Proofs and Models in Philosophical Logic Greg Restall All Chapters Instant Download

Full download ebooks at ebookmeta.
com
Proofs and Models in Philosophical Logic Greg

Restall
https://ebookmeta.com/product/proofs-and-models-in-
philosophical-logic-greg-restall/
OR CLICK BUTTON
DOWLOAD NOW
Download more ebook from https://ebookmeta.com

More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Logic and Language Models for Computer Science Fourth

Edition Dana Richards
https://ebookmeta.com/product/logic-and-language-models-for-
computer-science-fourth-edition-dana-richards/
Cavell s Must We Mean What We Say at 50 Cambridge

Philosophical Anniversaries Greg Chase
https://ebookmeta.com/product/cavell-s-must-we-mean-what-we-say-
at-50-cambridge-philosophical-anniversaries-greg-chase/
Seven Myths of the Spanish Conquest Matthew Restall
https://ebookmeta.com/product/seven-myths-of-the-spanish-
conquest-matthew-restall/
Towards Scientific Metaphysics Volume 2 Benedykt

Bornstein s Geometrical Logic and Modern Philosophy
Polish Contemporary Philosophy and Philosophical
Humanities Krzysztof Slezinski
https://ebookmeta.com/product/towards-scientific-metaphysics-
volume-2-benedykt-bornstein-s-geometrical-logic-and-modern-
philosophy-polish-contemporary-philosophy-and-philosophical-
Philosophical Fallacies: Ways of Erring in
Philosophical Exposition Nicholas Rescher
https://ebookmeta.com/product/philosophical-fallacies-ways-of-
erring-in-philosophical-exposition-nicholas-rescher/
Proofs & Theories 1994th Edition Louise Gluck
https://ebookmeta.com/product/proofs-theories-1994th-edition-
louise-gluck/
Program Proofs 1st Edition K. Rustan M. Leino
https://ebookmeta.com/product/program-proofs-1st-edition-k-
rustan-m-leino/
Mathematical Stories I – Graphs, Games and Proofs: For

Gifted Students in Primary School Susanne Schindler-
Tschirner
https://ebookmeta.com/product/mathematical-stories-i-graphs-
games-and-proofs-for-gifted-students-in-primary-school-susanne-
schindler-tschirner/
Cassirer and Heidegger in Davos: The Philosophical

Arguments Simon Truwant
https://ebookmeta.com/product/cassirer-and-heidegger-in-davos-
the-philosophical-arguments-simon-truwant/
Philosophy and Logic
Proofs and Models

in Philosophical
Logic
Greg Restall
ISSN 2516-418X (online)

ISSN 2516-4171 (print)
Elements in Philosophy and Logic
edited by
Bradley Armour-Garb
SUNY Albany
Frederick Kroon
The University of Auckland
P RO O F S A N D M O D E L S I N
PH I L O S O P H I C A L L O G I C
Greg Restall
University of St Andrews
University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
103 Penang Road, #05–06/07, Visioncrest Commercial, Singapore 238467
Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781009045384
DOI: 10.1017/9781009040457
© Greg Restall 2022
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2022
A catalogue record for this publication is available from the British Library.
ISBN 978-1-009-04538-4 Paperback
ISSN 2516-418X (online)
ISSN 2516-4171 (print)
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Proofs and Models in Philosophical Logic
Elements in Philosophy and Logic
DOI: 10.1017/9781009040457
First published online: March 2022
Greg Restall
University of St Andrews
Author for correspondence: Greg Restall, gr69@st-andrews.ac.uk
Abstract: This Element is an introduction to recent work on proofs and

models in philosophical logic, with a focus on the semantic paradoxes and
the sorites paradox. It introduces and motivates different proof systems
and different kinds of models for a range of logics, including classical logic,
intuitionistic logic, a range of three- and four-valued logics, and
substructural logics. It also compares and contrasts the different
approaches to substructural treatments of the paradox, showing how the
structural rules of contraction, cut and identity feature in paradoxical
derivations. It then introduces model theoretic treatments of the
paradoxes, including a simple fixed-point model construction that
generates three-valued models for theories of truth, which can provide
models for a range of different non-classical logics. The Element closes
with a discussion of the relationship between proofs and models, arguing
that both have their place in philosophers’ and logicians’ toolkits.
Keywords: proofs, models, logic, semantics, paradox
© Greg Restall 2022

