML, since, quite properly, it has instances that turn out
invalid: as we saw above, there are SQML models in which 2∃xF x→∃x2F x
is false. Let’s look at the other two schemas.
First, the converse Barcan schema. Like the Barcan schema, each of its
instances is valid given the SQML semantics (I’ll leave this to the reader to
demonstrate), and like the Barcan schema, this verdict faces a philosophical
challenge. The antecedent says that in every world, everything that exists in
that world is φ. Existents are thus always φ. It might still be that some object
isn’t necessarily φ: perhaps some object that is φ in every world in which it
exists, fails to be φ in worlds in which it doesn’t exist. This talk of an object
being φ in a world in which it doesn’t exist may seem strange, but consider
the following instance of the converse Barcan schema, substituting “∃y y=x”
(think: “x exists”) for φ:
2∀x∃y y=x→∀x2∃y y=x
This formula seems to be false. Its antecedent is clearly true; but its consequent
says that every object in the actual world exists necessarily, and hence seems
intuitively to be false.
Each instance of the fourth schema, ∃α2φ→2∃αφ, is also validated by
the SQML semantics (again, an exercise for the reader); and again, this is
philosophically questionable. Let’s suppose that physical objects are necessarily
physical. Then, ∃x2P x seems true, letting P mean ‘is physical’. But 2∃xP x
seems false—it seems possible that there are no physical objects. This coun-
terexample requires that there be worlds with fewer objects than those that
actually exist, whereas the counterexample to the Barcan formula involved the
possibility that there be more objects than those that actually exist.
9.5.3 Necessary existence defended

There are various ways to respond to the challenge of the previous section.
From a logical point of view, the simplest is to stick to one’s guns and defend
the SQML semantics. SQML-models accurately model the modal facts. The
Barcan formula, the converse Barcan formula, the fourth schema, and the state-
ment that everything necessarily exists are all logical truths; the philosophical
objections are mistaken. Contrary to appearances it is not contingent what
exists. Each possible world has exactly the same stock of individuals. Call this
the doctrine of Constancy.
One could uphold Constancy either by taking an narrow view of what is
possible, or by taking a broad view of what exists. On the former alternative,
one would claim that it is just not possible for there to be any ghosts, and that
it is just not possible in any sense for an actual object to have failed to exist. On
the latter alternative, which I’ll be discussing for the rest of this section, one
accepts the possibility of ghosts, dragons, and so on, but claims that possible
ghosts and dragons exist in the actual world.
Think of the objects in D as being all of the possible objects. In addition to
normal things—what one would normally think of as the actually existing entities:
people, tables and chairs, planets and electrons, and so on—our defender of
Constancy claims that there also exist objects that, in other possible worlds, are
ghosts, golden mountains, talking donkeys, and so forth; and these are included
in D as well. Call these further objects “merely possible things” (but don’t be
misled by this label; the claim is that merely possible things actually exist.) The
formula “∀xF x” means that every possible object is F in the actual world. It’s
not enough for the normal things to be F , for the normal things are not all
of the things that there are. There are also all the merely possible things, and
each of these must be F as well (must be F here in the actual world, that is), in
order for ∀xF x to be true. Hence, the objection to the Barcan formula from
the previous section fails. That objection assumed that ∀x2F x, the antecedent
of (an instance of) the Barcan formula, was true, when F symbolizes “is a
material object”. But this neglects the merely possible things. It’s true that all
the normal objects are necessarily material objects, but there are some further
things—merely possible things—that are not necessarily material objects.
Further: in ordinary language, when we say “Everything” or “something”,
we typically don’t mean to be talking about all possible objects; we’re typically
talking about just the normal things. Otherwise we would be speaking falsely
when we say, for example, “everything has mass”: merely possible unicorns
presumably have no mass (nor any spatial location, nor any other physical
feature.) Ordinary quantification is restricted to normal things. So if we want
to translate an ordinary claim into the language of QML, we must introduce
a predicate for the normal things, “N ”, and use it to restrict quantifiers. But
now, consider the following ordinary English statement:
If everything is necessarily a material object, then neces-

sarily: everything is a material object
If we mindlessly translate this into the language of QML, we would get
∀x2F x→2∀xF x—an instance of the Barcan schema. But since in every-
day usage, quantifiers are restricted to normal things, the thought in the mind
of an ordinary speaker who utters this sentence is more likely the following:
∀x(N x→2F x)→2∀x(N x→F x)
which says:
If every normal thing is necessarily a material object,

then necessarily: every normal thing is a material object.
And this formula is not an instance of the Barcan schema, nor is it valid, as may
be shown by the following countermodel:
+
1 0 0 0 1 0
∀ x(N x→2F x)→2∀x(N x→F x) N u
x
→2F ux
r
∗ †
N : {}
0 1 0 0
∀ x(N x→F x) N u
x
→F ux
a +
N : {u} F : {}
So in a sense, the ordinary intuitions that were alleged to undermine the Barcan
schema are in fact consistent with Constancy.
The defender of Constancy can defend the converse Barcan schema and the
fourth schema in similar fashion. The objection to the converse Barcan schema
assumed the falsity of ∀x2∃y y=x. “Sheer prejudice!”, according to the friend
of constancy. “And recall further that an ordinary utterance of ‘Everything exists
necessarily’ expresses, not ∀x2∃y y=x, but rather ∀x(N x→2∃y(N y∧y=x)),
(N for ‘normal’), the falsity of which is is perfectly compatible with Constancy.
It’s possible to fail to be normal; all that’s impossible is to utterly fail to exist.
Likewise for the fourth schema.”
This defense of SQML is hard to take. Let “G” stand for a kind of object
that, in fact, has no members, but which could have had members. Perhaps ghost
is such a kind. ∀x2∼Gx→2∀x∼Gx is an instance of the Barcan schema, and
so true according to the defender of Constancy. Since there could have existed
ghosts, the consequent of this conditional is false. Therefore, its antecedent
∀x2∼Gx must be false. That is, there exists something that could have been a
ghost. But this is a very surprising result. The alleged possible ghost couldn’t
be any material object, presumably, assuming it would be impossible for any
material object to be a ghost. The defender of Constancy, then, is committed to
the existence of objects which we wouldn’t otherwise have dreamed of accepting:
things that could have been ghosts, things that been dragons, things that could
have been gods, and so on.
The defender of Constancy might try to defend this conclusion by remind-
ing us that these “possible-ghosts”, “possible-dragons”, and so on, are not
normal objects. They aren’t in space and time, presumably, which explains why
no one has ever seen, heard, felt, or smelled one. He might even say that they
are non-actual, or even that they do not exist (though they are). We are quite cor-
rect, she might say, to scoff at the idea that some normal/actual/existing objects
are capable of being ghosts; but what’s the big deal about saying that some non-
normal/non-actual/non-existing objects have these capabilities? This move,
too, will be considered philosophically suspect by many. Many philosophers
regard the idea that there are some non-existent things, or some non-actual
things, as being anywhere from obviously false to conceptually incoherent, or
subversive, or worse.3 And how does it help to point out that the objects aren’t
normal? The postulation of non-normal objects—objects above and beyond
the objects that the rest of us believe in—was exactly what I was claiming is
philosophically suspect!
On the other hand, Constancy’s defenders can point to certain powerful
3
See Quine (1948); Lycan (1979).
arguments in its favor. Here’s a quick sketch of one such argument. First, the
following seems to be a logical truth:
Ted = Ted
But it follows from this that:
∃y y = Ted
This latter formula, too, is therefore a logical truth. But if φ is a logical truth
then so is 2φ (recall the rule of necessitation from chapter 6). So we may infer
that the following is a logical truth:
2∃y y = Ted
Next, notice that nothing in the argument for 2∃y y = Ted depended on any
special features of me. We may therefore conclude that the reasoning holds
good for every object; and so ∀x2∃y y = x is indeed a logical truth. Since,
therefore, every object exists necessarily, it should come as no surprise that
there are things that might have been ghosts, dragons, and so on—for if there
had been a ghost, it would have necessarily existed, and thus must actually exist.
This and other related arguments have apparently wild conclusions, but they
cannot be lightly dismissed, for it is no mean feat to say exactly where they go
wrong (if they go wrong at all!).4
9.6 Variable domains

We now consider a way of dealing with the problems discussed in section 9.5.2
above that does not require embracing Constancy.
SQML models contain a single domain, D, over which the quantifiers range
in each possible world. Since it was this feature that led to the problems of
section 9.5.2, let’s introduce a new semantics that instead provides different
domains for different possible worlds. And let’s also reinstate the accessibility
relation, for reasons to be made clear below:5 The new semantics is called
VDQML (“variable-domains quantified modal logic”):
4
On this topic see Prior (1967, 149-151); Plantinga (1983); Fine (1985); Linsky and Zalta
(1994, 1996); Williamson (1998, 2002).
5
More care than I take is needed in converting the earlier definition of validity if one is
worried about the validity of formulas with free variables. See ?, p. 275.
Definition of VDQML-model: A VDQML-model is a 5-tuple 〈W , R, D, Q, I 〉

such that:
· W is a nonempty set (“possible worlds”)

· R is a binary relation on W (“accessibility relation”)
· D is a set (“super-domain”)
· Q is a function that assigns to any w ∈ W a non-empty subset of D. Let
6
us refer to Q(w) as “Dw ”. Think of Dw as w’s “sub-domain”—the set of

objects that exist at w.
· I is a function such that: (“interpretation function”)

· if Π is an n-place predicate then I (Π) is a set of ordered n +1-tuples
〈u1 , . . . , un , w〉, where u1 , . . . , un are members of D, and w ∈ W .
Definition of valuation: The valuation function VM ,g , for VDQML-model

M (= 〈W , R, D, Q, I 〉) and variable assignment g , is defined as the function
that assigns either 0 or 1 to each wff relative to each member of W , subject to
the following constraints:
· for any terms α and β, VM , g (α=β, w) = 1 iff [α]M ,g = [β]M ,g

· for any n-place predicate, Π, and any terms α1 , . . . , αn ,
VM ,g (Πα1 . . . αn , w) = 1 iff 〈[α1 ]M ,g , . . . , [αn ]M ,g , w〉 ∈ I (Π)
· for any wffs φ and ψ, and variable, α,
VM ,g (∼φ, w) = 1 iff VM ,g (φ, w) = 0

VM , g (φ→ψ, w) = 1 iff either VM ,g (φ, w) = 0 or VM ,g (ψ, w) = 1
VM ,g (∀αφ, w) = 1 iff for each u ∈ Dw , VM ,g α (φ, w) = 1
u
VM ,g (2φ, w) = 1 iff for each v ∈ W , if R wv then VM ,g (φ, v) = 1

6
One could drop this assumption. But if subdomains can be empty then 2∃x(F x→F x) will
be invalid. Since ∃x(F x→F x) is valid given the chapter 4 semantics for non-modal predicate
logic, we would have the odd result of a logical truth whose necessitation isn’t a logical truth.
One could modify the chapter 4 semantics by allowing predicate logic models with empty
domains, thus invalidating ∃x(F x→F x). This approach is known as free logic.
The definitions of denotation, validity and semantic consequence remain un-

changed. The obvious derived clauses for ∃ and 3 are as follows:
VM ,g (∃αφ, w) = 1 iff for some u ∈ Dw , VM , g α (φ, w) = 1

u
VM ,g (3φ, w) = 1 iff for some v ∈ W , R wv and VM , g (φ, v) = 1
Thus, we have introduced introduced subdomains. We still have D, a set

that contains all of the possible individuals. But for each possible world w,
we introduce a subset of the domain, Dw , to be the domain for w. When
evaluating a quantified sentence at a world w, the quantifier ranges only over
Dw . Notice that we also reinstated the accessibility relation. This isn’t necessary
for introducing subdomains; I did this in order to be able to make a certain
point about subdomains in section 9.6.2.
Note that if M is a SQML model, then we can construct a corresponding
VDQML model with the same set of worlds, (super-) domain, and interpre-
tation function, in which every world is accessible from every other, and in
which Q is a constant function assigning the whole super-domain to each world.
It is intuitively clear that the same sentences are true in this corresponding
model as are true in M . Hence, whenever a sentence is SQML-invalid, it is
VDQML-invalid. (The converse of course is not true.)
9.6.1 Countermodels to the Barcan and related formulas in

VDQML
What is the effect of this new truth definition on the Barcan formula and related
formulas? All of these formulas come out invalid:
∀x2F x→2∀xF x 3∃xF x→∃x3F x (Barcan)

2∀xF x→∀x2F x ∃x3F x→3∃xF x (converse Barcan)
∃x2F x→2∃xF x 3∀xF x→∀x3F x
2∃xF x→∃x2F x ∀x3F x→3∀xF x
The third one on the list was invalid before, and so is still invalid now. As for
the Barcan formula, here is a countermodel:
r Dr : {u} F : {u}
0

a Da : {u, v} F : {u}
0
Official model:
W = {r, a}
R = {〈r, r〉, 〈r, a〉, 〈a, a〉}
D = {u, v}
Dr = {u}
Da = {u, v}
I (F ) = {〈u, r〉, 〈u, a〉}
I leave the demonstrations of the invalidity of the remaining formulas as exer-

cises.
Exercise 9.2 Does the move to variable domain semantics change

whether any of the formulas in exercise set 9.1 are valid? Justify
your answers.
Exercise 9.3 Demonstrate the VDQML-invalidity of the follow-

ing formulas
a) 2∀xF x→∀x2F x
b) ∃x2F x→2∃xF x
c) ∀x2∃y y=x
9.6.2 Expanding, shrinking domains

There are several comments worth making about VDQML-models. First,
note that if we made certain restrictions on variable-domains models, then
the countermodels of the previous section would no longer be legal models.
For example, the first example, the counterexample to the Barcan formula,
required a model in which the domain expanded; world a was accessible from
world r, and had a larger domain. But suppose we made the decreasing domains
requirement:
if R wv, then Dv ⊆ Dw
The counterexample would then go away. Indeed, every instance of the Barcan
schema would then become VDSQML-valid, which may be proved as follows:
i) suppose for reductio that Vg (∀α2φ→2∀αφ, w) = 0. Then Vg (∀α2φ, w) =

1 and…
ii) …Vg (2∀αφ, w) = 0
iii) by ii), for some v, R wv and Vg (∀αφ, v) = 0
iv) and so, for some u ∈ Dv , Vg α (φ, v) = 0

u
v) given decreasing domains, Dv ⊆ Dw , and so u ∈ Dw
vi) by i), for every object in Dw , and so for u in particular, Vg α (2φ, w) = 1

u
vii) so, Vg α (φ, v) = 1. Contradicts iv)

u
Similarly, notice that the counterexamples to ∃x2F x→2∃xF x and the

converse Barcan formula assumed that domains can shrink. But the following
increasing domains requirement validates these formulas (as may be easily shown):
if R wv then Dw ⊆ Dv
Even after imposing the increasing domains requirement, the Barcan for-
mula remains VDQML-invalid; and after imposing the decreasing domains
requirement, the converse Barcan formula and also ∃x2F x→2∃xF x remain
VDQML-invalid (the original countermodels for these formulas establish this.)
However, in systems in which the accessibility relation is symmetric, this col-
lapses: imposing either of these requirements results in imposing the other.
That is, in B or S5, imposing either the increasing or the decreasing domains
requirement results in imposing both, and hence results in all three formulas
being validated.
Exercise 9.4 Show that every instance of each of the following

schemas is valid given the increasing domains requirement.
a) 2∀αφ→∀α2φ
b) ∃α2φ→2∃αφ
9.6.3 Strong and weak necessity

In order for 2φ to be true at a world, the VDQML semantics requires that φ
be true at every accessible world. It might be thought that this requirement is
too strong. In order for 2F a, say, to be true, our definition requires F a to be
true in all possible worlds. But what if a fails to exist in some worlds? In order
for “Necessarily, I am human” to be true, must I be human in every possible
world? Isn’t it enough for me to be human in all the worlds in which I exist?
This argument goes by a little too quickly. The main worry of its proponent
is that our semantics requires a to exist necessarily, in order for 2F a to come
out true. But our semantics doesn’t require this. It does require F a to be
true in every world, in order for 2F a to be true; but it does not require a
to exist in every world in which F a is true. The clause in the definition of a
VDQML-model for the interpretation of predicates was this:
· if Π is an n-place predicate then I (Π) is a set of ordered n + 1-tuples

〈u1 , . . . , un , w〉, where u1 , . . . , un are members of D, and w ∈ W .
This allows I (F ) to contain pairs 〈u, w〉, where u is not a member of Dw . So

one could say that 2F a is consistent with a’s failing to necessarily exist; it’s just
that a has to be F even in worlds where it doesn’t exist.
I doubt this really addresses the worry, since it looks like bad metaphysics
to say that a person could be human at a world where he doesn’t exist. One
could hard-wire a prohibition of this sort of bad metaphysics into VDQML
semantics, by replacing the old clause with a new one:
· if Π is an n-place predicate then I (Π) is a set of ordered n+1-tuples

〈u1 , . . . , un , w〉, where u1 , . . . , un are members of Dw , and w ∈ W .
thus barring objects from having properties at worlds where they don’t ex-
ist. But some would argue that this goes too far. The new clause validates
∀x2(F x→∃y y=x). “An object must exist in order to be F ”—sounds clearly
true if F stands for ‘is human’, but what if F stands for ‘is famous’? If Baconians
had been right and there had been no such person as Shakespeare, perhaps
Shakespeare might still have been famous.
The issues here are complex.7 But whether or not we should adopt the new
clause, it looks as though there are some existence-entailing English predicates
π: predicates π such that nothing can be a π without existing. ‘Is human’ seems
to be such a predicate. So we’re back to our original worry about VDQML-
semantics: its truth condition for 2φ requires truth of φ at all worlds, which is
allegedly too strong, at least when φ is a sentence like πa, where π is existence-
entailing.
One could modify the clause for the 2 in the definition of the valuation
function, so that in order for 2F a to be true, a only needs to be F in worlds in
which it exists:
VM ,g (2φ, w) = 1 iff for each v ∈ W , if R wv, and if [α]M ,g ∈ Dw for each
name or free variable α occurring in φ, then VM ,g (φ, v) = 1
(“Free variable” here means a variable not bound to any quantifier in φ.) This
would indeed have the result that 2F a gets to be true provided a is F in every
world in which it exists. But be careful what you wish for. Along with this result
comes the following: even if a doesn’t necessarily exist, the sentence 2∃x x=a
comes out true. For according to the new clause, in order for 2∃x x=a to be
true, it must merely be the case that ∃x x=a is true in every world in which a
exists, and of course this is indeed the case.
If 2∃x x=a comes out true even if a doesn’t necessarily exist, then 2∃x x=a
doesn’t say that a necessarily exists. Indeed, it doesn’t look like we have any way
of saying that a necessarily exists, using the language of QML, if the 2 has the
meaning provided for it by the new clause.
A notion of necessity according to which “Necessarily φ” requires truth in
all possible worlds is sometimes called a notion of strong necessity. In contrast,
a notion of weak necessity is one according to which “Necessarily φ” requires
merely that φ be true in all worlds in which objects named within φ exist. The
new clause for the 2 corresponds to weak necessity, whereas our original clause
corresponds to strong necessity.
As we saw, if the 2 expresses weak necessity, then one cannot even express
the idea that a thing necessarily exists. That’s because one needs strong necessity
7
The question is that of so-called “serious actualism” (Plantinga, 1983).
to say that a thing necessarily exists: in order to necessarily exist, you need to
exist at all worlds, not just at all worlds at which you exist! So this is a serious
deficiency of having the 2 of QML express weak necessity. But if we allow the
2 to express strong necessity instead, there is no corresponding deficiency, for
one can still express weak necessity using the strong 2 and other connectives.
For example, to say that a is weakly necessarily F (that is, that a is F in every
world in which it exists), one can say: 2(∃x x=a→F a).
So it would seem that we should stick with our original truth condition for
the 2, and live with the fact that statements like 2F a turn out false if a fails
to be F at worlds in which it doesn’t exist. Those who think that “Necessarily,
Ted is human” is true despite Ted’s possible nonexistence can always translate
this natural language sentence into the language of QML as 2(∃x x=a→F a)
(which requires a to be F only at worlds at which it exists) rather than as 2F a
(which requires a to be F at all worlds).
Chapter 10
Two-dimensional modal logic
n this chapter we consider an extension to modal logic with considerable