ISBNs: 9781009045384 (PB), 9781009040457 (OC)
ISSNs: 2516-418X (online), 2516-4171 (print)
Contents
1 Context 1
2 Proofs 9
3 Models 38
4 Connections 58
Glossary of Symbols 73
References 77
Proofs and Models in Philosophical Logic 1
1 Context
As far as academic disciplines go, logic is strange. In the western academy,
its roots go back to Aristotle, to Euclid, to the Stoics, through medievals, the
Arabic world, and into a flowering complexity in the nineteenth and twentieth
centuries, as philosophers and mathematicians grappled with understanding the
power and limits of deductive reasoning. The field we now know as modern
logic took root in the project of systematising and securing the foundations of
mathematics1 and in giving an account of the relationship between those math-
ematical theories and our experience of the world around us. In the twentieth
century, new connections emerged with the nascent fields of linguistics, digital
systems and computer science. There is no way that an Element on the use of
proofs and models in philosophical logic could do justice to anything more than
a tiny fragment of this sprawling edifice.2
So what small fragment of the discipline of logic will this Element address?
As the title states, our focus is philosophical logic and the twin roles of proofs
and models in the development of logic. The philosophical concern will also
be twofold: we will reflect on the application of logic to some questions in
philosophy and, at the same time, consider a philosophical reflection on the
discipline of logic itself. Philosophical logic provides both a set of sensibilities
and processes and tools for application in philosophical discourse (among other
kinds of discourse), and at same time, it is a site of philosophical reflection. We
will maintain these dual perspectives on our topic throughout this Element.
In this first section, I set the scene by way of an introduction to how we can
use the different tools of proofs and models to form judgements about logical
validity and invalidity. Then I outline how attention to proofs and models plays
a role in some of the current debates in philosophical logic, concerning the
semantic paradoxes and vagueness. I then end the section by looking ahead to
the argument of the remainder of the Element.
1.1 Proofs and Models

There are many ways to look at logic and the constellation of concepts that
logicians have attempted to analyse using proofs and models. One way to do
1 J. Alberto Coffa’s The Semantic Tradition from Kant to Carnap (1993) tells the compelling
story of the growth of modern logic beyond its Aristotelian bounds as Bolzano, Weierstrass and
others attempted to make sense of the mathematically important notions of convergence and
continuity.
2 So I will not cover the rich tradition of proof complexity, Gentzen’s consistency proof for arith-
metic, the connections between proof search and decidability and many more interesting topics
in proof theory. Neither will I discuss a great deal of model theory, such as significant meta-
theoretical results including compactness, cardinality of models or ultrapowers, and other model
construction techniques. This Element is only so long.
2 Philosophy and Logic
this is to focus on the production and the evaluation of argumentation, reasoning

and inference. Let’s start with a simple example, involving two mathematicians,
who are reasoning about some newfangled binary relation R they are exploring.
They have discovered that the relation R is transitive (i.e., for any objects x, y
and z if R relates x to y and R relates y to z, then R relates x to z too, so is
older than is an example of a transitive relation, while is a parent of is not)
and symmetric (if R relates x to y, then R relates y back to x too, so neither is a
parent of nor is older than are symmetric relations but is a sibling of is), and
finally, the relation is directed (for each object x, there is some object y where
R relates x to y, so, on the collection of non-negative natural numbers (natural
numbers for short) 0, 1, 2, . . ., the relation is smaller than what is directed, since
for every number x we can find some whole number y where x is smaller than
y, but the relation is larger than what is not directed if we restrict our attention
to the natural numbers. There is no natural number y where 0 is larger than y).
So our two mathematicians agree that this newfangled relation R is transitive,
symmetric and directed. One of our mathematicians exclaims: ‘The relation
R is reflexive, too!’ (A relation R is reflexive if, for every object x, R relates
that object to itself ). Our second mathematician asks: why is that? The first
responds:
(1) R is transitive. It’s symmetric. It’s directed. It must be reflexive, too!
The second mathematician doesn’t see why this is the case, so they ask for the
reasoning to be spelt out. Can the leap from the premises to the conclusion be
broken down into smaller steps? It can. Our quick thinker responds:
(2) Take an object a. Since R is directed, there is some object b where R

relates a to b. Since R is symmetric, R also relates b to a. Now, R relates
a to b and R relates b to a, so since R is transitive, R relates a to a. So
we have just shown that for any object a at all, R relates a to itself. That
is, R is reflexive.
This elaboration is one way to spell out the jump from the premises to the con-
clusion. It is what we call a proof. It fills out that large jump in thought in
terms of smaller steps. In this case, the smaller steps involve the applications
of agreed-upon definitions (unpacking the definitions of the terms reflexivity,
transitivity, directedness and symmetry), the operations of individual logical
concepts like the universal and existential quantifiers (in the concept of direct-
edness, e.g., for every x, there is some y where R relates x to y), and other logical
concepts like conjunction (the and in the definition of transitivity) and condi-
tionality (the if in the definitions of transitivity and symmetry). If you were
to question this proof at any of the steps in the explanation, it would seem to
be a very different problem than not understanding a large leap in thought. It

would be a problem of understanding the concepts in use and not a problem
of understanding how those concepts are combined. (That is the idea, anyway.
Exactly how proofs work, and what options we have in understanding them is
the topic of the next section.)
Proofs are one side of the logical coin. Models are the other. To explain
models, we might consider another possible transition in thought: suppose our
mathematicians have another newfangled relation S, and they have concluded
that S is transitive and symmetric, but they do not know whether S is directed
or not. They wonder, does it follow from these two properties (transitivity and
symmetry) that S is reflexive? We can see that the reasoning supplied previ-
ously concerning R does not apply in the case of S, since we do not know that
S is directed. But having one potential proof that doesn’t work does not mean
that there isn’t another proof that does. Is it the case that when a relation S is
symmetric and transitive, then it must also be reflexive? Our mathematicians
think for a while, and sketching some ideas on a blackboard, they draw the
following diagram:
This diagram represents a way a relation S could be. Each dot is a different
object in the domain of the relation, and a line connecting dot x with dot y
indicates that S relates object x to object y. By design, the relation depicted is
symmetric, since if there is a line connecting x with y it is line that connects y
with x. We can see, too, that whenever we can get from x to y and from y to
z by lines, there is a direct line from x to z. This holds even in the case where
we can get from x to y and back – whenever there is a line from x to anywhere
at all, there is a line looping around from x to itself. So the relation depicted
here is transitive too. However, it is not reflexive because we have an isolated
dot in our diagram. This lonely object is not a counterexample to the claim of
symmetry or of transitivity for S, but it shows that if S were like this, it would
not be reflexive. We have a counterexample to the rule that symmetric and
transitive relations must be reflexive.
This counterexample is a model. It is not a claim about how the origi-
nal relation S is. It is a sketch, a representation, showing that if we want an
explanation why a relation S is reflexive, we cannot appeal merely to its transi-
tivity and its symmetry, since the argument breaks down in circumstances like
those depicted in this model. A counterexample is a barrier through which our
argument cannot pass.
Here, in a nutshell, we have the distinction between proofs and models. For
the first argument, we have provided a proof, leading from the premises to the
conclusion, and for the second, we have provided a counterexample, a model
that renders the premises true and the conclusion false.3
The proofs and models we have seen so far are relatively simple, involving
reasoning with the quantifiers all and some and familiar logical connectives
like if and and. There are important questions about the role of these concepts
in our argumentation and in our construction of proofs and models. There is
broad agreement that these quantifiers and connectives are important. There is
less agreement over whether there is anything categorically distinctive about
those concepts.4