I philosophical interest.
10.1 Actuality
The word ‘actually’, in one of its senses anyway, can be thought of as a one-place
sentence operator: “Actually, φ.”
‘Actually’ might at first seem redundant. “Actually, snow is white” basically
amounts to: “snow is white”. But the actuality operator interacts with modal
operators in interesting ways. The following two sentences, for example, clearly
have different meanings:
Necessarily, if snow is white then snow is white

Necessarily, if snow is white then snow is actually white
The first sentence expresses the triviality that snow is white in any possible
world in which snow is white. But the second sentence makes the nontrivial
statement that if snow is white in any world, then snow is white in the actual
world.
So, ‘actually’ is nonredundant, and consequently, worth thinking about.
Let’s add a symbol to modal logic for it. “@φ” will symbolize “Actually, φ”.
We can now symbolize the pair of sentences above as 2(S→S) and 2(S→@S),
272
CHAPTER 10. TWO-DIMENSIONAL MODAL LOGIC 273
respectively. For some further examples of sentences we can symbolize using

‘actually’, consider:1
It might have been that everyone who is actually rich is

poor
3∀x(@Rx→P x)
There could have existed something that does not actu-
ally exist
3∃x@∼∃y y=x
10.1.1 Kripke models with designated worlds

For the purposes of this chapter, the logic of iterated boxes and diamonds isn’t
relevant, so let’s simplify things by dropping the accessibility relation from
models; we will thereby treat every world as being accessible from every other.
Before laying out the semantics of @, let’s examine a slightly different way
of laying out standard modal logic. For propositional modal logic, instead of
defining a model as an ordered pair 〈W , I 〉 (no accessibility relation, remem-
ber), one could instead define a model as a triple 〈W , w@ , I 〉, where W and I
are as before, and w@ is a member of W , thought of as the actual, or designated
world of the model. The designated world w@ plays no role in the definition of
the valuation for a given model; it only plays a role in the definitions of truth
in a model and validity:
Definitions of truth in a model and validity with designated worlds:
· φ is true in model M (= 〈W , w@ , I 〉) iff VM (φ, w@ ) = 1

· φ is valid in system S iff φ is true in all models for system S
One could add a designated world to models for quantified modal logic in a
parallel way.
The old definition of validity for a system (section 6.3), recall, never em-
ployed the notion of truth in a model; rather, it proceeded via the notion of
validity in a frame. The nice thing about the new definition is that it’s parallel
In certain special cases, we could do without the new symbol @. For example, instead of
1
symbolizing “Necessarily, if snow is white then snow is actually white” as 2(S→@S), we could
symbolize it as 3S→S. But the @ is not in general eliminable; see Hodes (1984b,a).
to the way validity is usually defined in model theory: one first defines truth in
a model, and then defines validity as truth in all models. But the new definition
doesn’t differ in any substantive way from the old definition, in that it yields
exactly the same class of valid formulas:
Proof. It’s obvious that everything valid on the old definition is valid on the
new definition (the old definition says that validity is truth in all worlds in
all models; the addition of the designated world w@ doesn’t play any role in
defining truth at worlds, so each of the new models has the same distribution
of truth values as one of the old models.) Moreover, suppose that a formula is
invalid on the old definition—i.e., suppose that φ is false at some world, w, in
some model M . Now construct a model of the new variety that’s just like M
except that its designated world is w. φ will be false in this model, and so φ
turns out invalid under the new definition.
10.1.2 Semantics for @

Now for the semantics of @. We can give @ a very simple semantics using
models with designated worlds. Further, the designated world will now be
involved in the notion of truth in a model, not just in the definition of validity.
We’ll move straight to quantified modal logic, bypassing propositional logic. To
keep things simple, let the models have a constant domain and no accessibility
relation. (It will be obvious how to add these complications back in, if they are
desired.)
Definition of a Designated-world SQML-model: A designated-world
SQML-model is a four-tuple 〈W , w@ , D, I 〉, where:
· W is a non-empty set (“worlds”)
· w@ is a member of W (“designated/actual world”)
· D is a non-empty set (“domain”)
· I is a “interpretation” function that assigns semantic values as before (to
names: members of D; to predicates: extensions relative to worlds)
In the definition of the valuation for such a model, the semantic clauses for the
old logical constants run just as with SQML (section 9.3); and we now add a
clause for the new operator @:
· VM , g (@φ, w) = 1 iff VM ,g (φ, w@ ) = 1
i.e., @φ is true at any world iff φ is true in the designated world of the model.
10.1.3 Establishing validity and invalidity

The strategies for establishing the validity or invalidity of a given formula are
similar to those from chapter 9.
Example 10.1: Show that ∀x(F x∨2Gx) → 2∀x(Gx∨@F x)
i) Suppose for reductio that this formula is not valid. Then for some model
and some variable assignment g , Vg (∀x(F x∨2Gx) → 2∀x(Gx∨@F x), w@ ) =
0.
ii) Then Vg (∀x(F x∨2Gx), w@ ) = 1 and…
iii) …Vg (2∀x(Gx∨@F x), w@ ) = 0
iv) Given the latter, there is some world, call it “a”, such that R w@ a and
Vg (∀x(Gx∨@F x), a) = 0. And so, there is some object, call it “u”, in the
model’s domain, D, such that Vg x (Gx∨@F x, a) = 0
u
v) And so Vg x (Gx, a) = 0 and…

u
vi) …Vg x (@F x, a) = 0

u
vii) Given the latter, Vg x (F x, w@ ) = 0 (by the clause in the truth definition
u
for @)
viii) Given ii), for every object in D, and so for u in particular, Vg x (F x∨2Gx, w@ ) =
u
1.
ix) And so, either Vg x (F x, w@ ) = 1 or Vg x (2Gx, w@ ) = 1
u u
x) From ix) and vii), Vg x (2Gx, w@ ) = 1

u
xi) And so, Vg x (Gx, a) = 1, which contradicts v)

u
Example 10.2: Show that 2 2∀x(Gx∨@F x)→2∀x(Gx∨F x): here is a

model in which this formula is false:
W = {w@ , a}
D = {u}
I (F ) = {〈u, w@ 〉}
I (G) = ∅
The formula turns out false in this model, which means that it turns out false
in w@ : the consequent is false in @ because at world a, something (namely, u)
is neither G nor F ; but the antecedent is true there: since u is F at w@ , it’s
necessary that u is either G or actually F .
10.2 ×
Adding @ to the language of quantified modal logic is a step in the right
direction, since it allows us to express certain kinds of comparisons between
possible worlds that we couldn’t express otherwise. But it doesn’t go far enough;
we need a further addition.2 Consider this sentence:
It might have been the case that, if all those then rich
might all have been poor, then someone is happy
What it’s saying, in possible worlds terms, is this:
For some world w, if there’s a world v such that (every-

one who is rich in w is poor in v), then someone is happy
in w.
This is a bit like “It might have been that everyone who is actually rich is poor”;
in this new sentence the word ‘then’ plays a role a bit like the role ‘actually’
played in the earlier sentence. But the ‘then’ does not take us back to the actual
world of the model; it rather takes us back to the world, w, that is introduced
by the first possibility operator, ‘it might have been the case that’. We cannot,
therefore, symbolize our new sentence thus:
3(3∀x(@Rx→P x)→∃xH x)
for this has the truth condition that there is a world w such that, if there’s a
world v such that (everyone who is rich in w@ is poor in v), then someone is
happy in w. The problem is that the @, as we’ve defined it, always takes us
back to the model’s designated world, whereas what we need to do is to “mark”
a world, and have @ take us back to the “marked” world:
3×(3∀x(@Rx→P x)→∃xH x)
× marks the spot: it is a point of reference for subsequent occurrences of @.
2
See Hodes (1984a) on the limitations of @; see Cresswell (1990) on × (his symbol is “Ref”),
and further related additions.
10.2.1 Two-dimensional semantics for ×

So let’s further augment the language of QML with another one-place sentence
operator, ×. The idea is that ×φ means the same thing as φ, except that
subsequent occurrences of @ in φ are to be interpreted as picking out the world
that was the “current world of evaluation” when the × was encountered. (This
will become clearer once we lay out the semantics for × and @.)
To lay out this semantics, let’s return to the old SQML models (i.e., without
a designated world). Thus, a model is a triple 〈W , D, I 〉, W a non-empty set,
D a non-empty set, I a function assigning referents to names and extensions
to predicates at worlds as before. But now we change the definition of truth.
We no longer evaluate formulas at worlds. Instead we evaluate a formula at a
pair of worlds (hence: “two-dimensional semantics”). One world is the world
we’re used to; it’s the world that we’re evaluating the formula for truth in. Call
this the “world of evaluation”. The other world is a “reference world”—it’s
the world that we’re currently thinking of as the actual world, and the world
that will be relevant to the evaluation of @. Thus, “VM , g (φ, w1 , w2 )” will mean
that φ is true at world w2 , with reference world w1 . We define [α]M ,g , the
denotation of term α relative to model M and variable assignment g , as before.
And we define the valuation function as follows.
Definition of two-dimensional valuation function: The two-dimensional

valuation function, VM ,g , for an SQML-model M (= 〈W , D, I 〉) is defined as
the three-place function that assigns to each wff, relative to each pair of worlds,
either 0 or 1 subject to the following constraints, for any n-place predicate Π,
terms α1 . . . αn , wffs φ and ψ, and variable β:
· VM , g (Πα1 . . . αn , v, w) = 1 iff 〈[α1 ]M ,g , . . . , [αn ]M ,g , w〉 ∈ I (Π)

· VM , g (∼φ, v, w) = 1 iff VM , g (φ, v, w) = 0
· VM , g (φ→ψ, v, w) = 1 iff VM ,g (φ, v, w) = 0 or VM ,g (ψ, v, w) = 1
· VM ,g (∀βφ, v, w) = 1 iff for all u ∈ D, VM ,g α (φ, v, w) = 1
u
· VM ,g (2φ, v, w) = 1 iff for all w ∈ W , VM ,g (φ, v, w 0 ) = 1

0
· VM ,g (@φ, v, w) = 1 iff VM , g (φ, v, v) = 1

· VM ,g (×φ, v, w) = 1 iff VM ,g (φ, w, w) = 1
Note what the × does: change the reference world. When evaluating a formula,
it says to forget about the old reference world, and make the new reference
world whatever the current world of evaluation happens to be.
We can define validity and consequence thus:
Two-dimensional definitions of validity and consequence:

· φ is 2D-valid (“2D φ”) iff for every model M , every world w in that
model, and every assignment g based on that model, VM ,g (φ, w, w) = 1
· φ is a 2D-semantic consequence of Γ (“Γ 2D φ”) iff for every model
M , every assignment g based on that model, and every world w in that
model, if VM ,g (γ , w, w) = 1 for each γ ∈ Γ, then VM , g (φ, w, w) = 1
Valid formulas are thus defined as those that are true at every pair of worlds of
the form 〈w, w〉; semantic consequence is truth-preservation at every such pair.
Notice, however, that these aren’t the only notions of validity and con-
sequence that one could introduce. There is also the notion of truth, and
truth-preservation, at every pair of worlds:3
Definitions of general 2D validity and consequence:

· φ is generally 2D-valid (“G2D φ”) iff for every model M , any worlds
v and w in that model, and every assignment g based on that model,
VM ,g (φ, v, w) = 1
· φ is a general 2D-semantic consequence of Γ (“Γ G2D φ”) iff for every
model M , every assignment g based on that model, and any worlds v and
w in that model, if VM , g (γ , v, w) = 1 for each γ ∈ Γ, then VM ,g (φ, v, w) =
1
Validity and general validity, and consequence and general consequence, come
apart in various ways, as we’ll see below.
As we saw, moving to this new language increases the flexibility of the @;
we can symbolize
It might have been the case that, if all those then rich
might all have been poor, then someone is happy
3
The term ‘general validity’ is from Davies and Humberstone (1980); the first definition of
validity corresponds to their “real-world validity”.
as
3×(3∀x(@Rx→P x)→∃xH x)
Moreover, it costs us nothing. For we can replace any sentence φ of the old
language with ×φ in the new language (i.e. we just put the × operator at the
front of the sentence.)4 For example, instead of symbolizing
It might have been that everyone who is actually rich is

poor
as 3∀x(@Rx→P x) as we did before, we symbolize it now as:
×3∀x(@Rx→P x)
Example 10.3: Show that if φ then @φ. Suppose for reductio that
φ is valid but @φ is not. That means that in some model and some world,
w (and some assignment g , but I’ll suppress this since it isn’t relevant here),
V(@φ, w, w) = 0. Thus, given the truth condition for @, V(φ, w, w) = 0. But
that violates the validity of φ.
Example 10.4: Show that every instance of φ↔@φ is 2D-valid, but not
every instance of 2(φ↔@φ) is. (Moral: any proof theory for this logic had
better not include the rule of necessitation!) For the first, the truth condition
for @ insures that for any world w in any model (and any variable assignment),
V(@φ, w, w) = 1 iff V(φ, w, w) = 1, and so V(φ↔@φ, w, w) = 1. Thus,
φ↔@φ.
But some instances of 2(φ↔@φ) aren’t valid. Let φ be ‘F a’; here’s a
countermodel:
W = {c, d}
D = {u}
I (a) = u
I (F ) = {〈u, c〉}
4
This amounts to the same thing as the old symbolization in the following sense. Let
φ be any wff of the old language. Thus, φ may have some occurrences of @, but it has no
occurrences of ×. Then, for every SQML-model M = 〈W , D, I 〉, and any v, w ∈ W , ×φ is
true at 〈v, w〉 in M iff φ is true in the designated-world SQML model 〈W , w, D, I 〉.
In this model, V(2(F a↔@F a), c, c) = 0, because V(F a↔@F a, c, d) = 0. For

‘F a’ is true at 〈c, d〉 iff the referent of ‘a’ is in the extension of ‘F ’ at world d (it
isn’t) whereas ‘@F a’ is true at 〈c, d〉 iff the referent of ‘a’ is in the extension of
‘F ’ at world c (it is).
Note that this same model shows that φ↔@φ is not generally valid. General
validity is truth at all pairs of worlds, and the formula F a↔@F a, as we just
showed, is false at the pair 〈c, d〉.
Exercise 10.1 Demonstrate the following facts:
a) φ→2@φ (for any wff φ)
b) 2×∀x3@F x→2∀xF x
10.3 Fixedly
The two-dimensional approach to semantics—evaluating formulas at pairs of
worlds rather than single worlds—raises an intriguing possibility. The 2 is a
universal quantifier over the world of evaluation; we might, by analogy, follow
Davies and Humberstone (1980) and introduce an operator that is a universal
quantifier over the reference world. Davies and Humberstone call this operator
F, and read “Fφ” as “fixedly, φ”. Grammatically, F is a one-place sentential
operator. Its semantic clause is this:
· VM ,g (Fφ, v, w) = 1 iff for every v 0 , VM , g (φ, v 0 , w) = 1
All the other two-dimensional semantic definitions, including the definitions
of validity and consequence, remain the same.5
Humberstone and Davies point out that given F, @, and 2, we can introduce
two new operators: F@ and F2. It’s easy to show that:
· VM , g (F@φ, v, w) = 1 iff for every v 0 ∈ W , VM ,g (φ, v 0 , v 0 ) = 1

5
Humberstone and Davies don’t use two-dimensional semantics; they instead use designated-
world QML models (and they don’t include ×). Say that designated-world QML models are
variants iff they are alike except perhaps for the designated world. Their truth condition for F
is then this: VM , g (Fφ, w) = 1 iff for every model M 0 that is a variant of M , VM 0 , g (φ, w) = 1.
This approach isn’t significantly different from the present two-dimensional one.
· VM , g (F2, v, w) = 1 iff for v 0 , w 0 ∈ W , VM ,g (φ, v 0 , w 0 ) = 1
Thus, we can think of F@ and F2, as well as 2 and F themselves, as expressing

“kinds of necessities”, since their truth conditions introduce universal quantifiers
over worlds of evaluation and reference worlds. (What about 2F? It’s easy to
show that 2F is just equivalent to F2.)
As with the semantics of the previous section, validity and general validity
do not always coincide, as the following example shows.
Example 10.5: For example, F@φ→φ is 2D-valid for each wff φ (exer-
cise 10.2). But not every instance of this wff is generally valid. The formula
F@(@Ga↔Ga)→(@Ga↔Ga) is not generally valid, for example. General va-
lidity requires truth at all pairs 〈v, w〉 in all models. But in the following model,
Vg (F@(@Ga↔Ga)→(@Ga↔Ga), c, d) = 0 (for any variable assignment g ):
W = {c, d}
D = {u}
I (a) = u
I (G) = {〈u, c〉}
In this model, the referent of ‘a’ is in the extension of ‘G’ in world c, but not
in world d. That means that @Ga is true at 〈c, d〉 whereas Ga is false at 〈c, d〉,
and so @Ga↔Ga is false at 〈c, d〉. But F@φ means that φ is true at all pairs
of the form 〈v, v〉, and the formula @Ga↔Ga is true at any such pair (in any
model). Thus, F@(@Ga↔Ga) true at 〈c, d〉 in this model.
Exercise 10.2 Show that 2D F@φ→φ, for each wff φ
Exercise 10.3 Show that for some φ, 22D φ→Fφ.
Exercise 10.4 Show that if φ has no occurrences of @, then 2D

φ→Fφ (difficult).
Example 4:
10.4 A philosophical application: necessity and a

priority
The two-dimensional modal framework has been put to significant philosophi-
cal use in the past twenty-five or so years.6 This is not the place for an extended
survey; rather, I will briefly present the two-dimensional account of just one
philosophical issue: the relationship between necessity and a priority.
In Naming and Necessity, Saul Kripke famously presented putative examples
of necessary a posteriori statements and of contingent a priori statements:
Hesperus = Phosphorus
B (the standard meter bar) is one meter long
The first statement, Kripke argued, is necessary because whenever we try to

imagine a possible world in which Hesperus is not Phosphorus, we find that
we have in fact merely imagined a world in which ‘Hesperus’ and ‘Phosphorus’
denote different objects than they in fact denote. Given that Hesperus and
Phosphorus are in fact one and the same entity—namely, the planet Venus—
there is no possible world in which Hesperus is different from Phosphorus, for
such a world would have to be a world in which Venus is distinct from itself.
Thus, the statement is necessary, despite its a posteriority: it took astronomical
evidence to learn that Hesperus and Phosphorus were identical; no amount
of pure rational reflection would have sufficed. As for the second statement,
Kripke argues that one can know its truth as soon as one knows that the phrase
‘one meter’ has its reference fixed by the description: “the length of bar B”.
Thus it is a priori. Nevertheless, he argues, it is contingent: bar B does not
have its length essentially, and thus could have been longer or shorter than one
meter.
On the face of it, the existence of necessary a priori or contingent a posteriori
statements is paradoxical. How can a statement that is true in all possible worlds
be in principle resistant to a priori investigation? Worse, how can a statement
that might have been false be known a priori?
The two-dimensional framework has been thought by some to shed light
on all this. Let’s consider the contingent a priori first. Let’s define the following
6
For work in this tradition, see Stalnaker (1978, 2003a, 2004); Evans (1979); Davies and
Humberstone (1980); Hirsch (1986); Chalmers (1996); ?); Jackson (1998); see Soames (2004)
for an extended critique.
notion of contingency:
Definition of superficial contingency: φ is superficially contingent in model

M at world w iff, for every variable assignment g , VM ,g (2φ, w, w) = 0 and
VM ,g (2∼φ, w, w) = 0.
This corresponds, intuitively, to this: if we were sitting at w, and we uttered

3φ∧3∼φ, we’d speak the truth.
How should we formalize the notion of a priority? As a rough and ready
guide, let’s think of a sentence as being a priori iff it is 2D-valid—i.e., true at
every pair 〈w, w〉 of every model. In defense of this guide: we can think of the
truth value of an utterance of a sentence as being the valuation of that sentence
at the pair 〈w, w〉 in a model that accurately models the genuine possibilities,
and in which w accurately models the (genuine) possible world of the speaker.
So any 2D-valid sentence is invariably true whenever uttered; hence, if φ is
2D-valid, any speaker who understands his or her language is in a position to
know that an utterance of φ would be true.
Under these definitions, there are sentences that are superficially contingent
but nevertheless a priori. Consider any sentence of the form: φ↔@φ. In any
model in which φ is true at w and false at some other world, the sentence
is superficially contingent. But it is a priori, since, as we showed above, it is
2D-valid (though it’s not generally 2D-valid, as we also showed above.)
That was a relatively simple example; but one can give other examples that
are similar in spirit both to Kripke’s example of the meter bar, and to a related
example due to Gareth Evans (1979):
Bar B is one meter

Julius invented the zip
where bar B is the standard meter bar, and the “descriptive names” ‘one meter’
and ‘Julius’ are said to be “rigid designators” whose references are “fixed” by the
descriptions ‘the length of bar B’ and ‘the inventor of the zip’, respectively. Now,
whether or not these sentences, understood as sentences of everyday English,
are indeed genuinely contingent and a priori depends on delicate issues in the
philosophy of language concerning descriptive names, rigid designation, and
reference fixing. Rather than going into all that, let’s construct some examples
that are similar to Kripke’s and Evans’s. Let’s simply stipulate that ‘one meter’
and ‘Julius’ are to abbreviate “actualized descriptions”: ‘the actual length of bar
B’ and ‘the actual inventor of the zip’. With a little creative reconstruing in the
first case, the sentences then have the form: “the actual G is G”:
the actual length of bar B is a length of bar B

the actual inventor of the zip invented the zip
Now, these sentences are not quite a priori, since for all one knows, the G might
not exist—there might exist no unique length of bar B, no unique inventor of
the zip. So suppose we consider instead the following sentences:
If there is exactly one length of bar B, then the actual

length of bar B is a length of bar B
If there is exactly one inventor of the zip, then the actual
inventor of the zip invented the zip
Each has the form:
If there is exactly one G, then the actual G is G

Or, in symbols:
∃x(Gx∧∀y(Gy→y=x)) → ∃x(@Gx∧∀y(@Gy→y=x)∧Gx)
Any sentence of this form is 2D-valid (though not generally 2D-valid), and is
superficially contingent. So we have further examples of the contingent a priori
in the neighborhood of the examples of Kripke and Evans.
Various philosophers want to concede that these sentences are contingent
in one sense—namely, in the sense of superficial contingency. But, they claim,
this is a relatively unimportant sense (hence the term ‘superficial contingency’,
which was coined by Evans). In another sense, they’re not contingent at all.
Evans calls the second sense of contingency “deep contingency”, and defines it
thus (1979, p. 185):
If a deeply contingent statement is true, there will exist some state of
affairs of which we can say both that had it not existed the statement
would not have been true, and that it might not have existed.
The intended meaning of ‘the statement would not have been true’ is that the
statement, as uttered with its actual meaning, would not have been true. The
idea is supposed to be that ‘Julius invented the zip’ is not deeply contingent,
because we can’t locate the required state of affairs, since in any situation in
which ‘Julius invented the zip’ is uttered with its actual meaning, it is uttered
truly. So the Julius example is not one of a deeply contingent a priori truth.
Evans’s notion of deep contingency is far from clear. One of the nice things
about the two-dimensional modal framework is that it allows us to give a
clear definition of deep contingency. Davies and Humberstone (1980) give a
definition of deep contingency which is parallel to the definition of superficial
contingency, but with F@ in place of 2:
Definition of deep contingency: φ is deeply contingent in M at w iff (for

all g ) VM , g (F@φ, w, w) = 0 and VM , g (F@∼φ, w, w) = 0.
Under this definition, the examples we have given are not deeply contingent.
To be sure, this definition is only as clear as the two-dimensional notions of
fixedness and actuality. The formal structure of the two-dimensional framework
is of course clear, but one can raise philosophical questions about how that
formalism is to be interpreted. But at least the formalism provides a clear
framework for the philosophical debate to occur.
Our discussion of the necessary a posteriori will be parallel to that of the
contingent a priori. Just as we defined superficial contingency as the falsity of
the 2, so we can define superficial necessity as the truth of the 2:
Definition of superficial necessity: φ is superficially necessary in M at w

iff (for all g ) VM ,g (2φ, w, w) = 1
How shall we construe a posteriority? Let’s follow our earlier strategy, and take
the failure to be 2D-valid as our guide.
But here we must take a bit more care. It’s quite a trivial matter to construct
models in which 2D-invalid sentences are necessarily true; and we don’t need
the two-dimensional framework to do it. We clearly don’t want to say that
‘Everything is a lawyer” is an example of the necessary a posteriori. But let F
stand for ‘is a lawyer’; we can construct a model in which the predicate F is
true of every member of the domain at any world, ∀xF x is true, and so is
superficially necessary at every world, despite the fact that it is not 2D-valid.
But this is too cheap. We began by letting the predicate F stand for a predicate
of English, but then constructed our model without attending to the modal
fact that it’s simply not the case that it’s necessarily true that everything is a
lawyer. If F is indeed to stand for ‘is a lawyer’, we would need to include in any
realistic model—any model faithful to the modal facts—worlds in which not

everything is in the extension of F .
To provide nontrivial models of the necessary a posteriori, when we have
chosen to think of the nonlogical expressions of the language of QML as
standing for certain expressions of English, our strategy will be provide realistic
models—models that are faithful to the real modal facts in relevant respects,
given the choice of what the nonlogical expressions stand—in which 2D-invalid
sentences are necessarily true. Now, since the notion of a “realistic model”
has not been made precise, the argument here will be imprecise; but in the
circumstances this is inevitable.
So: consider now, as a schematic example of an a posteriori and superficially
necessary sentence:
If the actual F and the actual G exist, they are identical

In symbols:
[∃x(@F x∧∀y(@F y→x=y)) ∧ ∃z(@Gz∧∀y(@Gy→z=y)] →
∃x[@F x∧∀y(@F y→x=y) ∧ ∃z(@Gz∧∀y(@Gy→z=y) ∧ z=x)]
This sentence isn’t 2D-valid. Nevertheless, it is superficially necessary in any
model and any world w in which F and G each have a single object in their
extension, no matter what the extensions of F and G are in other worlds in the
model. So whenever such a model is realistic (given what we let F and G stand
for), we will have our desired example.
We can fill in this schema and construct an example similar to Kripke’s
Hesperus and Phosphorus example. Set aside controversies about the semantics
of proper names in natural language; let’s just stipulate that ‘Hesperus’ is to
be short for ‘the actual F ’, and that Phosphorus is to be short for ‘the actual
G’. And let’s think of F as standing for ‘the first heavenly body visible in the
evening’, and G for ‘the last heavenly body visible in the morning’. Then
(HP) If Hesperus and Phosphorus exist then they are identical
has the form ‘If the actual F and the actual G exist then they are identical’,
which was discussed in the previous paragraph. We may then construct a
realistic model in which F and G each have a single object in their extension in
some world, w, but in which they have different objects in their extensions in
other worlds. In such a model, the sentence
(2HP) 2(If Hesperus and Phosphorus exist then they are identical)
is true at 〈w, w〉, and so we again have our desired example: (HP) is superficially
necessary, despite the fact that it is a posteriori (2D-invalid).
Isn’t it strange that (HP) is both a posteriori and necessary? The two-
dimensional response is: no, it’s not, since although it is superficially necessary,
it isn’t deeply necessary in the following sense:
Definition of deep neceessity: φ is deeply necessary in M at w iff (for all g )

VM ,g (F@φ, w, w) = 1
It isn’t deeply necessary because in any realistic model (given what F and G
currently stand for), there must be worlds and objects other than c and u, that
are configured as they are in the model below:
W = {c, d}
D = {u, v}
I (F ) = {〈u, c〉, 〈u, d〉}
I (G) = {〈u, c〉, 〈v, d〉}
In this model, even though (2HP) is false at 〈c, c〉, still, F@(HP), i.e.:
F@{[∃x(@F x∧∀y(@F y→x=y)) ∧ ∃z(@Gz∧∀y(@Gy→z=y)] →

∃x[@F x∧∀y(@F y→x=y) ∧ ∃z(@Gz∧∀y(@Gy→z=y) ∧ z=x)]}
is false at 〈c, c〉 (and indeed, at every pair of worlds), since (HP) is false at 〈d, d〉.
And so, (HP) is not deeply necessary in this model.
One might try to take this two-dimensional line further, and claim that in
every case of the necessary a posteriori (or the contingent a priori), the necessity
(contingency) is merely superficial. But defending this stronger line would
require more than we have in place so far. To take one example, return again
to ‘Hesperus = Phosphorus’, but now, instead of thinking of ‘Hesperus’ and
‘Phosphorus’ as abbreviations for actualized descriptions, let us represent them
by names in the logical sense (i.e., the expressions called “names” in the defini-
tion of well-formed formulas, which are assigned denotations by interpretation
functions in models). Thus, ‘Hesperus = Phosphorus’ is now represented as:
a=b . Consider the following model:
W = {c, d}
D = {u, v}
I (a) = u
I (b ) = u
The model is apparently realistic; it falsifies no relevant modal facts. But the
sentence a=b is deeply necessary (at any world in the model). And yet it is a
posteriori (2D-invalid).
Exercise 10.5 Show that the symbolization of “If there is exactly

one G, then the actual G is G”, that is:
∃x(Gx∧∀y(Gy→y=x)) → ∃x(@Gx∧∀y(@Gy→y=x)∧Gx)
is valid, though not generally valid, and superficially contingent in

any world in any model.
Exercise 10.6 Show that a formula is capable of being superficially

contingent (i.e., for some model and some world, it is superficially
contingent at that world) iff it fails to be generally valid.
Appendix A
Answers to Selected Exercises
Exercise 5.6 We must show that for any PC+DD model 〈D, I , E 〉, and any
variable assignment g , [α] g (relative to this model) is either E or a member of
D. We’ll do this by induction on the grammar of α. So, we’ll show that the
result holds when α is a variable, constant, or ι term (base cases), and then show
that, assuming the result holds for simpler terms (inductive hypothesis), it also
holds for complex terms made up of the simpler terms using a function symbol.
Base cases. If α is a variable then [α] g is g (α), which is a member of D
given the definition of a variable assignment. If α is a constant then [α] g is
I (α), which is a member of D given the definition of a model’s interpretation
function. If α has the form ιβφ then [α] g is either the unique u ∈ D such that
V g β (φ) = 1 (if there is such a u) or E (if there isn’t). So in all three cases, [α] g
u
is either E or a member of D. (Note that even though ι terms are syntactically
complex, we treated them here as a base case of our inductive proof. That’s
because we had no need for any inductive hypothesis; we could simply show
directly that the result holds for all ι terms.)
Next we assume the inductive hypothesis:
(ih) The denotations of terms α1 . . . αn are either E or members of D
and show that the same goes for the complex term f (α1 . . . αn ). Well,
[ f (α1 . . . αn )] g is defined as I ( f )([α1 ] g . . . [αn ] g ). And the inductive hypothesis
tells us that each [αi ] g is either E or a member of D. And we know from the
definition of a model that I ( f ) is a function that maps any n-tuple of members
289
APPENDIX A. ANSWERS TO SELECTED EXERCISES 290
of D ∪ {E } to a member of D ∪ {E }. So I ( f )([α1 ] g . . . [αn ] g ) is a member of

D ∪ {E }—i.e. it is either E or a member of D.
Exercise 6.2a 2[P →3(Q→R)]→3[Q→(2P →3R)]:
D-Countermodel (hence the formula is invalid in K as well):
∗ ∗
r 1 0 0
2[P →3(Q→R)]→3[Q→(2P →3R)]

∗ ∗
1 1 1 1 0 1 1 0 0 0
a
P →3(Q→R) Q→(2P →3R)
† ∗
0

0 1 0 1
b Q→R P
†
0
Official model:
W = {r, a, b}
R = {〈r, a〉, 〈a, b〉, 〈a, a〉, 〈b, b〉}
I (P, a) = I (Q, a) = I (P, b) = 1, all else 0
T-validity proof (establishes validity in B, S4, and S5 as well):

i) Suppose for reductio that for some T-model 〈W , R, I 〉 and some r ∈ W ,
V(the formula, r ) = 0.
ii) So V(2[P →3(Q→R)], r ) = 1, and…
iii) …V(3[Q→(2P →3R)], r ) = 0
iv) From iii), for any v such that R r v, V(Q→(2P →3R), v) = 0
v) But R r r (reflexivity), so V(Q→(2P →3R), r ) = 0

vi) So V(2P →3R, r ) = 0 (truth condition for →)
vii) And so, V(2P, r ) = 1 (truth condition for →)
viii) And so, since R r r , V(P, r ) = 1.
ix) From ii), given R r r , we know that V(P →3(Q→R), r ) = 1.
x) Given viii) and the truth condition for →, V(3(Q→R), r ) = 1
xi) So for some world, call it “a”, R ra and V(Q→R, a) = 1
xii) Since R ra, from iv) we have V(Q→(2P →3R), a) = 0.
xiii) Given the truth condition for the →, V(Q, a) = 1 and …
xiv) …V(2P →3R, a) = 0
xv) That means that V(3R, a) = 0; so, since Raa (reflexivity), V(R, a) = 0.
xvi) Lines xiii), xv), and xi) contradict (truth condition for →)
Exercise 6.2d 2(P ↔Q)→2(2P ↔2Q):
B-countermodel (establishes invalidity in S5 as well):
∗
1 1 1 1 0 0
2(P ↔Q)→2(2P ↔2Q)
r
† ∗
0 O

∗
1 1 1 1 1 0 0
2P ↔2Q
a
P ↔Q
† † ∗
0 O

0 1
b
Q P
0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈a, r〉, 〈r, b〉, 〈b, r〉}
I (P, r) = I (Q, r) = I (P, a) = I (Q, a) = I (P, b) = 1, all else 0
Validity proof for S4 (and so for S5 as well):
i) Suppose for reductio that in some world r of some S4-model, the formula
is false.
ii) Then V(2(P ↔Q), r ) = 1 and…
iii) …V(2(2P ↔2Q), r ) = 0.
iv) Given iii), for some world a, R ra and V(2P ↔2Q, a) = 0.
v) Given iv), 2P and 2Q must have different truth values in world a. With-
out loss of generality (given the symmetry between P and Q elsewhere
in the problem), let’s suppose that …
vi) …V(2P, a) = 1 and …
vii) …V(2Q, a) = 0.
viii) Given vii), for some world b , Rab and V(Q, b ) = 0.
ix) Given vi), V(P, b ) = 1.
x) We already know that R ra and Rab . By transitivity, R r b .
xi) But then, given ii), V(P ↔Q, b ) = 1. This contradicts viii) and ix).
Exercise 6.2g 332P ↔2P :
B-countermodel (this establishes invalidity in S5 as well):

1 1 0 0
r 332P ↔2P
∗ † ∗
0 ? <] <
<<
<<
<<
<<
<<
∗ <<
1 1 1 0
32P
a b
P
0
∗
0
Official model:
W = {r, a, b}
I (P, r) = I (P, a) = 1, all else 0
S4-countermodel:
1 0 0
r 332P ↔2P
∗ † ∗
0 <<
<<
<<
<<
<<
<<
∗ <<
1 1 1 0
32P
a b
P
0
∗
0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈r, b〉}
I (P, a) = 1, all else 0
S5-validity proof:
i) We must show that in any world r of any S5-model, 332P and 2P have
the same truth value.
ii) So first suppose for reductio that V(332P, r ) = 1 and …
iii) …V(2P, r ) = 0.
iv) From ii), for some a, R ra and V(32P, a) = 1.
v) So, for some b , Rab and V(2P, b ) = 1.
vi) From iii), for some c, R r c and V(P, c) = 0.
vii) Since R ra and Rab , R r b (transitivity), and so R b r (symmetry); since

R r c, we have R b c (transitivity).
viii) But given R b c and v), V(P, c) = 1, which contradicts vi).
ix) So it cannot be that 332P is true at r while 2P is false. Suppose next

for reductio that the former is false and the latter is true; that is, …
x) V(332P, r ) = 0 and …
xi) …V(2P, r ) = 1.
xii) By reflexivity, R r r . So, given xi), V(32P, r ) = 1; and so, V(332P, r ) =

1, contradicting x).
Exercise 6.4a `K 3(P ∧Q)→(3P ∧3Q):
1. (P ∧Q)→P PL
2. 2[(P ∧Q)→P ] 1, NEC
3. 2[(P ∧Q)→P ]→[3(P ∧Q)→3P ] K3
4. 3(P ∧Q)→3P 2,3 MP
5. 3(P ∧Q)→3Q Similar to 1-4
6. 3(P ∧Q)→(3P ∧3Q) 4, 5, PL
Exercise 6.4c `K ∼3(Q∧R)↔2(Q→∼R):

1. ∼(Q∧R)→(Q→∼R) PL
2. 2[∼(Q∧R)→(Q→∼R)] 1, NEC
3. 2∼(Q∧R)→2(Q→∼R) 2, K, MP
4. 2∼(Q∧R)↔∼3(Q∧R) MN (modal negation, proved in book)
5. (Q→∼R)→∼(Q∧R) PL
6. 2(Q→∼R)→2∼(Q∧R) 5, NEC, K, MP
7. ∼3(Q∧R)↔2(Q→∼R) 3, 4, 6 PL
Exercise 6.4g We’re to show that `K 3(P →Q)↔(2P →3Q). This one’s
a bit tough. The trick for the first half is getting the right order for the PL
tautologies, and for the second half, getting the right PL strategy.
1. P →[(P →Q)→Q] PL
2. 2P →2[(P →Q)→Q] 1, NEC, K, MP
3. 2[(P →Q)→Q]→[3(P →Q)→3Q] K3
4. 2P →[3(P →Q)→3Q] 2, 3, PL
5. 3(P →Q)→(2P →3Q) 4, PL
I must now prove the right-to-left direction, namely, (2P →3Q)→3(P →Q).
Note that the antecedent of this conditional is PL-equivalent to ∼2P 3Q. So
my goal will be to get two conditionals, ∼2P →3(P →Q), and 3Q→3(P →Q),
from which the desired conditional follows by PL.
6. ∼P →(P →Q) PL
7. 3∼P →3(P →Q) 1, NEC, K3, MP
8. 3∼P →∼2P MN
9. Q→(P →Q) PL
10. 3Q→3(P →Q) 9, NEC, K3, MP
11. (2P →3Q)→3(P →Q) 7, 8, 10, PL
12. 3(P →Q)↔(2P →3Q) 5, 11, PL
Exercise 7.2 We must show that if Γ I φ then Γ PL φ. Suppose Γ I φ,

and let I be a PL-interpretation in which every member of Γ is true; we must
show that VI (φ) = 1. (VI , recall, is the classical valuation for I .) Consider
the intuitionist model with just one stage, r, in which formulas have the same
valuations as they have in the classical interpretation—i.e., 〈{r}, {〈r, r〉}, I *〉,
where I *(α, r) = I (α) for each sentence letter α. It’s easy to check that since
the intuitionist model has only one stage, the classical and intuitionist truth
conditions collapse in this case, so that for every wff φ, VI * (φ, r) = VI (φ). So,
since every member of Γ is true in I , every member of Γ is true at r in the
intuitionist model. Since Γ I φ, it follows that φ is 1 at r in the intuitionist
model; and so, φ is true in the classical interpretation—i.e., VI (φ) = 1.
Exercise 10.4 We’re to show that if φ has no occurrences of @, then 2D