However, there are concepts other than the familiar connectives and quantifiers
that have proved important for philosophical logic, and which are amenable to
treatment by way of proofs and models. One example is provided by modal
concepts, such as possibility and necessity. However, the proofs and models
appropriate for modal operators seem qualitatively different to the models we
have seen so far. They do not just represent a way things could have been but
also represent more than one such way that things could be. To see how they
arise, consider the difference between two different arguments:
(3) It’s possible that either p or q. So either it’s possible that p or it’s
possible that q.
We can fill in this reasoning into a proof in the following way:
(4) Since it’s possible that either p or q, we grant some possibility where
either p or q holds. Suppose it’s p. In that case, we can conclude that
(back where we started) it’s possible that p, and so it’s either possible
that p or it’s possible that q. On the other hand, suppose that the pos-
sibility we granted makes q hold. In that case (also, back where we
started), it’s possible that q, and again, either it’s possible that p or it’s
possible that q. So, in either case, it’s possible that p or it’s possible that
q, and we’re done.
3 Distinguishing validity as defined by way of proofs and validity defined by way of models was
a great conceptual advance in the twentieth century. Consult Zach (1999) for a discussion of the
early days of that development of the distinction.
4 There is extensive literature attempting to characterise the logical constants from other concepts.
We will not explore it here. Gila Sher’s The Bounds of Logic is a good place to start (1991).
In this proof, we broke down the leap from premise to conclusion into smaller
steps, using more fundamental principles governing possibility and disjunction.
This is a good candidate for being a proof.
Suppose, on the other hand, we tried a similar argument, with necessity in

place of possibility.
(5) It’s necessary that either p or q. So either it’s necessary that p or it’s
necessary that q.
This argument is less convincing. We can propose a model as a counterexample.
(6) Suppose we have a range of possibilities, where some of them (not all)
make p true, and the all the others (again, not all of possibilities) make
q true. In that case, in each possibility, we have either p or q – so from
the point of view of any possibility at all, it is necessary that either p or
q. Nonetheless, we don’t have that it is necessary that p (since in some
possibilities, p fails), and we also do not have that it is necessary that q
(since in other possiblities, q fails). So, in any possibility in our model,
we do not have either that it is necessary that p or that it is necessary
that q. So this model, at any possibility, the premise is true, and the
conclusion is not.
Modal reasoning works just like the other reasoning we have seen. Valid argu-
ments can be broken down into proofs, while invalid arguments can be given
models as counterexamples. The models have a richer structure than the mod-
els we saw at first. We used not only a representation of one way that things
might be but also a range of such representations, a system of different possible
worlds.
There are many more concepts that can be rigorously explored with proofs
and models, like the identity predicate; definite and indefinite descriptions;
quantifiers (over objects) beyond the existential and universal quantifier; quan-
tifiers ranging over other domains, such as functions, propositions or properties
and much more. However, this will be more than enough to be going on with
for what follows.
1.2 Paradoxes
As soon as the field developed accounts of proofs and models – in fact, before
these tools took distinct shapes – some natural questions arose. How do we
evaluate those tools? Do they distinguish good and bad arguments correctly
(whatever that would mean), or should the main candidates for the correct
account of proofs or the correct account of models be revised or rejected? Some
of the most active revisionary arguments concerning proofs and models have
involved different kinds of paradoxes. After all, a paradoxical argument is one
where the premises seem true, the argument seems valid and the conclusion
seems false. If we want to find a good reason to take some argument that is
traditionally thought to be valid to be, in fact, invalid, then the paradoxes are
where we should look.
Example 1: The liar paradox

Consider this sentence, which says of itself that it is not true.
(7) Sentence (7) is not true.
It seems that we can reason like this. Suppose (7) is true. Then, since (7) says
that (7) is not true, then it would follow that (7) is not true, which would contra-
dict (7) being true. In other words, if (7) is true, we have a contradiction. This
means that (7) is not true, since we reduced the supposition that (7) is true to a
contradiction. We’ve refuted it. But this means that we have proved that (7) is
not true and that is what (7) itself says. So we’ve proved (7). It’s true. And so,
it isn’t. We have proved a contradiction.
This is the liar paradox, one example of a semantic paradox, and a paradox
of self-reference.5 As we will see in the next section, we have used very few
logical principles in this line of reasoning, and it very much looks like we have
made some kind of mistake, though it has proved very difficult to locate the
mistake to everyone’s satisfaction. For some, the semantic paradoxes like the
liar have been seen as reasons to curtail our rules of proof for the logical con-
nectives in some way or other, so as to stop the contradictory conclusion or to
render the contradictory conclusion palatable. For others, the fact that logical
principles like these are involved in the proof means that the problem must lie
elsewhere, either in the so-called definition of the liar sentence (say, we attempt
to ban self-reference) or to say that despite appearances, the logic of the truth
predicate cannot satisfy the rules used here in the derivation of the paradoxes.
There are many different kinds of response to the liar paradox, and we will dis-
cuss a representative sample of these in the coming sections, since doing so will
give us a range of perspectives on what we are doing when we use the logical
tools of proofs and models.
5 Another example, which we will also consider, is Curry’s paradox, which uses the conditional,
where the liar uses negation. Pick some statement p. Consider c, the class of all classes x where
if x is a member of itself, then p. Suppose c is a member of itself. Then, it follows that if c is a
member of itself, then p. So, combining those two facts, we have p. In other words, we have just
proved that if c is a member of itself, then p. So, it follows that c is a member of itself. Again,
putting these together, we have proved p. And we made no assumption about p at all.
Example 2: The sorites paradox