φ→Fφ. Let’s prove by induction that if φ has no occurrences of @, then
φ→Fφ is generally valid (i.e., true in any model at any world pair 〈v, w〉 under
any variable assignment). The result then follows, because general validity
(truth at all pairs) obviously implies validity (truth at pairs 〈w, w〉).
First, let φ be atomic. Then we’re trying to show that for any worlds
v, w, and any variable assignment g , Vg (Πα1 . . . αn →F Πα1 . . . αn , v, w) =
1. Suppose otherwise—suppose that (i) Vg (Πα1 . . . αn , v, w) = 1 and
(ii) Vg (F Πα1 . . . αn , v, w) = 0. Given (ii), for some world, call it
v 0 , Vg (Πα1 . . . αn , v 0 , w) = 0, and so the ordered n-tuple of the denotations
of α1 . . . αn is not in the extension of Π at w, which contradicts (i).
Now the inductive step. We must assume that φ and ψ obey our statement,
and show that complex formulas built from φ and ψ also obey our statement.
That is, we assume the inductive hypothesis:
(ih) φ and ψ have no occurrences of @, G2D φ→Fφ and G2D ψ→Fψ
and we must show that the following are also generally valid: ∼φ→F∼φ,
(φ→ψ)→F(φ→ψ), ∀αφ→F∀αφ, 2φ→F2φ, Fφ→FF φ, and ×φ→F ×φ:
∼ : Suppose otherwise—suppose V(∼φ→F∼φ, v, w) = 0 for some v, w.
So V(∼φ, v, w) = 1 and V(F∼φ, v, w) = 0. So V(φ, v, w) = 0, and for some
v 0 , V(∼φ, v 0 , w) = 0; and so V(φ, v 0 , w) = 1. By (ih), V(φ→Fφ, v 0 , w) = 1, and
so V(Fφ, v 0 , w) = 1, and so V(φ, v, w) = 1—contradiction.
→ : Suppose for some v, w, V(φ→ψ)→ F(φ→ψ), v, w) = 0. So (i)
V(φ→ψ, v, w) = 1 and V(F(φ→ψ), v, w) = 0. So, for some world, call it u,
V(φ→ψ, u, w) = 0, and so V(φ, u, w) = 1 and V(ψ, u, w) = 0. Given the former
and the inductive hypothesis, V(Fφ, u, w) = 1, and so V(φ, v, w) = 1. And so,
given (i), V(ψ, v, w) = 1, and so, given the inductive hypothesis, V(Fψ, v, w) = 1,
and so V(ψ, u, w) = 1, which contradicts (ii).
∀ : Suppose for some v, w, Vg (∀αφ, v, w) = 1, but Vg (F∀αφ, v, w) = 0.
Given the latter, for some v 0 , Vg (∀αφ, v 0 , w) = 0; and so, for some u in the
domain, Vg α (φ, v 0 , w) = 0. Given the former, Vg α (φ, v, w) = 1; given (ih) it
u u
follows that Vg α (Fφ, v, w) = 1, and so, Vg α (φ, v 0 , w) = 1. Contradiction.
u u
2 : suppose (i) V(2φ, v, w) = 1 and (ii) V(F2φ, v, w) = 0, for some v, w.
From (ii), V(2φ, v 0 , w) = 0 for some v 0 , and so V(φ, v 0 , w 0 ) = 0 for some
w 0 . Given (i), V(φ, v, w 0 ) = 1; and so, given (ih), V(Fφ, v, w 0 ) = 1, and so
V(φ, v 0 , w 0 ) = 1. Contradiction.
F: suppose V(Fφ, v, w) = 1 and V(FFφ, v, w) = 0, for some v, w. From the
latter, V(Fφ, v 0 , w) = 0 for some v 0 , and so V(φ, v 00 , w) = 0 for some v 00 , which
contradicts the former.
×: suppose Vg (×φ, v, w) = 1 but Vg (F×φ, v, w) = 0, for some v, w. Given
the latter, Vg (×φ, v 0 , w) = 0 for some v 0 , and so Vg (φ, w, w) = 0, which con-
tradicts the former.
Appendix B
Answers to Remaining Exercises
Exercise 2.1 We’re to show that the defined symbols ∨ and ↔ get the right
truth conditions. We must first show that V (ψ∨χ ) = 1 iff either V (ψ) = 1 or
V (χ ) = 1, for any valuation V . ψ ∨ χ is short for ∼ψ→χ , so we need to show
that for any V ,V (∼ψ→χ ) = 1 iff V (ψ) = 1 or V (χ ) = 1.
First suppose that V (∼ψ→χ ) = 1. We must now show that V (ψ) = 1
or V (χ ) = 1. So suppose for reductio that this is not the case—i.e., suppose
that V (ψ) = 0 and V (χ ) = 0. Since V (ψ) = 0 then V (∼ψ) = 1, by the clause
in the definition of truth-in-a-valuation function for the ∼. And then, since
V (∼ψ) = 1 and V (χ ) = 0, V (∼ψ→χ ) = 1, by the clause in the definition
of truth-in-a-valuation function for the →. That contradicts the reductio
assumption.
Next suppose that either V (ψ) = 1 or V (χ ) = 1, and suppose for reductio
that V (∼ψ→χ ) = 0. Given the latter, V (∼ψ) = 1 and V (χ ) = 0 (clause for → );
and then given the clause for ∼, V (ψ) = 0. But if both V (ψ) = 0 and V (χ ) = 0,
that contradicts the initial supposition that either V (ψ) = 1 or V (χ ) = 1.
Next we must show that V (ψ↔χ ) = 1 iff V (ψ) = V (χ ) (for any V ).
“ψ↔χ ” is short for: “(ψ→χ )∧(χ →ψ)”. The ∧is still not part of our basic
vocabulary; however, we showed in class that, given how ∧is defined, V (α∧β) =
1 iff V (α) = 1 and V (β) = 1. Given this fact, V (ψ↔χ ) = 1 iff V (ψ→χ ) = 1
and V (χ →ψ) = 1. But it is true that both V (ψ→χ ) = 1 and also V (χ →ψ) = 1,
iff ψ and χ have the same truth value in V —i.e., iff V (ψ) = V (χ ). For if they
have different truth values then one of the conditionals ψ→χ or χ →ψ must
be false (whichever one is 1→0); and conversely, if ψ and χ have the same truth
value then each of these conditionals must be true (since both 0→0 and 1→1
298
APPENDIX B. ANSWERS TO REMAINING EXERCISES 299
are 1.)
Exercise 2.2a Sequent proof of P, Q, R ` P :

1 (1) P As.
2 (2) Q As.
3 (3) R As.
2,3 (4) Q∧R 2,3 ∧I
1,2,3 (5) P ∧(Q∧R) 1,4 ∧I
1,2,3 (6) P 5, ∧E
Exercise 2.2b Sequent proof of P →(Q→R) ` (Q∧∼R)→∼P :

1 (1) P →(Q→R) As
2 (2) Q∧∼R As (for conditional proof)
1,3 (4) Q→R 1, 3 →E
2 (5) Q 2, ∧E
1,2,3 (6) R 4,5 →E
2 (7) ∼R 2, ∧E
1,2,3 (8) R∧∼R 6,7 ∧I
1,2 (9) ∼P 8, RAA
1 (10) (Q∧∼R)→∼P 9, →I
Exercise 2.2c Sequent proof of P →Q, R→Q ` (P ∨R)→Q:

1 (1) P →Q As
2 (2) R→Q As
3 (3) P ∨R As (for conditional proof)
4 (4) P As (for use with ∨E)
1,4 (5) Q 1, 4→E
6 (6) R As (for use with ∨E)
2,6 (7) Q 2,6 →E
1,2,3 (8) Q 3, 5, 7 ∨E
1,2 (9) (P ∨R)→Q 8, →I
Exercise 2.3a Axiomatic proof of P →P :
1. (P →((P →P )→P )→ From (A2); φ = P ,ψ = P →P , χ = P

((P →(P →P ))→(P →P ))
2. P →((P →P )→P ) From (A1); φ = P, ψ = P →P
3. (P →(P →P ))→(P →P ) 1,2 MP
4. P →(P →P ) From (A1); φ, ψ = P
5. P →P 3,4 MP
Exercise 2.3b Axiomatic proof of: (∼P →P )→P :
1. ∼P →∼P Repeat the proof in problem #1, but

substitute ‘∼P ’ everywhere for ‘P ’
2. (∼P →∼P )→((∼P →P )→P ) From (A3); φ, ψ = P
3. (∼P →P )→P 1, 2 MP
Exercise 2.3c Axiomatic proof of P from ∼∼P :
1. ∼∼P Premise
2. ∼∼P →(∼P →∼∼P ) A1
3. ∼P →∼∼P 1,2, MP
4. (∼P →∼∼P )→[(∼P →∼P )→P ] A3
5. [(∼P →∼P )→P ] 3,4 MP
6. ∼P →∼P Repeat the proof of #1
7. P 5,6 MP
Exercise 2.4a Axiomatic proof of ` P →[(P →Q)→Q]:
1. (P →Q)→(P →Q) Repeat the proof of #1

2. [(P →Q)→(P →Q)]→(P →[(P →Q)→Q]) “Swapping antecedents”
3. P →[(P →Q)→Q] 1, 2 MP
Exercise 2.4b Axiomatic proof that ` ∼∼P →P : Here’s my strategy. Look

back to problem 3. Its conclusion is P , whereas what I want is ∼∼P →P . So
I’m going to convert the proof of problem 3 by sticking a “∼∼P → ” in front

of a number of its lines. When problem 3 uses modus ponens, I can just use
the modus ponens technique. When problem 3 gets to its conclusion, I’ll be
at the conclusion I want. I’m going to start at line 3 of problem 3, since I can
stick a “∼∼P → ” in front of that easily, by A1. Here is the proof:
1. ∼∼P →(∼P →∼∼P ) A1 (counterpart of line 3

of old proof)
2. (∼P →∼∼P )→[(∼P →∼P )→P ] A3
3. ∼∼P →[(∼P →∼∼P )→[(∼P →∼P )→P ]] 2, adding an antecedent
(cpt of old line 4)
4. ∼∼P →[(∼P →∼P )→P ] 1, 3, MP technique
5. ∼P →∼P Repeat the proof of
problem 1
6. ∼∼P →(∼P →∼P ) 5, adding an antecedent
(cpt of old line 6)
7. ∼∼P →P 4, 6, MP technique
Exercise 2.4c Axiomatic proof that ` P →∼∼P :
1. ∼∼∼P →∼P repeat proof of problem 5

2. (∼∼∼P →∼P )→(P →∼∼P ) contraposition 1
3. P →∼∼P 1, 2, MP
Exercise 2.5 The system we are to consider uses the same definition of wffs
and the same rule (MP), but has different axioms:
φ→φ
(φ→ψ)→(ψ→φ)
Let’s first prove that a) every theorem of this system has an even number of
“∼”s:
A theorem of a system is the last line of a proof. So I’ll prove by induction
that every line of every proof in this system has an even number of ∼s. To do
that, I need to prove i) that every axiom of the system has an even number of
∼s (base case), and ii) that if we assume that φ and φ→ψ each have an even
number of tildes, then it follows that what you get from those formulas by
MP—i.e., ψ—must also have an even number of ∼s (inductive step).
Base case: that’s easy. In each axiom schema, each Greek letter occurs twice.
In the first schema, φ occurs twice, and in the second schema, both φ and ψ
occur twice. So whenever we construct an axiom from either schema, each ∼
that occurs in any wff that we stick in for φ or for ψ will appear in the axiom
twice. Thus, that axiom will have an even number of ∼s.
Inductive step: assume the inductive hypothesis: that φ and φ→ψ each
have an even number of ∼s. Let n be the number of ∼s in φ, and let m be the
number of ∼s inφ→ψ. The inductive hypothesis tells us that both n and m
are even. That means that m − n is even. But m − n is the number of ∼s in ψ.
So ψ has an even number of ∼s.
Next I’ll show that b) not every theorem of this system is valid. To do this,
I just need to produce a single theorem of this system and a single valuation
in which this theorem is false. I choose the theorem (P →Q)→(Q→P ). That’s
a theorem of the system because it’s an axiom (second axiom schema). The
valuation I choose is one in which Q is 1 and P is 0. In this valuation P →Q is
true (because P is 0), and Q→P is 0 (because Q is 1 and P is 0); so that means
that the whole thing is 0.
NOTE: you can’t just say that the axiom schema (φ→ψ)→(ψ→φ) is invalid.
First, it’s just a schema, not a wff, so the notion of validity doesn’t apply to
it. Second, there are instances of this schema that are valid, for instance:
[(P →P )→Q]→[Q→(P →P )].
Exercise 2.6 We are to show (for regular propositional logic) that the truth
value of a formula depends only on the truth values of the sentence letters in
that formula.
Let φ be any wff and let V and V 0 be valuations that agree on the sentence
letters in φ (i.e., for any sentence letter α, if α is in φ then V (α) = V 0 (α)).
Show that V (φ) = V 0 (φ).
Let V and V 0 be as described. Let’s show by induction that every formula
φ containing only sentence letters on which V and V 0 agree is such that
V (φ) = V 0 (φ). Since we’re trying to show something of the form “all formulas
φ are blah blah blah”, our base case is to show that all atomic formulas are blah
blah blah; and our induction step will be to show that if ψ and χ are blah blah
blah, then so are ∼ψ and ψ→χ .
Base case: we must show that if φ is atomic then V (φ) = V 0 (φ). But it’s
given in the problem that V (α) = V 0 (α) if α is an atomic in φ. Since in this

case, φ is atomic, φ is one of those αs, and so we have V (φ) = V 0 (φ).
Inductive step: assume that ψ and χ obey the theorem we’re trying to
prove—i.e., assume thatV (ψ) = V 0 (ψ), and also that V (χ ) = V 0 (χ ), where
ψ and χ are formulas made up of atomics over which V and V 0 agree. We
now need to show that both ∼ψ and ψ→χ obey the theorem also—i.e., that
V (∼ψ) = V 0 (∼ψ), and that V (ψ→χ ) = V 0 (ψ→χ ). Well, in any valuation, the
truth value of ∼ψ is just the opposite of the truth value of ψ. So, since we’re
given that V (ψ) = V 0 (ψ), the truth of ψ in both V and V 0 is the opposite of
whatever truth value ψ has in both V and V 0 . Thus, V (∼ψ) = V 0 (∼ψ). Finally,
in any valuation, the truth value of ψ→χ is 0 if ψ is true and χ is false; otherwise
ψ→χ is 1. But we’re given that V (ψ) = V 0 (ψ), and also that V (χ ) = V 0 (χ ).
So if ψ is 1 and χ is 0 in both V and V 0 then ψ→χ is false in both V and V 0 ;
otherwise ψ→χ is true in both V and V 0 . Either way, V (ψ→χ ) = V 0 (ψ→χ ).
End of inductive proof.
Exercise 2.7 We must show that for any set of formulas, Γ, and any formula
φ, if Γ ` φ then Γ φ (i.e., if φ is provable from Γ then φ is a semantic
consequence of Γ.) Like the proof of the original version of soundness, let’s
do this by induction. Here we’re not proving that every formula has a certain
property; we’re trying to prove that anything that is provable from Γ has a certain
property. So our inductive proof will concern the successive addition of lines
to a growing proof according to the rules of proof, not the successive addition
of more formulas to a growing formula by the rules of grammar.
Remember that a formula φ is provable from Γ iff there exists a proof from Γ
(i.e., a proof in which each line is either an axiom, a member of Γ , or follows
from earlier lines in the proof by MP) whose last line is φ. So let’s prove by
induction that the last line of every proof from Γ is a semantic consequence of Γ . And
we do that, in essence, by showing that every time you add to a proof from Γ,
you must always add a formula that is a semantic consequence of Γ. Formulas
you add to the proof fall into two categories: i) axioms and members of Γ , and
ii) formulas following by MP from earlier lines.
Base case: we must show that axioms and members of Γ are semantic
consequences of Γ. What does it mean to say that a formula ψ is a semantic
consequence of Γ ? It means that for any valuation, V , if every member of Γ
is true in V , then so is ψ. Well, it’s then obvious that any member of Γ is a
semantic consequence of Γ (obviously, a member of Γ is true in any valuation
that counts all of Γ s members true.) What about axioms—are they all true
in every valuation that counts every member of Γ true? Yes—because axioms

are true in all valuations whatsoever (we proved this when proving the original
soundness theorem).
Inductive case: Here we start by assuming that the inputs to modus ponens,
namely φ and φ→ψ, are both semantic consequences of Γ, and then show that
the output of modus ponens, namely ψ, is a semantic consequence of Γ—i.e.,
that ψ is true in any valuation that counts every member of Γ as true. (We do
this because we’re assuming that a certain proof has the feature we’re interested
in—each line is a semantic consequence of Γ—and then trying to prove that it
follows that we can add one more line to the proof and be assured that it will
continue to have the feature.) Well, let V be any such valuation. Since φ and
φ→ψ are semantic consequences of Γ, V (φ) = 1 and V (φ→ψ) = 1. But then,
V (ψ) must be 1 (since if it were 0, since V (φ) = 1, V (φ→ψ) would have to be
0, which it isn’t.) End of inductive proof.
Exercise 2.8 We must show that if Γ ` φ is a provable sequent, then Γ φ.

A provable sequent is the last line of any sequent proof, where a sequent
proof is a list of sequents, each of which is either the rule of assumptions,
or comes from earlier lines in the proof by one of the rules (∧I, →E, etc.)
Let’s call a sequent Γ ` φ a valid sequent iff Γ φ (i.e., iff φ is true in every
interpretation in which all the members of Γ are true.) So all we need to do is
prove by induction that every sequent in every sequent proof is a valid sequent.
Base: prove that sequents coming from the rule of assumptions are valid
sequents. That’s easy. Such sequents look like this: φ ` φ. Any such sequent is
obviously valid, since φ is obviously true in every valuation in which φ is true.
Induction: prove that each rule preserves validity of sequents. I’ll do this for
a few of them to illustrate the idea:
∧E: We assume that the input to this rule—i.e., Γ ` φ∧ψ—is a valid sequent,
and must show that its outputs (namely, Γ ` φ and Γ ` ψ) are valid sequents.
To show that Γ ` φ is a valid sequent, consider any valuation V in which every
member of Γ is true; we must show that V(φ) = 1. By the assumption (namely,
Γ ` φ∧ψ is a valid sequent), we know that V(φ∧ψ) = 1. Given the derived
truth conditions for ∧(section 2.3), this means that V(φ) = 1. The proof for ψ
is parallel.
∨E: We assume that the inputs to this rule, i.e., Γ ` φ∨ψ; ∆1 ,φ ` χ ; and
∆2 , ψ ` χ are valid sequents, and show that the rule’s output, i.e., Γ, ∆1 , ∆2 ` χ
is also a valid sequent. To show this, consider any valuation V in which the
members of Γ, ∆1 and ∆2 are all true. Since Γ ` φ∨ψ is a valid sequent,
V(φ∨ψ) = 1, so either V(φ) = 1 or V(ψ) = 1. If the former, then since ∆1 , φ ` χ

is a valid sequent, and we have assumed that all the members of ∆1 are true
in V, then V(χ ) = 1. Similarly, if the latter then V(χ ) = 1 (since ∆2 , ψ ` χ is a
valid sequent). Either way, V(χ ) = 1.
RAA: We assume that the input to this rule, Γ, φ ` ψ∧∼ψ, is a valid sequent,
and show that its output, Γ ` ∼φ, is also a valid sequent. To show this, let V
be any valuation in which every member of Γ is true. Then V(∼φ) must be 1.
For if it were 0, then V(φ) would be 1. But then, since Γ, φ ` ψ∧∼ψ is a valid
sequent, V(ψ∧∼ψ) would be 1, which is impossible.
→I: We assume that the input to this rule, Γ, φ ` ψ, is a valid sequent, and
show that its output, Γ ` φ→ψ, is also a valid sequent. To show this, let V be
any valuation in which every member of Γ is true, and suppose for reductio
that V(φ→ψ) = 0. Then V(φ) = 1 and V(ψ) = 0. But if V(φ) = 1, then since
Γ, φ ` ψ is a valid sequent (and since every member of Γ is true in V), V(ψ) = 1.
contradiction.
Exercise 3.1 The connective % is stipulated to have the following truth
table:
% 1 0
1 0 1
0 1 0
Let’s show that negation cannot be defined using just %. ∼P is 1 when P is 0,

whereas we can show by induction that any φ constructed from just P and %
is always 0 whenever P is 0. Let V be any valuation such that V(P ) = 0.
Base: φ is atomic. Then φ is just P , and so V(φ) = 0
Induction: assume result holds for φ and ψ (i.e., each is 0 in V). Show result
holds for φ%ψ—i.e., show that this too is 0 in V. That follows from the truth
table for %; 0%0 = 0.
Exercise 3.2 We are to express these truth functions:
f (1, 1) = 1 g (1, 1, 1) = 1
f (1, 0) = 0 g (1, 1, 0) = 0
f (0, 1) = 0 g (1, 0, 1) = 1
f (0, 0) = 1 g (1, 0, 0) = 1
g (0, 1, 1) = 1
g (0, 1, 0) = 1
g (0, 0, 1) = 0
g (0, 0, 0) = 1
Function f is expressed in standard propositional logic by: P ↔Q. To get the

Sheffer stroke equivalent, note that P ↔Q is equivalent to (P →Q)∧(Q→P ),
which is in turn equivalent to
∼(P ∧∼Q)∧∼(∼P ∧Q)
Now, | is equivalent to “not both”, so we can rewrite this as follows:
(P |∼Q) ∧ (∼P |Q)
And since ∼φ is equivalent to φ|φ, we can rewrite this as follows:
[P |(Q|Q)] ∧ [(P |P )|Q]
Finally, φ∧ψ is equivalent to ∼∼(φ∧ψ), which is equivalent to ∼(φ|ψ), which

is equivalent to (φ|ψ)|(φ|ψ). So the final answer is:
([P |(Q|Q)]|[(P |P )|Q]) | ([P |(Q|Q)]|[(P |P )|Q])
As for function g , we want a sentence, S, containing the sentence letters P , Q,

and R, that has the listed truth table. Our sentence S ought to be false in just
two cases:
· P and Q are true and R is false
· P and Q are false and R is true

Thus, the whole thing is false exactly when one of the following is true:
P ∧Q∧∼R
∼P ∧∼Q∧R
i.e., when the following is true:
(*) (P ∧Q∧∼R) ∨ (∼P ∧∼Q∧R)
Thus, the whole thing is true when (*) is false—i.e., when the negation of (*) is
true. But the negation of (*) is equivalent to:
∼(P ∧Q∧∼R) ∧ ∼(∼P ∧∼Q∧R)
So S should be equivalent to this last sentence. Let’s arbitrarily pick some

groupings for these three-way conjunctions (officially ∧ is a binary connective;
I’ve been sloppy in putting in the parentheses because ∧ is associative):
∼[(P ∧Q)∧∼R] ∧ ∼[(∼P ∧∼Q)∧R]
This is equivalent to:
[(P ∧Q)|∼R] ∧ [(∼P ∧∼Q)|R]
which in turn is equivalent to:
[(P ∧Q)|(R|R)] ∧ [((P |P )∧(Q|Q))|R]
The only remaining thing to do is eliminate the ∧s with their equivalents using
the |, via the equivalence mentioned above: φ∧ψ is equivalent to (φ|ψ)|(φ|ψ).
That would take forever, and I’m getting a bit lazy, so I won’t write it out.
Note: a simpler expression of function g is: (P ↔Q)→(Q↔R), which
could then be expressed using the Sheffer stroke.
Exercise 3.3 The truth table for ↓ is the following:
% 1 0
1 0 0
0 0 1
We know that all the truth functions can be defined using just ∼ and ∨. So all
we need to do is show that ∼ and ∨ can be defined using ↓.
First, ∼φ can be defined as φ ↓ φ. (The ↓ generates a false sentence from
two trues, and a true sentence from two falses.)
Second, note that φ ↓ ψ is equivalent to ∼(φ∨ψ). So φ∨ψ is equivalent to
∼(φ ↓ ψ), and hence to (φ ↓ ψ) ↓ (φ ↓ ψ).
Exercise 3.4a P ↔∼P becomes: ↔P ∼P

Exercise 3.4b (P →(Q→(R→∼∼(S∨T )))) becomes: →P →Q→R∼∼∨ST
Exercise 3.4c [(P ∧∼Q)∨(∼P ∧Q)]↔∼[(P ∨∼Q)∧(∼P ∨Q)] becomes:

↔∨∧P ∼Q∧∼P Q∼∧∨P ∼Q∨∼P Q
Here’s how to work this out step by step. Start with the conjuncts on the left
hand side:
P ∧∼Q : ∧P ∼Q
∼P ∧Q : ∧∼P Q
then plug these after an ∨ to get the entire left hand side:
(P ∧∼Q)∨(∼P ∧Q) : ∨∧P ∼Q∧∼P Q
now go to work on the small formulas on the right hand side:
P ∨∼Q : ∨P ∼Q
∼P ∨Q : ∨∼P Q
now to get the conjunction on the right hand side, put an ∧ followed by these
last two formulas:
(P ∨∼Q)∧(∼P ∨Q) : ∧∨P ∼Q∨∼P Q
put a ∼ in front of this to negate it, and so obtain the entire right hand side:
∼[(P ∨∼Q)∧(∼P ∨Q)] : ∼∧∨P ∼Q∨∼P Q
Now, for the whole formula. The major connective is the ↔, so we start with
a ↔, then put the whole symbolization for the left hand side, followed by the
whole symbolization for the right hand side:
↔ ∨∧P ∼Q∧∼P Q ∼∧∨P ∼Q∨∼P Q
Exercise 3.5 Indefinability of the → for Łukasiewicz. We must show that

there is no wff φ such that i) φ contains just the sentence letters P and Q, plus
the connectives ∼, ∧, and ∨ (plus parentheses), and ii) φ has the same truth
table as P →Q (i.e., φ is true in exactly the same Łukasiewicz-valuations as
P →Q).
P →Q is true in the Łukasiewicz tables when P and Q are both #. I’ll now
show by induction that any wff φ containing just P , Q, ∼, ∧ and ∨ gets assigned
# by any Łukasiewicz-valuation that assigns both P and Q #, and hence has a
different truth table from P →Q.
Let I be any trivalent interpretation in which both P and Q are #. Here
is the inductive proof that for any φ made up of just P , Q, ∼, ∧, and ∨,
ŁVI (φ) = #:
Base: φ is atomic. That means that φ is either P or Q, and so ŁVI (φ) = #.
Induction: assume the result holds for φ and ψ (i.e., ŁVI (φ) = #, ŁVI (ψ) =
#), and show that the result also holds for ∼φ, φ∧ψ, and φ∨ψ—i.e., show that
each of these is # in ŁVI . This follows from the truth tables for the ∼, ∧and ∨:
whenever φ is #, so is ∼φ; and whenever φ and ψ are both #, so are φ∧ψ and
φ∨ψ.
Exercise 3.8 We are to show that no wff in which no sentence letter is

repeated is supervaluationally valid. Call a sentence “fresh” if no sentence letter
is repeated in it, and consider the trivalent interpretation I that assigns # to
each sentence letter. I’ll show by induction that each wff φ has the following
property (which I’ll call “the property”)
If φ is fresh then SI (φ) = #
Base case: suppose φ is a sentence letter. Then by stipulation, I (φ) = #, and so

SI (φ) = #.
Induction: suppose that φ and ψ have the property; we must show that (a)
φ→ψ and (b) ∼φ have the property.
(a): To show that φ→ψ has the property, we assume φ→ψ is fresh; we
must show that SI (φ→ψ) = #. In order to do that, we need to construct two
precisifications of I : one in which φ→ψ is 1, one in which φ→ψ is 0.
Since φ→ψ is fresh, we know that φ and ψ are each fresh. Since by the
inductive hypothesis each has the property, we know that SI (φ) = # and
SI (ψ) = #. But now: if SI of any formula is #, that means that the formula is 1
in some precisification of I , and also that the formula is 0 in some precisification
of I . So, since SI (φ) and SI (ψ) are both #, we know that there exist four
precisifications of I : B, C , D, and E , such that:
VB (φ) = 1 VC (φ) = 0
VD (ψ) = 1 VE (ψ) = 0
Note that VC (φ→ψ) = 1. So this is the first of the two precisifications of I

that we need. The other one we need is a precisification of I in which φ→ψ
is false. Let F be the precisification that is just like B except that it assigns
whatever E assigns to any sentence letter occurring in ψ. Notice that, since
φ→ψ was fresh, no sentence letter occurs in both φ and ψ; and so all three
precisifications F , B, and E assign the same truth values to all the sentence
letters occurring in either φ or ψ. But that means that φ has the same truth
value in F , B, and E ; and likewise for ψ (since, as shown in exercise 2.6, in
any PL-valuation the truth value of an entire sentence is a function solely of
the truth values of the sentence letters that occur in that sentence.) And that
means that φ is 1 in F and ψ is 0 in F ; hence, VF (φ→ψ) = 0.
(b): It remains to show that ∼φ has the property. That’s easy: to show it
has the property we must assume that ∼φ is fresh, and show that SI (∼φ) = #.
But if ∼φ is fresh then so is φ; so by the inductive hypothesis, SI (φ) = #. That
means that φ is 1 in some precisifications of I and 0 in others. But ∼φ is 0 in
the first precisifications and 1 in the second precisifications, and so SI (∼φ) = #.
Exercise 4.1a What we are trying to show is that ∀x(F x→(F x∨Gx)) is
valid—i.e., that this formula is true in every model—i.e., that for any model
M (= 〈D, I 〉), and any assignment to the variables g defined on that model,
V g ,M (∀x(F x→(F x∨Gx))) = 1. (I’ll leave the subscript M implicit from now
on.) So, suppose for reductio that this is not true—that is, suppose for some
model and some g defined on that model, we have:
V g (∀x(F x→(F x∨Gx))) = 0
We then reason as follows:
i) So, for some u ∈ D,V g x (F x→(F x∨Gx)) = 0. Call one such u, “u”.
u
ii) V g x (F x) = 1 and V g x (F x∨Gx) = 0 (clause for →)

u u
iii) Given ii), V g x (F x) = 0 (derived clause for ∨)

u
iv) Lines ii) and iii) contradict.
Exercise 4.1b ∀x(F x∧Gx)→(∀xF x∧∀xGx):
i) Suppose for reductio that for some model and some g , V g (∀x(F x∧Gx)→
(∀xF x∧∀xGx)) = 0
ii) Then V g (∀x(F x∧Gx)) = 1 and V g (∀xF x∧∀xGx)) = 0
iii) Given the second, either V g (∀xF x) = 0 or V g (∀xGx) = 0
iv) Suppose the former (i.e., V g (∀xF x) = 0). Then for some u in the domain,
V g x (F x) = 0. From the first part of ii), we know that for every object in
u
the domain, and so for u in particular, V g x (F x∧Gx) = 1. From the clause
u
in the definition of V for ∧, we know that V g x (F x) = 1. Contradiction.
u
v) Suppose the latter (i.e., V g (∀xGx) = 0). Then, for some v in the domain,
V g x (Gx) = 0. From the first part of ii), we know that for every object in
v
the domain, and so for v in particular, V g x (F x∧Gx) = 1. So V g x (Gx) = 1.
v v
contradiction.
vi) So either way we get a contradiction.
Exercise 4.1c What we’re trying to show is that the set of formulas
{∀x(F x→Gx), ∀x(Gx→H x)} logically implies the sentence ∀x(F x→H x).
That is, we’re trying to show that ∀x(F x→H x) is true in every model, M , in
which all the premises in the set are true. So, we proceed as follows: suppose
for reductio that in some model, M , each of the premises are true and the
conclusion is false. We then reason as follows:
i) Since the conclusion is false in this model, we know that for some g ,
V g (∀x(F x→H x)) = 0.
ii) Since the premises are true, we know that for each variable assignment,
and so for g in particular, V g (∀x(F x→Gx)) = 1, and V g (∀x(Gx→H x)) =
1
iii) From i), for some u ∈ D, V g x (F x→H x) = 0 (clause for ∀ ). Call it “u”
u
iv) So, V g x (F x) = 1 and V g x (H x) = 0 (clause for →)

u u
v) Given ii), we know that for all v ∈ D, V g x (F x→Gx) = 1 and

v
V g x (Gx→H x) = 1.
v
vi) Since u ∈ D, we can conclude from v) that: V g x (F x→Gx) = 1 and

u
V g x (Gx→H x) = 1.
u
vii) Given the clause for →, we have:
a) V g x (F x) = 0 or V g x (Gx) = 1; and
u u
b) V g x (Gx) = 0 and V g x (H x) = 1
u u
viii) Given lines iv) and vii)a), we have V g x (Gx) = 1

u
ix) And so, given line vii)b), we have V g x (H x) = 1.

u
x) That contradicts line iv).
Exercise 4.1d ∃x∀yRxy→∀y∃xRxy:
i) Suppose for reductio that for some model and some

g ,V g (∃x∀yRxy→∀y∃xRxy) = 0.
ii) So, V g (∃x∀yRxy) = 1 and V g (∀y∃xRxy) = 0
iii) Given the former, for some member of the domain, call it u,
V g x (∀yRxy) = 1
u
iv) Given the latter, for some member of the domain, call it v, V g y (∃xRxy) =
v
0
v) From line iii), we know that for each member of the domain, and so for
v in particular, V g xy (Rxy) = 1
uv
vi) From line iv) we know that for each member of the domain, and so for u
in particular, V g y x (Rxy) = 0.
vu
vii) The function g uxvy is the same function as the function gvyux (each is the
function just like g , except that it assigns u to x and v to y.) So the
previous two lines contradict.
Exercise 4.2a We must show that 2 ∀x(F x→Gx)→∀x(Gx→F x). To say

that a sentence is valid is to say that it is true in all models. So in order to show
that a sentence is invalid, all we need to do is to produce one model in which
the sentence is false. Our given sentence says that if all F s are Gs, then all Gs
are Fs. So let’s produce a model in which the set of Gs contains the set of F s as
a subset, but more objects in addition. I’ll use numbers in the domains of my
models:
D = {0, 1}
I (F ) = {0}
I (G) = {0, 1}
In this model, everything that is in F ’s extension is also in G’s extension; so

the antecedent of our conditional, namely: ∀x(F x→Gx), is true. But since
the object 1 is in the extension of G without being in the extension of F , the
consequent of the conditional, namely: ∀x(Gx→F x), is false. So the condi-
tional is false in this model. So we’ve found a model in which the conditional
∀x(F x→Gx)→∀x(Gx→F x) is false. So it’s not true in all models. So it’s
invalid.
Exercise 4.2b We must show that 2 ∀x(F x∨∼Gx)→(∀xF x ∨∼∃xGx). The
antecedent says that everything is either an F or not a G. The consequent says
that either everything is an F , or nothing is a G. So let’s choose a model with
some things that are F and also G, and other things that are neither F nor G.
Then the antecedent will be true and the consequent will be false:
D = {0, 1}
I (F ) = {0}
I (G) = {0}
Exercise 4.2c To show that Rab does not semantically imply ∃xRx x, we
need to find a model in which the first is true and the second is false. Here is
such a model:
D = {0, 1}
I (a) = 0
I (b ) = 1
I (R) = {〈0, 1〉}
Exercise 4.2d We must show that ∀x∀y∀z[(Rxy∧Ry z)→Rx z], ∀x∃yRxy 2

∃x Rx x. So we need a model in which the premises are true and the conclusion is
false. The first premise says that R is transitive; the second says that everything
Rs something. To make the conclusion false, we must make sure in our model
that nothing Rs itself.
There is no way to satisfy all these constraints with a finite model. Suppose
you start with just a single object in the domain:
It must R something, given the second premise. It can’t R itself, since we

want the conclusion to be false. So we must posit a second thing that it Rs:
• / •
But now, given the second premise, this new thing has to R something. It
can’t R back to the first thing, because given transitivity, that first thing would
then need to R itself. So we need to add a third thing:
• / • / •
Also, given transitivity, the first thing Rs the third thing. But now this third
thing needs to R something. It can’t R itself, or any of the things earlier in the
sequence, because each of those things Rs it; so given transitivity, if the third
thing Rs any of those things, it would have to R itself.
And so on. We can never stop with any finite model, since the second
premise will always force us to add another object.
But we can have an infinite model, in which each object Rs all the later
objects:
• / • / • / • / ...
Here’s the official model:
D = {0, 1, 2, . . . } (i.e., the set of natural numbers)

I (R) = {〈i, j 〉 | i, j ∈ D, i < j } (i.e., the set of ordered pairs in which the
first member of the pair is a smaller number than the second.)
In essence, R is interpreted in this model as meaning “is less than”. The first
premise is true in the model because “is less than” is a transitive relation. The
second premise is true because for each natural number there exists a greater
natural number. The conclusion is false because no number is less than itself.
Exercise 5.1a F ab ∀x(x=a→F x b ):
i) Suppose for reductio that for some model and some g , V g (F ab ) = 1, but
…
ii) …V g (∀x(x=a→F x b )) = 0.
iii) Given ii), for some u ∈ D (call it: u) we have: V g x (x=a→F x b ) = 0

u
iv) And so we have: V g x (x=a) = 1 and V g x (F x b ) = 0 (clause for →)

u u
a) Given V g x (x=a) = 1 we have: [x] g x is (identical to) [a] g x

u u u
b) But [x] g x is just g ux (x)—that is, u

u
c) And [a] g x is just I (a)

u
d) So, we have: u is I (a)

e) Given V g x (F x b )=0, we have 〈[x] g x , [b ] g x 〉 ∈
/ I (F ) (clause for atom-
u u u
ics)
f) But [x] g x is u—i.e., I (a), as we just showed
u
g) And [b ] g x is I (b )
u
h) So we have: 〈I (a), I (b )〉 ∈
/ I (F )
v) Given line i), we have 〈[a] g , [b ] g 〉 ∈ I (F )
a) [a] g is just I (a)

b) [b ] g is just I (b )
c) so 〈I (a),I (b )〉 ∈ I (F )
d) this contradicts line iv) h)
Exercise 5.1b We are to show that ∃x∃y∃z(F x∧F y∧F z∧x6=y∧x6= z∧y6= z),
∀x(F x→(Gx∨H x) 2 ∃x∃y∃z(Gx∧Gy∧Gz∧x6=y∧x6= z∧y6= z). We need a
model in which the two premises:
∃x∃y∃z(F x∧F y∧F z ∧x6=y∧x6= z∧y6= z)

∀x(F x→(Gx∨H x)
are true, but in which the conclusion:
∃x∃y∃z(Gx∧Gy∧Gz ∧ x6=y ∧ x6= z ∧ y6= z)
is false. Let’s think about what these sentences mean. The first premise says
that “there are at least three F s”. The second premise says that “every F is
either a G or an H ”. The conclusion says that “there are at least three Gs”.
Well, if we include three objects in our model, and make each of them F s, then
the first premise is true. If we make some but not all of those objects Gs, and
make the rest H s, then the second premise will be true. But since we haven’t
made all three objects in the model Gs, the conclusion will be false. Here is
the model:
D = {0, 1, 2}
I (F ) = {0, 1, 2}
I (G) = {0, 1}
I (H ) = {2}
Exercise 5.2a “Everyone who loves someone else loves everyone”:

∀x[∃y(y6= x ∧ Lxy)→∀yLxy]
Exercise 5.2b “The only truly great player who plays in the NBA is Allen
Iverson”: ∀x[(Gx∧P xn)→x=i ]
Exercise 5.2c We must symbolize “If a person shares a solitary confinement

cell with a guard, then they are the only people in the cell”. Letting ‘C ’ stand
for ‘is a solitary confinement cell’, ‘S xy z’ stand for ‘x shares y with z’ (it’s a
three place predicate), and ‘I ’ stand for ‘is in’:
∀x∀y∀z[(P x∧C y∧Gz∧S xy z)→∀x1 ([P x1 ∧I x1 y]→[x1 =x∨x1 =z])]
Exercise 5.2d The shortest symbolization of “there are at least five dinosaurs”
I could find:
∃xD x ∧ ∼∃x1 ∃x2 ∃x3 ∃x4 (D x1 ∧D x2 ∧D x3 ∧D x4 ∧

∀y(D y → (y=x1 ∨y=x2 ∨y=x3 ∨y=x4 ))
Exercise 5.3a “The product of an even number and an odd number is an

even number.”: ∀x∀y[(E x∧O y)→E p(x, y)]
Exercise 5.3b “If the square of a number that is divisible by each

smaller number is odd, then that number is greater than all numbers.”:
∀x[(N x∧∀y(S y x→D xy)∧O s(x)) → ∀z(N z→S z x)]
Exercise 5.4a To show that ∀xF x→F f (a), let M be any model, let g be
any assignment in that model.
i) suppose for reductio that V g (∀xF x→F f (a)) = 0.
ii) Then V g (∀xF x) = 1 and V g (F f (a)) = 0
iii) Given the second, [ f (a)] g ∈

/ I (F )
iv) Given the first, for all u ∈ D,V g x (F x) = 1.

u
v) So, letting u = [ f (a)] g (which we know to be in D because all denotations

are in D), we have V g x (F x) = 1
[ f (a)] g
vi) And so, [x] g x ∈ I (F )

[ f (a)] g
a) But [x] g x is g[xf (a)] (x) …

[ f (a)] g g
b) …which is just [ f (a)] g

vii) so, [ f (a)] g ∈ I (F )—contradicts line iii)
Exercise 5.4b To show that {∀x f (x)6= x} 2 ∃x∃y( f (x)=y ∧ f (y)=x),

we must find a model in which ∀x f (x)6= x is true, but in which
∃x∃y( f (x)=y∧ f (y)=x) is false.
In any model, the interpretation of the function symbol f will some one-
place function defined on the domain. Given that ∀x f (x)6= x is to be true in
our desired model, the function assigned to f must map each object to an
object other than itself. And given that ∃x∃y( f (x)=y∧ f (y)=x) is to be false,
there can be no two objects that this function maps to each other. So let’s just
choose a model with a “triangle” of three objects, each mapped by the function
to the next vertex in the clockwise direction:
@0=
===
==
==

2 o 1
The arrows represent the function.