Consider a colour strip of colour, shading evenly from red on the left to yellow
on the right. Let’s divide the strip up into 10 000 evenly sized tiny patches, from
left to right, labelled 1 to 10 000. For each number n from 1 to 10 000, consider
the claim that patch number n looks red (to me). The first such claim, ‘patch 1
looks red (to me)’, is true. The last such claim, patch 10 000 looks red (to me), is
false. The claim ‘if patch 1 looks red to me, so does patch 2’ also seems true, not
just because both patches look red but also because they look indistinguishable
to me. This generalises: three distinct features conspire to make it that each
claim of the form ‘if patch n looks red to me, so does patch n + 1’ seems just
as true. First, the strip shades evenly from red to yellow, with no sharp changes
in observable colour. Second, we chose very many subdivisions, so each patch
differs from its neighbours by at most a tiny difference, and third, my powers of
visual discrimination have their limits. So the premises of this argument seem
true:
(8) Patch 1 looks red to me.
If patch 1 looks red to me, so does patch 2.
If patch 2 looks red to me, so does patch 3.
..
.
If patch 9 999 looks red to me, so does patch 10 000.
Therefore, patch 10 000 looks red to me.
From these premises, we can draw the conclusion that patch 10 000 looks red
to me, using one very simple principle of logic, the inference rule of modus
ponens, which takes us from a conditional claim of the form if A then B and
its antecedent A to its consequent, B. Unfortunately, for us, we seem to have
a logically valid argument from premises that seem true to a conclusion that
seems false.

If we wish to find a counterexample to the validity of the sorites argument, we

need to find some model in which the premises hold and in which the con-
clusion fails. If there is no such counterexample, then either we grant that the
argument is valid or we reject the constraint that invalidity must be witnessed
by a model as a counterexample. Classical ‘two-valued’ logical systems have
proved difficult to adapt to this task. A two-valued model will assign ‘true’ or
‘false’ to each sentence ‘patch n looks red to me’, which will involve assign-
ing either ‘true’ to each such claim (so this model represents the whole strip as
looking red to me) or ‘false’ to each claim (so the model represents the whole
strip as not looking red to me), or there are two adjacent patches, and the model
represents one as looking red to me and the other as not looking red to me. But,
as we said, the set-up is designed to make each patch indistinguishably differ-
ent from its neighbours. So, if that is unpalatable, a natural reaction involves
expanding the picture of semantic evaluation to allow for more than the two
values of ‘true’ and ‘false’: logics with truth-value gaps or a whole panoply
of degrees of truth might provide ways to understand the sorites paradoxes. In
Section 3, we will examine options for the sorites paradox, as well as other
reflections on models that have proved fruitful in philosophical logic in recent
decades.
1.3 The Plan

The semantic paradoxes and the sorites paradox are two examples of paradoxes
over which a great deal of ink has been spilt in recent decades. The philo-
sophical literature concerning the paradoxes provides one entry point – among
many – to the different approaches to understanding the foundations of logical
consequence, and this is our entryway into the broader landscape of the use of
proofs and models in philosophical logic. The paradoxes are sites where what
seemed for all the world to be fundamental principles about proofs and about
models come into tension and give rise to what seem to be absurd conclusions.
Different proposals for revising those fundamental principles or for defusing
the tension provide different approaches to understanding these principles of
logic, and they will provide a suitable set of lenses through which to view key
ideas in logic as they have developed.