Here is the official model:
D = {0, 1, 2}
I ( f ) = the function g such that g (0) = 1, g (1) = 2, g (2) = 0
Exercise 5.5a We must show that ∀xL(x, ιyF xy)→∀x∃yLxy. It’s easy to
get confused by the complexity of the antecedent here, “∀xL(x, ιyF xy)”. This
just has the form: ∀xLxα , where α is “ιyF xy”. L is a two-place predicate;
it applies to the terms x and α. If you think of “F xy” as meaning that x is a
father of y, and “Lxy” as meaning that x loves y, then ∀xL(x, ιyF xy) means
“everyone x loves the y that he (x) is the father of”.
Now for the proof. Suppose for reductio that in some model, and some
assignment g in that model:
i) V g (∀xL(x, ιyF xy)→∀x∃yLxy) = 0
ii) So, V g (∀xL(x, ιyF xy)) = 1 and V g (∀x∃yLxy) = 0
iii) Given the second, for some u ∈ D,V g x (∃xLxy) = 0. Call this u “u”.
u
iv) Given the first, for every v ∈ D,V g x (L(x, ιyF xy)) = 1
v
v) Letting v = u, we have: V g x (L(x, ιyF xy)) = 1

u
vi) So, 〈[x] g x , [ιyF xy] g x 〉 ∈ I (L)

u u
a) [x] g x is g ux (x)—i.e., u.
u
b) Let’s call [ιyF xy] g x “v”. Thus, we have 〈u, v〉 ∈ I (L).

u
Aside: It doesn’t really matter for this problem, but we can in-
fer something about v. Remember that E , the emptiness
marker, is never in the extension of any predicate. That goes
for two-place predicates like L, as well as one-place predi-
cates. What that means is that for any ordered pair 〈o1 , o2 〉,
if 〈o1 , o2 〉 ∈ I (L), then neither o1 nor o2 can be E . Thus, since
〈[x] g x , [ιyF xy] g x 〉 ∈ I (L), we can conclude that [ιyF xy] g x —
u u u
i.e., v—is not E . What’s more, given the definition of deno-
tation for ι terms, there must exist exactly one object v ∈ D
such that V g xy (F xy) = 1, and that [ιyF xy] g x is this v. (If there
uv u
weren’t exactly one such v, then [ιyF xy] g x would be E .) Sum-
u
mary: we know that v is not the emptiness marker, but rather is
the one and only object in the domain such that V g xy (F xy) = 1.)
uv
vii) Thus, we have: 〈u, v〉 ∈ I (L)
viii) Now, from line iii) we have: for every o ∈ D,V g xy (Lxy) = 0
uo
ix) Letting o = v, we have: V g xy (Lxy) = 0.

uv
x) And so, 〈[x] g xy , [y] g xy 〉 ∈

/ I (L)
uv uv
xi) But [x] g xy is u and [y] g xy is v

uv uv
/ I (L), contradicting line vii).

xii) So, 〈u, v〉 ∈
Exercise 5.5b We must show that 2 GιxF x→F ιxGx. To make this formula
false in a model, we need to make GιxF x true and F ιxGx false. Let’s think
about the denotation of ιxF x. To make GιxF x true, the denotation of ιxF x
must be in the extension of G; that means that it can’t be the emptiness marker.
So let’s let the denotation of ιxF x be the number 0. Now, 0 must be the one
and only object in the extension of F (since it is not the emptiness marker and
is the denotation of ιxF x.) So we have this so far:
D = {E , 0, . . . ?
I (F ) = {0}
I (G) = {0, . . . ?
Now let’s ask: can we stop there? Can we let G’s extension just contain 0 and
nothing else? The answer is no. For if 0 is the one and only object in G’s
extension, then 0 will be the denotation of ιxGx. But since 0 is in the extension
of F , that would make F ιxGx be true, whereas we want it to be false. So we
need to add something else to G’s extension:
D = {E , 0, 1}
I (F ) = {0}
I (G) = {0, 1}
Now, the denotation of ιxGx is the emptiness marker, E . Since E is not in the
extension of F , F ιxGx is false, which is what we want.
Exercise 5.7a “If a person commits a crime, then the judge that sentences
him/her wears a wig”: ∀x[(P x∧∃y(C y∧M xy)) → ∃y(W y∧Eιz(J z∧S z x)y)]
(“E x1 x2 ” = “x1 wears x2 ”)
Exercise 5.7b “The tallest spy is a spy": Sιx(S x∧∀y((S y∧y6= x)→T xy))
Exercise 5.8 “The ten-feet-tall man is not happy”, symbolized first using
the ι, and then (under two readings) using Russell’s method:
∼H ιx(T x∧M x)
∼∃x(T x∧M x∧∀y([T y∧M y)→y=x) ∧ H x)
∃x(T x∧M x∧∀y([T y∧M y)→y=x) ∧ ∼H x)
The first Russellian symbolization says that it’s not true that: there is exactly
one ten-feet-tall man who is happy. The second says that there is exactly one
ten-feet-tall man, and he his not happy. So if there isn’t exactly one ten-feet-tall
man (whether because no man is ten-feet-tall, or because more than one man
is ten-feet-tall), then the first is true while the second is false. Given that the
null individual is not in the extension of any predicate, the ι symbolization is
also true if there is not exactly one ten-feet-tall man; so the first Russellian
symbolization is like the ι symbolization.
Exercise 5.9 The semantics of ∃prime : For any model, M , and any assignment
to the variables g , VM , g (∃prime αφ) = 1 iff |φM, g ,α | is prime.
Exercise 5.10 Symbolize “the number of people multiplied by the number

of cats that bite at least one dog is 198”, inventing any desired generalized
quantifiers:
This sentence concerns the cardinalities of two sets, the set of people and
the set of cats that bite at least one dog. So we need a binary quantifier. I’ll
invent one called Ted’s quantifier, Ted. The idea is that (Ted α : φ)ψ is to be
true iff the number of φs multiplied by the number of ψs is 198. The semantic
clause for this quantifier is this: for any model, M , and variable assignment g ,
VM ,g ((Ted α : φ)ψ) = 1 iff |φM ,g ,α | · |ψM , g ,α | = 198
The symbolization of the sentence is then:
(Ted x : P x)(C x∧∃y(D y∧B xy))
Exercise 6.2b 2(P ∨3Q)→(2P ∨3Q):
B-countermodel (hence the formula is invalid in K, D, and T as well):

∗ ∗
1 1 1 0 0 0 0 0
2(P ∨3Q)→(2P ∨3Q)
r
† ∗
0 O

0 0 1 1
a P Q P ∨3Q
†∗
0 O

1
b
Q
0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈a, r〉, 〈a, b〉, 〈b, a〉}
I (P, r) = I (Q, b) = 1, all else 0
Validity proof in S4 (establishes validity in S5 as well):
i) Suppose for reductio that in some S4 model 〈W, R,I〉, for some world
r ∈ R, V(2(P ∨3Q)→(2P ∨3Q), r ) = 0
ii) So V(2(P ∨3Q), r ) = 1 and …
iii) …V(2P ∨3Q), r ) = 0
iv) Given iii), V(2P, r ) = 0. So for some world, call it a, R ra and V(P, a) = 0
v) Given ii), V(P ∨3Q, a) = 1, and so, given 4, V(3Q, a) = 1. So there is

some world, call it b , such that Rab and V(Q, b ) = 1
vi) Since R ra and Rab , by transitivity, R r b . Given iii), V(3Q, r ) = 0.

And so V(Q, b ) = 0, contradicting v)
Exercise 6.2c 3(P ∧3Q)→(23P →32Q):

S5-countermodel (hence, the formula is invalid in all systems):
∗ ∗
1 1 1 1 0 1 1 1 0 0 0 0
3(P ∧3Q)→(23P →32Q)
r
∗ ∗ ∗ ∗
0 O

1 1 0
a Q 3P 2Q
∗ ∗
0
Official model:
W = {r, a}
R = {〈r, r〉, 〈a, a〉, 〈r, a〉, 〈a, r〉}
I (P, r) = I (Q, a) = 1; all else 0
Exercise 6.2e 2(P ∧Q)→22(3P →3Q):
T-countermodel (establishes invalidity in K and D as well):
∗
1 1 1 1 0 0
r
2(P ∧Q)→22(3P →3Q)
∗
0

1 1 1 0
a P ∧Q 2(3P →3Q)
∗
0

∗
1 1 0 0 0
3P →3Q
b
∗
0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈a, b〉}
S4-validity proof (establishes validity in S5 as well):
i) Suppose for reductio that the formula is false at some world r in some
B-model
ii) Then V(2(P ∧Q), r ) = 1 and …
iii) V(22(3P →3Q), r ) = 0
iv) Given ii), for some world, call it a, R ra and V(2(3P →3Q), a) = 0. And
so, for some world, call it b , Rab and V(3P →3Q, b ) = 0
v) Thus, V(3P, b ) = 1 and …
vi) …V(3Q, b ) = 0
vii) Since R ra and Rab , by transitivity R r b . Given i), V(P ∧Q, b ) = 1, and
so V(Q, b ) = 1.
viii) Given reflexivity, R b b , and so by vi), V(Q, b ) = 0, contradicting vii)
B-validity proof: like the S4-validity proof through the first 6 steps; then:
vii) Since Rab , by symmetry, R b a. So, given vi), V(Q, a) = 0.
viii) Since R ra, given ii), V(P ∧Q, a) = 1, and so V(Q, a) = 1, contradicting
vii)
Exercise 6.2f 2(2P →Q)→2(2P →2Q):
B-countermodel (establishes invalidity in S5 as well):

∗
1 1 1 1 1 0 0
2(2P →Q)→2(2P →2Q)
r
† ∗
0 O

∗
1 1 1 1 1 0 0
2P →Q 2P →2Q
a
† ∗
0 O

0 1
b
Q P
0
Official model:
W = {r, a, b}
i) Suppose for reductio that the formula is false at some world r of some
S4-model.
ii) Then V(2(2P →Q), r ) = 1 and…
iii) …V(2(2P →2Q), r ) = 0.
iv) Given iii), for some world a, R ra and V(2P →2Q, a) = 0.
v) By the truth condition for the →, V(2P, a) = 1 and…
vi) …V(2Q, a) = 0.
vii) Given vi), for some b , Rab and V(Q, b ) = 0.
viii) Given v), V(P, b ) = 1.

ix) Since R ra and Rab , R r b , by transitivity.

x) But then, given ii), V(2P →Q, b ) = 1.
xi) Given x) and vii), by the truth condition for the →, V(2P, b ) = 0.
xii) So, for some c, R b c and V(P, c) = 0.
xiii) Since Rab and R b c, Rac (transitivity).
xiv) So, given v), V(P, c) = 1. This contradicts xii).
Exercise 6.2h 33P →23P :

S4-countermodel (establishes invalidity in K, D, and T as well):
1 0 0
r 33P →23P
∗ ∗
0 @@
~~ @@
~~~ @@
~ @@
~~~ @@
~ @@
~~
1 1 ~ ∗
a 3P b 0 0
∗ 3P
0 0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈r, b〉}
B-countermodel:
1 0 0 0
r 33P →23P
∗ ∗
~>
0 `@@
@@
~~~ @@
~~ @@
~~~ @@
~ @@
~~
1 1 ~ ∗
a 3P b 0 0
∗ 3P
0 0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈r, b〉, 〈a, r〉, 〈b, r〉}
S5 validity proof:
i) Suppose for reductio that the formula is false at some world r in some
S5 model
ii) Then V(33P, r ) = 1 and …
iii) …V(23P, r ) = 0
iv) Given i), V(3P, a) = 1 for some a such that R ra
v) Given ii), V(3P, b ) = 0 for some b such that R r b .
vi) Given iv), V(P, c) = 1, for some c such that Rac.
vii) Since R r b , R b r by symmetry. Since R b r , R ra, and R r c, by transi-

tivity: R b c. But then by v), V(P, c) = 0, contradicting vi)
Exercise 6.2i 2[2(P →2P )→2P ]→(32P →2P ):
S4-model (establishes invalidity in K, D, and T as well):
∗
1 0 1 0 0 1 0 1 0 0
2[2(P →2P )→2P ]→(32P →2P )
r
∗ ∗ † ∗ ∗
0 Nf NN
NNN
NNN
NNN
NNN
∗ &
0 1 0
1 1 1
2(P →2P )→2P
2(P →2P )→2P 2P o
a b
∗ †
† 0
0
Official model:
W = {r, a, b}
R = {〈r, r〉, 〈a, a〉, 〈b, b〉, 〈r, a〉, 〈r, b〉, 〈b, r〉, 〈b, a〉}
I (P, r) = I (P, a) = 1, all else 0
B-model (establishes invalidity in K, D, and T as well):
∗
1 0 1 0 0 1 0 1 0 0
2[2(P →2P )→2P ]→(32P →2P )
r
∗ ∗ † ∗ ∗
0 O Nf NN
NNN
NNN
NNN
NNN
∗ &
0 1 0
1 1 1
2(P →2P )→2P
2(P →2P )→2P 2P
a b
∗ †
† 0
0
Official model:
W = {r, a, b}
I (P, r) = I (P, a) = 1, all else 0
i) Suppose for reductio that the formula is false in some world, r , of some
S5 model
ii) Then V(2[2(P →2P )→2P ], r ) = 1 and …
iii) V(32P →2P, r ) = 0
iv) Given iii), V(32P, r ) = 1, so V(2P, a) = 1 for some a such that R ra
v) Given iii), V(P, c) = 0 for some c such that R r c

vi) Since R ra, Ra r by symmetry; and then since R r c, by transitivity we

have Rac, whence V(P, c) = 1 by iv), contradicting v)
Exercise 6.4b `K 2∼P →2(P →Q):

1. ∼P →(P →Q) PL
2. 2(∼P →(P →Q)) 1, Nec
3. 2∼P →2(P →Q) 2, K
Exercise 6.4d `K 2(P ↔Q)→(2P ↔2Q):

1. (P ↔Q)→(P →Q) PL
2. 2(P ↔Q)→2(P →Q) 1, NEC, K, MP
3. 2(P →Q)→(2P →2Q) K
4. 2(P ↔Q)→(2P →2Q) 2, 3, PL
5. 2(P ↔Q)→(2Q→2P ) repeat 1-4 starting with (P ↔Q)→(Q→P )
6. 2(P ↔Q)→(2P ↔2Q) 4, 5, PL
Exercise 6.4e `K [2(P →Q)∧2(P →∼Q)]→∼3P

1. (P →Q)→[(P →∼Q)→∼P ] PL
2. 2(P →Q)→2[(P →∼Q)→∼P ] 1, NEC, K, MP
3. 2[(P →∼Q)→∼P ]→[2(P →∼Q)→2∼P ] K
4. 2(P →Q)→[2(P →∼Q)→2∼P ] 2, 3, PL
5. ∼3P ↔2∼P MN
6. [2(P →Q)∧2(P →∼Q)]→∼3P 4,5 PL
Exercise 6.4f `K (2P ∧2Q)→2(P ↔Q):

1. P →(Q→(P ↔Q)) PL
2. 2[P →(Q→(P ↔Q))] 1, Nec
3. 2P →2(Q→(P ↔Q)) 2, K, MP
4. 2(Q→(P ↔Q))→(2Q→2(P ↔Q)) K
5. 2P →(2Q→(2(P ↔Q)) 3, 4, PL
6. (2P ∧2Q)→2(P ↔Q) 5, PL
Exercise 6.4h `K 3P →(2Q→3Q):
1. Q→(P →Q) PL
2. 2Q→2(P →Q) 1, NEC, K, MP
3. 2(P →Q)→(3P →3Q) K3
4. 3P →(2Q→3Q) 2, 3, PL
Exercise 6.5a ` D ∼(2P ∧2∼P ):
1. 2P →3P D
2. 2P →∼2∼P rewrite of 1 given def of 3
3. ∼(2P ∧2∼P ) 2, PL
Exercise 6.5b `D ∼2[2(P ∧Q)∧2(P →∼Q)]:
1. (P ∧Q)→∼(P →∼Q) PL
2. 2(P ∧Q)→2∼(P →∼Q) 1, NEC, K, MP
3. 2∼(P →∼Q)→3∼(P →∼Q) D
4. 3∼(P →∼Q)↔∼2(P →∼Q) MN
5. ∼[2(P ∧Q)∧2(P →∼Q)] 2, 3, 4, PL
6. 2∼[2(P ∧Q)∧2(P →∼Q)] 6, NEC
7. 3∼[2(P ∧Q)∧2(P →∼Q)] D, 6, MP
8. 3∼[2(P ∧Q)∧2(P →∼Q)]↔ MN
∼2[2(P ∧Q)∧2(P →∼Q)]
9. ∼2[2(P ∧Q)∧2(P →∼Q)] 7, 8, PL
Exercise 6.6a `T 32P →3(P ∨Q):
1. 2P →P T
2. 2P →(P ∨Q) 1, PL
3. 32P →3(P ∨Q) 2, NEC, K3, MP
Exercise 6.6b `T [2P ∧32(P →Q)]→3Q:

1. 2(P →Q)→(P →Q) T

2. P →(2(P →Q)→Q) 1, PL
3. 2P →2(2(P →Q)→Q) 2, NEC, K, MP
4. 2(2(P →Q)→Q)→(32(P →Q)→3Q) K3
5. 2P →(32(P →Q)→3Q) 3, 4, PL
6. [2P ∧32(P →Q)]→3Q 5, PL
Exercise 6.6c `T 3(P →2Q)→(2P →3Q):

1. 2Q→Q T
2. P →[(P →2Q)→Q] 1, PL
3. 2P →2[(P →2Q)→Q] 2, NEC, K, MP
4. 2[(P →2Q)→Q]→[3(P →2Q)→3Q] K3
5. 3(P →2Q)→(2P →3Q) 3, 4, PL
Exercise 6.7a `B 32P ↔3232P :

1. 2P →232P B3
2. 32P →3232P 1, Nec, K3, MP
3. 3232P →32P B
4. 32P ↔3232P 2, 3 PL
Exercise 6.7b `B [2P ∧232(P →Q)]→2Q:

1. 32(P →Q)→(P →Q) B
2. 232(P →Q)→2(P →Q) 1, NEC, K, MP
3. 2(P →Q)→(2P →2Q) K
4. [2P ∧232(P →Q)]→2Q 2, 3, PL
Exercise 6.8a `S4 2P →232P :

1. 2P →32P T3
2. 22P →232P 1, NEC, K, MP
3. 2P →22P S4
4. 2P →232P 2, 3, PL
Exercise 6.8b `S4 2323P →23P :
1. 23P →3P T
2. 323P →33P 1, NEC, K3, MP
3. 33P →3P S43
4. 323P →3P 2, 3, PL
5. 2323P →23P 4, NEC, K, MP
This is also provable in B:
1. 323P →3P B
2. 2323P →23P 1, NEC, K, MP
Exercise 6.8c `S4 32P →3232P :
1. 2P →32P T3
2. 22P →232P 1, NEC, K, MP
3. 2P →22P S4
4. 2P →232P 2, 3, PL
5. 32P →3232P 4, NEC, K3, MP
Exercise 6.9a We are to show that `S5 (2P ∨3Q)↔2(P ∨3Q). My strat-
egy for the first half, in lines 1-7, uses the fact from propositional logic that
(φ∨ψ)→χ follows from φ→χ and ψ→χ . My strategy for the second half uses
MN, plus the fact from propositional logic that χ →(φ∨ψ) is equivalent to
χ →(∼ψ→φ):
1. P →(P ∨3Q) PL
2. 2P →2(P ∨3Q) 1, Nec, K, MP
3. 3Q→(P ∨3Q) PL
4. 23Q→2(P ∨3Q) 3, Nec, K, MP
5. 3Q→23Q S53
6. 3Q→2(P ∨3Q) 4, 5, PL
7. (2P ∨3Q)→2(P ∨3Q) 2, 6, PL (done left-to-right; now
for the other direction. Goal: get
2(P ∨3Q)→(2∼Q→2P ))
8. 2∼Q↔∼3Q MN
9. (P ∨3Q)→(2∼Q→P ) 8, PL
10. 2(P ∨3Q)→2(2∼Q→P ) 9, NEC, K, MP
11. 2(2∼Q→P )→(22∼Q→2P ) K
12. 2∼Q→22∼Q S4
13. 2(P ∨3Q)→(2∼Q→2P ) 10, 11, 12, PL
14. 2(P ∨3Q)→(2P ∨3Q) 13, 8, PL
15. (2P ∨3Q)↔2(P ∨3Q) 7, 14, PL
Exercise 6.9b `S5 3(P ∧3Q)↔(3P ∧3Q):