In Section 2, we will discuss logic from the standpoint of proof, giving a quick
introduction to the kinds of techniques philosophical logicians have adopted
in the study of proof and its application to issues in semantics, epistemology
and metaphysics. In this section, we will keep an eye on the kinds of responses
people have made to the semantic paradoxes, as these paradoxes provide ample
motivation for us to inquire into the costs and benefits of different fundamental
proof-theoretical principles. Then, in Section 3, we will do the same thing for
models, introducing not only the debate over the applicability of the standard
two-valued ‘classical’ semantic picture in the light of the paradoxes, but also
our sights that will involve discussions of other kinds of models of use in phil-
osophical logic, such as models featuring possible worlds, which have proved
so useful, and so controversial, in giving an account of the meanings of modal
expressions.
After those two sections, we will wrap up with Section 4, in which we explore
not only the ways that these tools are used in the discussions of the paradoxes
but also some other natural questions, including the relationship between proofs
and models themselves. In particular, we will ask which proofs and models
should be taken to be fundamental. But, first, let us turn our attention to proofs.
2 Proofs
Let’s start with the proof labelled (2) on page 2. That is a proof that R is reflex-
ive. It has three premises: R is directed, R is symmetric and R is transitive.
It lays out a path from those premises to the conclusion, leaving nothing out.
The aim of a proof is not just to convince us that some conclusion is true but
also, in some sense, to make explicit how that conclusion follows from the
premises.
Notice that this proof does not just start from the premises and lead to the
conclusion, with each intermediate step following from the ones granted before
it.6 There are some other important features of our reasoning that are worth
examining. First, our proof includes an imperative: ‘take an object a’. This
sentence is not a premise, nor is it a conclusion, and it is not another statement
that follows from the premises. It is an invitation. It cannot be true or false. We
cannot assert or deny it. Second, the term ‘b’ in the proof also has an interesting
status. We moved from the claim that R is directed (so, in particular, there must
be some object to which R relates a) to calling one such object b. The fact that
R relates a to a given object b does not logically follow from the claim that R is
directed. R could be directed without R relating a to this particular b (whichever
b it happens to be).7 So much more is going on in this proof than simply working
out conclusions from things we have granted. There are different steps where
objects are given names, and speech acts, other than asserting, are involved
as well. Proofs have complex structure. A crucial constraint, though, is that a
proof is not simply a statement of the premises and the conclusion – at least,
not in most cases. To prove some conclusion C from some premises P1 , . . . ,
Pn , you must somehow trace the connection from P1, . . . , Pn to C.
6 Proofs with that direct ‘linear’ structure Hilbert Proofs. We will see these in the next section.
7 If you find this puzzling, think of a concrete case. I have a son. As a matter of fact, Zachary is
my son. The fact that Zachary is my son does not follow as a logical consequence of the fact
that I have a son because there are other ways I could have had a son, other than Zac. Similarly,
if R relates a to some object, in any particular circumstance in which that is true, we could call
that object b. Given that choice, it would still not follow that in every circumstance where R
relates a to something, that R must relate a to that object b. It might have related a to some other
object instead.
2.1 Proof Structures