1. (P ∧3Q)→P PL
2. 3(P ∧3Q)→3P 1, NEC, K3, MP
3. (P ∧3Q)→3Q PL
4. 3(P ∧3Q)→33Q 3, NEC, K3, MP
5. 33Q→3Q S43
6. 3(P ∧3Q)→(3P ∧3Q) 2, 4, 5, PL
7. 3Q→(P →(P ∧3Q)) PL
8. 23Q→2(P →(P ∧3Q)) 7, NEC, K, MP
9. 2(P →(P ∧3Q))→(3P →3(P ∧3Q)) K3
10. 3Q→23Q S53
11. 3Q→(3P →3(P ∧3Q)) 8, 9, 10, PL
12. (3P ∧3Q)→3(P ∧3Q) 11, PL
13. 3(P ∧3Q)↔(3P ∧3Q) 6, 12, PL
Exercise 6.9c We must show that `S5 2(2P →2Q)∨2(2Q→2P ). Plan

of attack: show the PL equivalent: ∼2(2P →2Q)→2(2Q→2P ). Note the
equivalence of the antecedent of this conditional, via MN, with 3∼(2P →2Q).
1. ∼(2P →2Q)→2P PL
2. 3∼(2P →2Q)→32P 1, NEC, K3, MP
3. 32P →2P S5
4. 2P →22P S4
5. 2P →(2Q→2P ) PL
6. 22P →2(2Q→2P ) 5, NEC, K, MP
7. 3∼(2P →2Q)↔∼2(2P →2Q) MN
8. ∼2(2P →2Q)→2(2Q→2P ) 7, 2, 3, 4, 6 PL
9. 2(2P →2Q)∨2(2Q→2P ) 8, PL
Exercise 6.9d `S5 2[2(3P →Q)↔2(P →2Q)]:
1. 2(3P →Q)→(23P →2Q) K

2. P →23P B
3. 2(3P →Q)→(P →2Q) 1, 2, PL
4. 22(3P →Q)→2(P →2Q) 3, NEC, K
5. 2(3P →Q)→22(3P →Q) S4
6. 2(3P →Q)→2(P →2Q) 4, 5, PL
7. 2(P →2Q)→(3P →32Q) K3 (now the other direction)
8. 32Q→Q B
9. 2(P →2Q)→(3P →Q) 7, 8, PL
10. 22(P →2Q)→2(3P →Q) 9, NEC, K, MP
11. 2(P →2Q)→22(P →2Q) S4
12. 2(P →2Q)→2(3P →Q) 10, 11, PL
13. 2(3P →Q)↔2(P →2Q) 12, 6, PL
14. 2[2(3P →Q)↔2(P →2Q)] 13, NEC
Exercise 6.10 Lemma 6.4c is a trivial consequence of lemma 6.4b. As for

lemma 6.4d: since Γ is maximal, either φ or ∼φ is a member of Γ. But ∼φ
can’t be a member of Γ; otherwise, since Γ would contain the S-inconsistent

subset {∼φ} (this subset is S-inconsistent because `S φ). So φ ∈ Γ.
Exercise 6.11 Where S is any normal modal system, we must show that if
∆ is an S-consistent set of wffs containing the formula 3φ, then 2− (∆) ∪ φ is
also S-consistent.
3φ is an abbreviation of ∼2∼φ; so what we’re given is this: S is a normal
modal system, ∆ is an S-consistent set of wffs containing the formula ∼2∼φ.
By Lemma 6.6, 2− (∆) ∪ {∼∼φ} is S-consistent. Now suppose for reductio
that 2− (∆) ∪ {φ} is not S-consistent. So given the definition of S-consistency,
for some ψ1 . . . ψn in 2− (∆) ∪ {φ}, `S ∼(ψ1 ∧ · · · ∧ψn ). Since S includes PL,
`S ∼(ψ1 ∧ · · · ∧ψn ∧φ). If φ is one of the ψs, then the rest of the ψs are members
of 2− (∆); so for some δ1 . . . δ m in 2− (∆) (namely, all the ψs other than φ),
`S ∼(δ1 ∧ · · · ∧δ m ∧φ). Since S includes PL, we have: `S ∼(δ1 ∧ · · · ∧δ m ∧∼∼φ),
which violates the S-consistency of 2− (∆) ∪ {∼∼φ}.
Exercise 6.12 We are to demonstrate completeness for the system that results
from adding to K every axiom of the form 3φ→2φ, where the frames for this
system are defined as those whose accessibility relation meets the condition
that every world can see at most one world. Let’s first show that
(*) in the canonical model for the strange system, every world sees at most
one world.
To do this, suppose for reductio that for some world, w, in this canonical
model, R wv and R wv 0 and v6=v 0 . Now, since v6=v 0 , and v and v 0 are maximal
consistent sets of sentences, there must be some sentence, φ, that is a member
of one set but not the other. Without loss of generality, suppose that φ ∈ v and
φ∈ / v 0 . Then, by theorem 6.7, V(φ, v) = 1 and V(φ, v 0 ) = 0. Since R wv and
R wv 0 , that means that V(3φ, w) = 1 (since φ is true in some world accessible
from w) and V(2φ, w) = 0 (since φ isn’t true at all worlds accessible from w).
So V(3φ→2φ, w) = 0. But 3φ→2φ is a theorem of the strange system, and
so by 6.4d is a member of w, and so by theorem 6.7 is true at w. Contradiction.
Now we use (*) to prove completeness for the strange system. Suppose φ
is valid in the strange system. That is, φ is true in any world of any model in
which the accessibility relation is such that every world sees at most one world.
Given (*), the canonical model for the strange system is such a model. So φ is
true at every world in the canonical model—i.e., is valid in the canonical model
for this system. By corollary 6.8, φ is a theorem of the strange system.
Exercise 7.1 We’re to show that φ I ψ iff I φ→ψ. For the left-to-right
direction, suppose φ ψ, and suppose for reductio that V(φ→ψ, s) = 0. Then
for some s 0 (that s sees), V(φ, s 0 ) = 1 and V(ψ, s 0 ) = 0—contradicts φ ψ.
For the other direction, suppose φ→ψ, and suppose for reductio that
V(φ, s) = 1 while V(ψ, s ) = 0. By φ→ψ, V(φ→ψ, s ) = 1. By reflexivity, either
V(φ, s ) = 0 or V(ψ, s) = 1. Contradiction.
Exercise 7.3a We’re to show that ∼(P ∧Q) 2 (∼P ∨∼Q):
∗
1 0 0 0 0 0
r
∼(P ∧Q) ∼P ∨∼Q
∗ ∗
0 JJ
tt JJ
t tt JJ
JJ
tt JJ
tt J%
zt
t
∗ ∗
a 1 0 0 b 1 0 0
P P ∧Q Q P ∧Q
0 0
Exercise 7.3b We’re to show that ∼P ∨∼Q ∼(P ∧Q). Suppose

V(∼P ∨∼Q, s) = 1 and V(∼(P ∧Q), s) = 0. Given the latter, for some s 0 , R s s 0
and V(P ∧Q, s 0 ) = 1. So, V(P, s 0 ) = 1 and V(Q, s 0 ) = 1. Given the former, either
V(∼P, s ) = 1 or V(∼Q, s ) = 1. If the former then V(P, s 0 ) = 0; if the latter
V(Q, s 0 ) = 0. Contradiction either way.
Exercise 7.3c We’re to show that P →(Q∨R) 2 (P →Q)∨(P →R). We begin

thus:
∗
1 0 0 0
r
P →(Q∨R) (P →Q)∨(P →R)
∗ ∗
0
We now discharge the bottom-asterisks:

∗
1 0 0 0
r
P →(Q∨R) (P →Q)∨(P →R)
∗ ∗
0 rr LLL
rrrrr LLL
LLL
r
rrrrr LLL
LLL
∗ x r
r LL
&
a 1 0 b P R
0
P Q
0
Exercise 7.4 We are assuming the inductive hypothesis (ih) that heredity
holds for formulas φ and ψ, and we must show that heredity then must also
hold for ∼φ, φ→ψ, and φ∨ψ.
∼ : Suppose for reductio that V(∼φ, s) = 1, R s s 0 , and V(∼φ, s 0 ) = 0. Given
the latter, for some s 00 , R s 0 s 00 and V(φ, s 00 ) = 1. By transitivity, R s s 00 . This
contradicts V(∼φ, s) = 1.
→ : Suppose for reductio that V(φ→ψ, s ) = 1, R s s 0 , and V(φ→ψ, s 0 ) = 0.
Given the latter, for some s 00 , R s 0 s 00 and V(φ, s 00 ) = 1 and V(ψ, s 00 ) = 0; but by
transitivity, R s s 00 —contradicts the fact that V(φ→ψ, s) = 1.
∨ : Suppose for reductio that V(φ∨ψ, s ) = 1, R s s 0 , and V(φ∨ψ, s 0 ) = 0.
Given the former, either V(φ, s) = 1 or V(ψ, s ) = 1; and so, given (ih), either φ
or ψ is 1 in s 0 . That violates V(φ∨ψ, s 0 ) = 0.
Exercise 7.5 We must show that ∧I, ∨I, DNI, RAA, →I, →E, and EF pre-
serve I-validity.
∧E: Assume that Γ ` φ∧ψ is I-valid, and suppose for reductio that V(Γ, s ) =
1 and V(φ, s) = 0, for some stage s in some model. By the I-validity of Γ ` φ∧ψ,
V(φ∧ψ, s) = 1, so V(φ, s) = 1. Contradiction. The case of ψ is parallel.
∨I: Assume that Γ ` φ is I-valid and suppose for reductio that V(Γ, s) = 1
but V(φ∨ψ, s) = 0. Thus V(φ, s) = 0—contradiction.
DNI: if Γ ` φ then Γ ` ∼∼φ): assume Γ ` φ is I-valid, and suppose for
reductio that V(Γ, s) = 1 but V(∼∼φ, s) = 1. From the latter, for some s 0 , R s s 0
and V(∼φ, s 0 ) = 1. So V(φ, s 0 ) = 0 (since R is reflexive). From the former and
the fact that Γ ` φ, V(φ, s) = 1. This violates general heredity.
RAA: Suppose that Γ,φ ` ψ∧∼ψ is I-valid, and suppose for reductio that
V(Γ, s) = 1 but V(∼φ, s ) = 0. Then for some s 0 , R s s 0 and V(φ, s 0 ) = 1. Since
V(Γ, s) = 1 (i.e., all members of Γ are 1 at s), by general heredity V(Γ, s 0 ) = 1 (all
members of Γ are 1 at s’). Thus, since Γ,φ ` ψ∧∼ψ is I-valid, V(ψ∧∼ψ, s 0 ) = 1.
But that is impossible. (If V(ψ∧∼ψ, s 0 ) = 1, then V(ψ, s 0 ) = 1 and V(∼ψ, s 0 ) = 1;
but from the latter and the reflexivity of R it follows that V(ψ, s 0 ) = 0.)
→I: Suppose that Γ,φ ` ψ is I-valid, and suppose for reductio that V(Γ, s ) = 1
but V(φ→ψ, s) = 0. Given the latter, for some s 0 , R s s 0 and V(φ, s 0 ) = 1 and
V(ψ, s 0 ) = 0. Given general heredity, V(Γ, s 0 ) = 1. And so, given that Γ,φ ` ψ is
I-valid, V(ψ, s 0 ) = 1—contradiction.
→E: Suppose that Γ ` φ and ∆ ` φ→ψ are both I-valid, and suppose for
reductio that V(Γ ∪ ∆, s) = 1 but V(ψ, s) = 0. Since Γ ` φ and ∆ ` φ→ψ,
V(φ, s ) = 1 and V(φ→ψ, s) = 1. Given the latter, and given that R is reflexive,
either V(φ, s) = 0 or V(ψ, s) = 1. Contradiction.
EF (ex falso): Suppose Γ ` φ∧∼φ is I-valid, and suppose for reductio that
V(Γ, s) = 1 but V(ψ, s ) = 0. Given the former and the I-validity of Γ ` φ∧∼φ,
V(φ∧∼φ, s) = 1, which is impossible.
Exercise 8.1a We’re to show that φ⇒ψ φ2→ψ:
i) Suppose φ⇒ψ is true at r , and suppose for reductio that φ2→ψ is false
at r .
ii) Then there’s a nearest-to-r φ world, a, at which ψ is false.
iii) But that can’t be. “φ⇒ψ” means 2(φ→ψ). So φ→ψ is true at every
world. So there can’t be a world like a, in which φ is true and ψ is false.
Exercise 8.1b We’re to show that φ2→ψ φ→ψ:
i) Suppose φ2→ψ is true at some world r in some SC-model, and …
ii) …suppose for reductio that φ→ψ isn’t true there.
iii) Then φ is true at r and …
iv) …ψ is false at r
v) Given “base”, for every world, x, r r x.
vi) Given iii) and v), r is a closest-to-r φ world. So, given i), ψ is true at r .
Contradicts iv).
Exercise 8.2a SC 3P →[(P 2→Q)↔∼(P 2→∼Q)]:
i) Suppose for reductio that V(3P →[(P 2→Q)↔∼(P 2→∼Q)], w) = 0,

for some world w in some Stalnaker model.
ii) Then V(3P, w) = 0, and so P is true at some world; and …
iii) …V[(P 2→Q)↔∼(P 2→∼Q)], w) = 0. So P 2→Q and ∼(P 2→∼Q)

have different truth values at w—either the first is true and the second is
false, or the first is false and the second is true.
iv) Suppose first that V(P 2→Q, w) = 1 and …
v) …V(∼(P 2→∼Q), w) = 0.
vi) Given ii) and the limit assumption, there is some nearest-to-w P -world,
call it v.
vii) Given iv), Q is true in every nearest-to-w P -world; thus, V(Q, v) = 1.
viii) Given v), V(P 2→∼Q, w) = 1; and so V(∼Q, v) = 1, and so V(Q, v) = 0,

contradicting vii).
ix) Suppose second that V(∼(P 2→∼Q), w) = 1 and…
x) …V(P 2→Q, w) = 0
xi) Given x), there is some nearest-to-w P -world, call it v 0 , such that
V(Q, v 0 ) = 0.
xii) Given ix), V(P 2→∼Q, w = 0). So there’s some nearest-to-w P -world,
call it v 00 , such that V(∼Q, v 00 ) = 0.
xiii) To say that v 0 is a “nearest-to-w” P -world (line xi)) is to say: V(P, v 0 ) = 1

and for every x, if V(P, x) = 1 then v 0 w x. So, since V(P, v 00 ) = 1 (line
xii)), we have v 0 w v 00 . Likewise, since v 00 is a nearest-to-w P -world and
P is true at v 0 , we know that v 00 w v 0 .
xiv) From xiii), by antisymmetry, v 0 = v 00 . And so by xii), V(∼Q, v 0 ) = 1, and

so V(Q, v 0 ) = 1, contradicting xi)
(Note that this wff is invalid given Lewis’s semantics, as the countermodel from
section 8.7 shows.
Exercise 8.2b 2SC [P 2→(Q→R)]→[(P ∧Q)2→R]:
/.1 1 1 -,
()P ∧Q R*+
0
b
O
/.1 -,
()P Q→R*+
0 1
O
a no P ∧Q
no P
/. -,
()[P 2→(Q→R)]→[(P ∧Q)2→ R]*+
1 0 0
r
Official model:
W = {r, a, b}
r = {〈a, b〉 . . . }
I (P, b) = I (Q, b) = I (P, a) = 1; all else 0
Exercise 8.2c 2SC [P 2→(Q2→R)]→[Q2→(P 2→R)]:

/. -,
d () *+
/. -,
c () *+
/.1 -,
()Q P 2→R*+
0
b
O
“view from r”:
/.1 -,
()P Q2→R*+
1
O
a no Q
no P
/. -,
()[P 2→(Q2→ R)]→[Q2→(P 2→ R)]*+
1 0 0
r
/. -,
d () *+
/. -,
r () *+
/. -,
b () *+
“view from a”:
/.1 -,
()Q R*+
1
c
O
no Q
/.1 -,
()P Q2→R*+
1
a
/. -,
c () *+
/. -,
r () *+
/. -,
a () *+
“view from b”:
/.1 -,
()P R*+
0
O d
no P
/.1 -,
()Q P 2→R*+
0
b
Official model:
W = {r, a, b, c, d}
r = {〈a, b〉, 〈b, c〉, 〈c, d〉 . . . }
a = {〈c, b〉, 〈b, r〉, 〈r, d〉 . . . }
b = {〈d, a〉, 〈a, r〉, 〈r, c〉 . . . }
I (P, a) = I (Q, b) = I (Q, c) = I (R, c) = I (P, d) = 1, all else 0
Exercise 8.3 We must show that in Lewis models where the limit and anti-
symmetry conditions hold, Lewis’s truth conditions reduce to Stalnaker’s. Con-
sider any Lewis model 〈W , , I 〉 in which the limit and anti-symmetry con-
ditions hold. Let VL and VS be the Lewis and Stalnaker valuation functions,
respectively, for this model. We must show that these are the same functions.
We’ll show by induction that for any wff φ: for any world w, VS (φ2→ψ) =
VS (φ2→ψ). Base case: show that VS and VL assign the same truth values to
sentence letters at each world. This follows from the fact that both functions
by definition assign the truth values I (α, w) for sentence letters α.
Induction step: assuming the inductive hypothesis:
(ih) VS and VL assign the same truth values at each world to wffs φ and ψ
we must show that VS and VL also assign the same truth values at each world
to ∼φ, φ→ψ, 2φ, and φ2→ψ. This is easy in the first three cases, since i)
the clauses in the definitions of VS and VL for the ∼, →, and 2 are identical,
and define the truth values of the complex formulas ∼φ, φ→ψ, and 2φ at a
given world as a function of the truth values of φ and ψ at that world and other
worlds; and ii) (ih) tells us that φ and ψ have the same VS and VL values at all
worlds.
It remains to show that, for a given world w, VS (φ2→ψ) = VL (φ2→ψ).
Given Stalnaker’s truth conditions, we know that VS (φ2→ψ, w) = 1 iff:
(S) for any x, IF [VS (φ, x) = 1 and for any y such that VS (φ, y) = 1, x w y]
THEN VS (ψ, x) = 1
And given Lewis’s truth conditions, we know that VL (φ2→ψ, w) = 1 iff:
(L) EITHER φ is trueL at no worlds, OR: there is some world, x, such that
VL (φ, x) = 1 and for all y, if y w x then VL (φ→ψ, y) = 1
So what we must show is that (S) holds iff (L) holds.