What kind of structure can a proof have? How can such connections be made,
leading from premises to conclusions? In philosophical logic, over recent
decades, the focus of research has not been on any general overarching theory
of the structure and properties of proofs as such.8 We have seen one example,
but in the study of proofs and the formal development of proofs through the
years, formal accounts of different kinds of proofs have been represented in
many different ways, each with different costs and benefits.
Let’s start with the standard logical connectives such as ∧ (conjunction), ∨
(disjunction), → (the conditional) and ¬ (negation) and the quantifiers such as
∀ (universal) and ∃ (existential) and consider proofs for statements exploiting
the concepts in this vocabulary. As mentioned previously, a simple analysis of
proofs is given by the structure of Hilbert Proofs. A Hilbert Proof from prem-
ises P1, . . . , Pn to conclusion C is a list of formulas, ending in the conclusion C,
and each of them is (a) one of the premises Pi , (b) one of the axioms or (c) a for-
mula that follows from earlier formulas in the list by way of the rules. Whether
something counts as a Hilbert proof depends on the choices of axioms and rules.
If we focus just on the connectives → and ¬, we can provide a very simple
Hilbert proof system for propositional logic. There are three axiom schemes:9
WEAKENING: A → (B → A)
DISTRIBUTION:(A → (B → C)) → ((A → B) → (A → C))
CONTRAPOSITION: (¬B → ¬A) → (A → B)
These are schemes rather than individual formulas because we count any for-
mula that fits the shape as an instance of the axiom. So, for example, p → (q →
p) and p → (p → p), and ¬(p → q) → (p → ¬(p → q)) are each instance of
the WEAKENING axiom scheme. Our Hilbert proof system has just one rule:
MODUS PONENS: A → B, A =⇒ B
We understand the rule like this: if, in a proof, we have already written down
A → B and A (in either order), then we can add B to our proof. With this
choice of axioms and rules, we can construct proofs in our axiom system
8 See some of Dag Prawitz’s papers to see what could be done in this direction (1973, 1974,
2019).
9 Hilbert proofs are named after German mathematician David Hilbert (1862–1943), for whom
the axiomatisation of mathematics, with a precise formally specified notion of proof, was a
central task to the foundation and justification of mathematical methods (see Sieg, 2013; Zach,
2019).
straightforwardly.10 A proof from a set X of premises to a conclusion A is a

list of formulas, where each formula is either an X or an axiom or follows from
earlier formulas in the proof by way of the rule and ends in A. Here is a proof
from p → q and q → r to p → r.
1. p→q PREM
2. q→r PREM
3. (q → r) → (p → (q → r)) WEAK
4. p → (q → r) 2, 3, MP
5. (p → (q → r)) → ((p → q) → (p → r)) DIST
6. (p → q) → (p → r) 4, 5, MP
7. p→r 1, 6, MP
I have numbered every formula in the list and annotated each line with a note
indicating the source of the formula. For example, the formula on line 3 is an
instance of the WEAKENING axiom, while the formula on line 4 follows from the
formulas on lines 2 and 3 using MODUS PONENS.
Recall our motivating conception of a proof as drawing out the connection
between the premises and the conclusion, leaving nothing out. Hilbert proofs
meet that aim, provided that we are happy to grant that the axioms need no
further justification and that the rules of the system are individual steps of
deduction, requiring no further analysis. In the case of this Hilbert system, the
rule MODUS PONENS seems like a good candidate for meeting that criterion, but
the axioms are another question entirely. These axioms look like the kinds of
statements that could stand in need of proofs themselves, and this is one reason
why, for the purposes of thinking of proofs as laying out justificatory connec-
tions as explicitly as possible, attention in philosophical logic has moved away
from Hilbert proofs to other proof structures in which there are fewer axioms
and more rules.11 We will consider natural deduction proofs, which correspond
more closely to the everyday proof from our example (2).12
Here is a short natural deduction proof from the three premises q → r, p → q
and p, using the one rule →E, which licences the inference from a conditional
(A → B) and its antecedent (A) to its consequent (B). Instead of having a linear
structure, this natural deduction proof is laid out as a tree.
10 The definition is straightforward. This does not mean that it is straightforward to find a proof
for any given argument.
11 There are many such proof structures to consider. We will set aside tableau (tree) proofs (see
Smullyan, 1968) and resolution proofs (see Genesereth & Kao, 2016) in what follows.
12 This is a natural deduction in Gentzen–Prawitz format (Prawitz, 1965). There are other ways to
format natural deduction proofs, too. See Pelletier’s history (1999) for more.
p→q p
→E
q→r q
→E
r
Here, we see that the conclusion r follows from the three premises, which are
laid out as leaves of the tree. So far, this is only a presentational difference with
Hilbert proofs, but instead of specifying the behaviour of the conditional by
adding axioms, we add another rule, showing how to introduce a conditional
statement given a proof. In the previously given proof, we have shown that
r holds, given the three assumptions. So, if we grant q → r and p → r and
withhold commitment to p, we can at least conclude that if p then r. That is,
we can conclude p → r. So we have the following two rules for the conditional:
[A]1
Π A→B A
→E
B B
→I1
A→B
The →E rule is straightforward. The →I rule needs some explaining. It says