First: (S) ⇒ (L):
i) Suppose (S) is true
ii) Suppose for reductio that (L) isn’t true. Then each disjunct of (L) is false;
so:
iii) φ is trueL at some world, and …
iv) …NOT: “there is some world, x, such that VL (φ, x) = 1 and for all y, if
y w x then VL (φ→ψ, y) = 1”. So: for every world, x, if VL (φ, x) = 1
then for some y, y w x and VL (φ→ψ, y) = 0
v) Given iii) and the limit condition (which we are assuming holds in this
model), there is some world, call it a, that is a nearest-to-w world in
which φ is trueL .
vi) That is, VL (φ, a) = 1 and …

vii) …for all y, if VL (φ, y) = 1 then a w y
viii) Given (ih) and vi), VS (φ, a) = 1
ix) Given (ih) and vii), for all y, if VS (φ, y) = 1 then a w y. (It’s crucial to
the success of this step that (ih) tells us that φ has the same value under
VS and VL at all worlds.)
x) Given i), viii), and ix), VS (ψ, a) = 1. Given (ih), VL (ψ, a) = 1
xi) Given iv) and vi), there is some world, call it b , such that b w a and …
xii) …VL (φ→ψ, b ) = 0
xiii) From xii), VL (φ, b ) = 1. So by vii), a w b . So, given xi), by antisymmetry

(assumed to hold in this model), a = b , and so, given x), VL (ψ, b ) = 1.
This contradicts xii).
Now: (L) ⇒ (S):
i) Suppose (L) is true
ii) Suppose for reductio that (S) isn’t true. So for some world, call it a,
VS (φ, a) = 1 and…
iii) …for any y such that VS (φ, y) = 1, a w y; and …
iv) VS (ψ, a) = 0. So by (ih), VL (ψ, a) = 0
v) Given ii), by (ih) VL (φ, a) = 1.
vi) v) tells us that φ is trueL at some world. So by i), there is some world,
call it b , such that VL (φ, b ) = 1 and …
vii) …for all y, if y w b then VL (φ→ψ, y) = 1
viii) Given v) and iv), VL (φ→ψ, a) = 0. So from vii), a w b . But given vi)
and the ih, VS (φ, b ) = 1; and so, given iii), a w b . Contradiction.
Exercise 9.1a SQML 3∀xF x→∃x3F x:

i) Suppose for reductio that in some QML model 〈W , D, I 〉, for some

w ∈ W , and for some variable assignment g based on this model,
V g (3∀xF x→∃x3F x, w) = 0
ii) Then V g (3∀xF x, w) = 1 and…
iii) …V g (∃x3F x, w) = 0
iv) From ii), for some w 0 ∈ W ,V g (∀xF x, w 0 ) = 1. So for every u ∈ D,

V g x (F x, w 0 ) = 1
u
v) From iii), for every u ∈ D,V g x (3F x, w) = 0

u
vi) The definition of a QML model specifies that the domain D cannot be
empty. So, D has at least one member; call some member of D “u 0 ”.
vii) From iv), V g x0 (F x, w 0 ) = 1

u
viii) From v), V g x0 (3F x, w) = 0, from which it follows that for every member
u
of W , and so for w 0 in particular, V g x0 (F x, w 0 ) = 0, contradicting vii)
u
Exercise 9.1b 2SQML ∃x3Rax→32∃x∃yRxy:

∗
1 0 0 0 1 1
∃ x3Rax→32∃x∃yRxy 3Ra ux
+
r
∗ ∗
D = {u}
R : {〈u, u〉}
I (a) = u
+ +
c 0 0 0
∃ x∃yRxy ∃ yR ux y R ux uy
Official model:
W = {r, c}
D = {u}
I (a) = u
I (R) = {〈u, u, r〉}
Exercise 9.1c We are to determine whether the formula

∃x(N x∧∀y(N y→y=x)∧2O x)→2∃x(N x∧∀y(N y→y=x)∧O x) is valid.
Think of N as meaning “numbers the planets”, and think of O as meaning “is
odd”. The sentence then says: “If the entity which is in fact the number of the
planets is necessarily-odd, then it’s necessary that: the number of the planets is
odd”. It is, therefore, intuitively invalid. I won’t write out the full process of
constructing the model (with overstars, etc.); I’ll just go directly to a picture of
a model. The model will have one world in which one and only one object, u,
numbers the planets; and we will make u be odd in every world in the model.
The model will also contain a second world in which one and only one object,
v, numbers the planets; but v won’t be odd in this second world:
r N : {u} O : {u} W = {r, a}

D = {u, v}
D : {u, v}
I (N ) = {〈u, r〉, 〈v, a〉}
a N : {v} O : {u} I (O) = {〈u, r〉, 〈u, a〉}
Exercise 9.2 Formulas 9.1b and 9.1c are SQML-invalid, and so remain
invalid in the variable domain semantics. But whereas 9.1a is SQML-valid, it
is VDQML-invalid:
+ ∗
1 0 0 0 0
3∀xF x→ ∃ x3F x 3F vx
r
∗ Official model:
Dr : {v} F : {} W = {r, a}
Dr = {v}
Da = {u}
+
I (F ) = {〈u, a〉}
1 1 0
u
a ∀ xF x F x
F vx
Da : {u} F : {u}
Exercise 9.3a 2VDQML 2∀xF x→∀x2F x:
Official model:
r Dr : {u, v} F : {u, v}
0
W = {r, a} R = {〈r, r〉, 〈r, a〉, 〈a, a〉}
D = {u, v} Dr = {u, v} Da = {v}
a Da : {v} F : {v} I (F ) = {〈u, r〉, 〈v, r〉, 〈v, a〉}
0
Exercise 9.3b 2VDQML ∃x2F x→2∃xF x:
Official model:
r Dr : {u, v} F : {u}
0
W = {r, a} R = {〈r, r〉, 〈r, a〉, 〈a, a〉}
D = {u, v} Dr = {u, v} Da = {v}
a Da : {v} F : {u} I (F ) = {〈u, r〉, 〈u, a〉}
0
Exercise 9.3c The model in exercise 9.3a shows that 2VDQML ∀x2∃y y=x.
Exercise 9.4a 2VDQML+ID 2∀αφ→∀α2φ:
i) suppose for reductio that Vg (2∀αφ, w) = 1, and …

ii) …Vg (∀α2φ, w) = 0
iii) so for some u ∈ Dw , Vg α (2φ, w) = 0

u
iv) so for some v, R wv and Vg α (φ, v) = 0

u
v) by i), Vg (∀αφ, v) = 1
vi) by increasing domains, u ∈ Dv
vii) by v), for every object in Dv , and so for u in particular, Vg α (φ, v) = 1.

u
Contradicts iv)
Exercise 9.4b 2VDQML+ID ∃α2φ→2∃αφ:
i) suppose for reductio that Vg (∃α2φ→2∃αφ, w) = 0, Then

Vg (∃α2φ, w) = 1 and …
ii) …Vg (2∃αφ, w) = 0
iii) by i), for some u ∈ Dw , Vgu/α (2φ, w) = 1
iv) by ii), for some world v, R wv and Vg (∃αφ, v) = 0
v) by the increasing domain requirement, Dw ⊆ Dv , and so u ∈ Dv
vi) by iv), for every object in Dv , and so for u in particular, Vg α (φ, v) = 0

u
vii) by iii), Vg α (φ, v) = 1. Contradicts vi)

u
Exercise 10.1a We are to show that φ→2@φ. Consider any model, world
w (and variable assignment), and suppose for reductio that V(φ, w, w) = 1
but V(2@φ, w, w) = 0. Given the latter, there is some world, v, such that
V(@φ, w, v) = 0. And so, given the truth condition for @, V(φ, w, w) = 0.
Contradiction.
Exercise 10.1b We are to show that 2×∀x3@F x→2∀xF x.
i) Suppose for reductio that for some world w, some variable assignment
g , and some model, Vg (2×∀x3@F x, w, w) = 1 and …
ii) …Vg (2∀xF x, w, w) = 0.
iii) Given the latter, for some world, call it “a”, Vg (∀xF x, w, a) = 0.
iv) And so for some u ∈ D (call it “u”), Vg x (F x, w, a) = 0.

u
v) Given i), Vg (×∀x3@F x, w, a) = 1
vi) Given the truth condition for ×, Vg (∀x3@F x, a, a) = 1
vii) Thus, for every object in the domain, and so for u in particular,
Vg x (3@F x, a, a) = 1
u
viii) Thus, for some world, call it b , Vg x (@F x, a, b ) = 1

u
ix) Given the truth condition for @, Vg x (F x, a, a) = 1

u
x) Given the truth condition for atomics, 〈[x] g x , a〉 ∈ I (F )

u
/ I (F ). Contradiction
xi) But given iv), 〈[x] g x , a〉 ∈
u
Exercise 10.2 We must show that 2D F@φ→φ. Suppose V(F@φ, w, w) = 1

but V(φ, w, w) = 0. Given the former, V(@φ, w, w) = 1 (given the truth
condition for ‘F’); but then V(φ, w, w) = 1 (given the truth condition for @).
Contradiction.
Exercise 10.3 We must find some φ such that 22D φ→Fφ. In example 10.5
model of the previous problem, the formula @Ga→F@Ga is false at 〈c, c〉. The
antecedent is true because the referent of ‘a’ is in the extension of ‘G’ at c. The
consequent is false because F@Ga means that ‘Ga’ is true at all pairs of the
form 〈v, v〉, whereas ‘Ga’ is not true at 〈d, d〉 (since the referent of ‘a’ is not in
the extension of ‘G’ at d).
Bibliography
Benacerraf, Paul and Hilary Putnam (eds.) (1983). Philosophy of Mathematics.

2nd edition. Cambridge: Cambridge University Press.
Boolos, George (1975). “On Second-Order Logic.” Journal of Philosophy 72:

509–527. Reprinted in Boolos 1998: 37–53.
— (1984). “To Be Is to Be the Value of a Variable (or to Be Some Values of

Some Variables).” Journal of Philosophy 81: 430–49. Reprinted in Boolos 1998:
54–72.
— (1985). “Nominalist Platonism.” Philosophical Review 94: 327–44. Reprinted

in Boolos 1998: 73–87.
— (1998). Logic, Logic, and Logic. Cambridge, MA: Harvard University Press.
Boolos, George and Richard Jeffrey (1989). Computability and Logic. 3rd edition.
Cambridge: Cambridge University Press.
Chalmers, David (1996). The Conscious Mind. Oxford: Oxford University Press.
Cresswell, M. J. (1990). Entities and Indices. Dordrecht: Kluwer.
Cresswell, M.J. and G.E. Hughes (1996). A New Introduction to Modal Logic.
London: Routledge.
Davies, Martin and Lloyd Humberstone (1980). “Two Notions of Necessity.”

Philosophical Studies 38: 1–30.
Dummett, Michael (1973). “The Philosophical Basis of Intuitionist Logic.” In

H. E. Rose and J. C. Shepherdson (eds.), Proceedings of the Logic Colloquium,
Bristol, July 1973, 5–49. Amsterdam: North-Holland. Reprinted in Benacerraf
and Putnam 1983: 97–129.
350
BIBLIOGRAPHY 351
Enderton, Herbert (1977). Elements of Set Theory. New York: Academic Press.
Evans, Gareth (1979). “Reference and Contingency.” The Monist 62: 161–189.
Reprinted in Evans 1985.
— (1985). Collected Papers. Oxford: Clarendon Press.
Fine, Kit (1985). “Plantinga on the Reduction of Possibilist Discourse.” In

J. Tomberlin and Peter van Inwagen (eds.), Alvin Plantinga, 145–186. Dor-
drecht: D. Reidel.
Gamut, L. T. F. (1991a). Logic, Language, and Meaning, Volume 1: Introduction

to Logic. Chicago: University of Chicago Press.
— (1991b). Logic, Language, and Meaning, Volume 2: Intensional Logic and Logical
Grammar. Chicago: University of Chicago Press.
Gibbard, Allan (1975). “Contingent Identity.” Journal of Philosophical Logic 4:

187–221. Reprinted in Rea 1997: 93–125.
Glanzberg, Michael (2006). “Quantifiers.” In Ernest Lepore and Barry C.

Smith (eds.), The Oxford Handbook of Philosophy of Language, 794–821. Oxford
University Press.
Harper, William L., Robert Stalnaker and Glenn Pearce (eds.) (1981). Ifs: Con-
ditionals, Belief, Decision, Chance, and Time. Dordrecht: D. Reidel Publishing
Company.
Hirsch, Eli (1986). “Metaphysical Necessity and Conceptual Truth.” In

Peter French, Theodore E. Uehling, Jr. and Howard K. Wettstein (eds.),
Midwest Studies in Philosophy XI: Studies in Essentialism, 243–256. Minneapolis:
University of Minnesota Press.
Hodes, Harold (1984a). “On Modal Logics Which Enrich First-order S5.”
Journal of Philosophical Logic 13: 423–454.
— (1984b). “Some Theorems on the Expressive Limitations of Modal Lan-

guages.” Journal of Philosophical Logic 13: 13–26.
Jackson, Frank (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis.

Oxford: Oxford University Press.
BIBLIOGRAPHY 352
Kripke, Saul (1972). “Naming and Necessity.” In Donald Davidson and Gilbert
Harman (eds.), Semantics of Natural Language, 253–355, 763–769. Dordrecht:
Reidel. Revised edition published in 1980 as Naming and Necessity (Cambridge,
MA: Harvard University Press).
Lemmon, E. J. (1965). Beginning Logic. London: Chapman & Hall.
Lewis, C. I. (1918). A Survey of Symbolic Logic. Berkeley: University of California

Press.
Lewis, C. I. and C. H. Langford (1932). Symbolic Logic. New York: Century

Company.
Lewis, David (1973). Counterfactuals. Oxford: Blackwell.
— (1977). “Possible-World Semantics for Counterfactual Logics: A Rejoinder.”

Journal of Philosophical Logic 6: 359–363.
— (1979). “Scorekeeping in a Language Game.” Journal of Philosophical Logic 8:

339–59. Reprinted in Lewis 1983: 233–249.
— (1983). Philosophical Papers, Volume 1. Oxford: Oxford University Press.
Linsky, Bernard and Edward N. Zalta (1994). “In Defense of the Simplest
Quantified Modal Logic.” In James Tomberlin (ed.), Philosophical Perspectives
8: Logic and Language, 431–458. Atascadero: Ridgeview.
— (1996). “In Defense of the Contingently Nonconcrete.” Philosophical Studies

84: 283–294.
Loewer, Barry (1976). “Counterfactuals with Disjunctive Antecedents.” Journal

of Philosophy 73: 531–537.
Lycan, William (1979). “The Trouble with Possible Worlds.” In Michael J.

Loux (ed.), The Possible and the Actual, 274–316. Ithaca: Cornell University
Press.
Mendelson, Elliott (1987). Introduction to Mathematical Logic. Belmont, Cali-

fornia: Wadsworth & Brooks.
Plantinga, Alvin (1983). “On Existentialism.” Philosophical Studies 44: 1–20.

BIBLIOGRAPHY 353
Priest, Graham (2001). An Introduction to Non-Classical Logic. Cambridge:

Cambridge University Press.
Prior, A. N. (1967). Past, Present, and Future. Oxford: Oxford University Press.
— (1968). Papers on Time and Tense. London: Oxford University Press.
Quine, W. V. O. (1948). “On What There Is.” Review of Metaphysics 2: 21–38.

Reprinted in Quine 1953a: 1–19.
— (1953a). From a Logical Point of View. Cambridge, Mass.: Harvard University

Press.
— (1953b). “Mr. Strawson on Logical Theory.” Mind 62: 433–451.
Rea, Michael (ed.) (1997). Material Constitution. Lanham, Maryland: Rowman

& Littlefield.
Russell, Bertrand (1905). “On Denoting.” Mind 479–93. Reprinted in Russell

1956: 41–56.
— (1956). Logic and Knowledge. Ed. Robert Charles Marsh. New York: G.P.
Putnam’s Sons.
Sher, Gila (1991). The Bounds of Logic: A Generalized Viewpoint. Cambridge,

Mass.: MIT Press.
Sider, Theodore (2003). “Reductive Theories of Modality.” In Michael J. Loux

and Dean W. Zimmerman (eds.), Oxford Handbook of Metaphysics, 180–208.
Soames, Scott (2004). Reference and Description: The Case against Two-
Dimensionalism. Princeton: Princeton University Press.
Stalnaker, Robert (1968). “A Theory of Conditionals.” In Studies in Logical

Theory: American Philosophical Quarterly Monograph Series, No. 2. Oxford:
Blackwell. Reprinted in Harper et al. 1981: 41–56.
— (1978). “Assertion.” In Peter Cole and Jerry Morgan (eds.), Syntax and Se-
mantics, Volume 9: Pragmatics, 315–332. New York: Academic Press. Reprinted
in Stalnaker 1999: 78–95.
BIBLIOGRAPHY 354
— (1981). “A Defense of Conditional Excluded Middle.” In Harper et al.

(1981), 87–104.
— (1999). Context and Content: Essays on Intentionality in Speech and Thought.

— (2003a). “Conceptual Truth and Metaphysical Necessity.” In Stalnaker

(2003b), 201–215.
— (2003b). Ways a World Might Be. Oxford: Oxford University Press.
— (2004). “Assertion Revisited: On the Interpretation of Two-Dimensional

Modal Semantics.” Philosophical Studies 118: 299–322. Reprinted in Stalnaker
2003b: 293–309.
von Fintel, Kai (2001). “Counterfactuals in a Dynamic Context.” In Ken Hale:

A Life in Language, 123–152,. Cambridge, MA: MIT Press.
Westerståhl, Dag (1989). “Quantifiers in Formal and Natural Languages.” In

D. Gabbay and F. Guenther (eds.), Handbook of Philosophical Logic, volume 4,
1–131. Dordrecht: Kluwer.
Williamson, Timothy (1998). “Bare Possibilia.” Erkenntnis 48: 257–273.
— (2002). “Necessary Existents.” In A. O’Hear (ed.), Logic, Thought and

Language, 233–51. Cambridge: Cambridge University Press.

Sider Theodore - Logic For Philosophy

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sider Theodore - Logic For Philosophy

Uploaded by

Copyright:

Available Formats

Logic for Philosophy

This book is an elementary introduction to the logic that students of contempo-

meanings of the logical symbols of first-order predicate logic without identity

· Propositional logic: Mendelson (1987)

· Descriptions, multi-valued logic: Gamut (1991a)

· Sequents: Lemmon (1965)

· Further quantifiers: Glanzberg (2006); Sher (1991, chapter 2); Wester-

· Modal logic: Gamut (1991b); Cresswell and Hughes (1996)

· Semantics for intuitionism : Priest (2001)

· Counterfactuals: Lewis (1973)

· Two-dimensional modal logic: Davies and Humberstone (1980)

3 Variations and Deviations from PL 63

5 Extensions of Predicate Logic 93

6 Propositional Modal Logic 118

6.3 Semantics for MPL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

7 Variations on MPL 191

7.2.3 Soundness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

9 Quantified Modal Logic 243

9.5.3 Necessary existence defended . . . . . . . . . . . . . . . . . 261

10 Two-dimensional modal logic 272

A Answers to Selected Exercises 289

B Answers to Remaining Exercises 298

English Propositional logic

English Predicate logic

1.1 Logical consequence and logical truth

1.2 Form and abstraction

It’s not the case that John is α

It’s not the case that John is happy

If we think of Argument A as having Form 1, then we can think of it as being

It’s not the case that John is happy

opposed to a Q in proofs or truth tables. That’s because P doesn’t symbolize

1.3 Formal logic

1.4 Correctness and application

1.5 The nature of logical consequence

competing philosophical accounts of its nature. Probably the most standard

Further, the notion of translation must be defined; further, an appropriate

intermediate proposal: φ is a logical consequence of the members of Γ iff, in

1.6 Extensions, deviations, variations4

We do this in order to get a better representation of logical consequence. There

‘Ted is happy’ logically implies ‘someone is happy’ using propositional logic

The king of the United States is bald

For example, in intro logic we show how:

1.7 Metalogic, metalanguages, and formalization

is then appropriately defined. (Think: introduction- and elimination- rules,

Soundness of propositional logic: In propositional logic, any proof-theoretic

Completeness of propositional logic: In propositional logic, any semantic

‘P ∧Q’ is a sentence with three symbols, one of which is

If there exists a derivation of a formula, then its truth

Thus, we formulate metalogical claims in the metalanguage, and our proofs in

1.8 Set theory5

· R∈/ R. Thus, R is a non-self-member. But R was said to be the set of all

· R ∈ R. So R is not a non-self-member. R, by definition, contains only

Thus, each possibility leads to a contradiction. But there are no remaining

members of f , then v = w; f (u1 , . . . , un ) is then defined as the object, v, such

e begin with the simplest logic commonly studied: propositional logic.

Whereas things like this do not:

· Sentence letters: P, Q, R . . . , with or without numerical subscripts

i) Sentence letters are wffs

We will be discussing a number of different logical systems throughout this

· P is a wff (clause i))

· so, ∼P is a wff (clause ii))

· Q is a wff (clause i))

We write “PL φ” for “φ is PL-valid”. (When it’s obvious which system