that if we already have a proof (call it ‘Π’) with B as a conclusion and with A
occurring some number of times among the assumptions,13 then we can extend
the proof by discharging those occurrences of A (surrounding them with brack-
ets) and conclude A → B. The active assumptions remaining in the proof are
whatever assumptions were active before, except that we remove the indicated
occurrences of A from that list. So we can extend our previously given proof
with one more step:
p→q [p]1
→E,
q→r q
→E
r
→I1
p→r
to now conclude p → r from the premises p → q and q → r. This is a very

different proof to the Hilbert proof seen previously, but it, in its own way, shows
how p → r follows from p → q and q → r. One big difference is that our
natural deduction proof uses no axioms at all (except for the idea that we can
assume anything at all – a formula A on its own is a ‘proof’ with premise A
and conclusion A), the →I rule shows how we could introduce a conditional
A → B (in terms of its constituents, A and B) and the →E rule shows how we
could eliminate or exploit such a conditional.
13 For reasons, we will discuss in Section 2.3 that number of occurrences can, in fact, be zero.
[A]i
A B A∧B A∧B Π A→B A
∧I ∧E ∧E →E
A∧B A B B
→Ii B
A→B
[A]i [B]i
A B
∨I ∨I A∨B Π1 Π2
A∨B A∨B ∨Ei
C
[A]1
Π ¬A A ⊥
¬E ⊥E
⊥ ⊥ A
¬I1
¬A
X [A(n)]i
Π ∀xA(x) A(t) Π
A(n) ∀E ∃I
∀I A(t) ∃xA(x) ∃xA(x) C
∃Ei
∀xA(x) C
s=t A(s) s=t A(t)

=I =E =E
t=t A(t) A(s)
Figure 1 Natural deduction rules
In fact, rules of these kinds can be given to the other connectives and quan-
tifiers (they are summarised in Figure 1), and using these rules, we can display
the structure of proofs like the everyday reasoning of proof (2).
∀x∀y(Rxy → Ryx)
∀E
∀x∀y∀z((Rxy ∧ Ryz) → Rxz) ∀y(Ray → Rya)
∀E ∀E
∀y∀z((Ray ∧ Ryz) → Raz) Rab → Rba [Rab]1
∀E →E
∀z((Rab ∧ Rbz) → Raz) [Rab]1 Rba
∀E ∧I
∀x∃yRxy (Rab ∧ Rba) → Raa Rab ∧ Rba
∀E →E
∃yRay Raa
∃E1
Raa
∀I
∀xRxx
Here, the leaves of the proof tree that haven’t been discharged are formulas stat-
ing that R is directed (∀x∃yRxy), transitive (∀x∀y∀z((Rxy ∧ Ryz) → Rxz)) and
symmetric (∀x∀y(Rxy → Ryx)), and the conclusion states that R is reflexive
(∀xRxx). Each step is either an introduction or an elimination rule, operating
on a single logical concept, a quantifier or a connective. The rules for the

conditional we have seen and the rule introducing a conjunction are straight-
forward. The other rules need some comment. The ∀E steps eliminate or apply
a universally quantified statement by substituting some term (in this case, the
terms a and b) for the variable bound by the quantifier at the head of the expres-
sion. In general, ∀E mandates the inference from some statement ∀xA(x) to
A(t) for some term t.14 This rule says that from a universal claim, we can infer
each of its instances. The remaining rules in this proof are ∀I and ∃E. The
last step in our proof is a ∀I step. It echoes the last step in our intuitive proof
(2). In the previous step, we concluded that Raa, and by this time, the object
a in question is arbitrary, in the sense that none of the remaining premises
in the proof contain the term a. We have made no assumptions about what
a is, and so the argument applies generally to anything. So we can conclude
∀xRxx. This is the shape of the rule ∀I. We can infer ∀xA(x) whenever we have
proved A(n) with some name n, provided that n does not occur in the premises
X used in that proof. This move corresponds nicely with the corresponding step
in proof (2).
The remaining rule is ∃E, and this step also corresponds to a move made in
proof (2), though not quite as directly. As we have seen, there is a kind of ‘step’
that we make from ∃yRay to Rab, but this step should not be thought of as an
inference in the way that other steps are, since there is no sense in which Rab
has to be true if ∃yRay is true. In our formal proof, the step is conceived of in
this way. If we have concluded ∃yRay (as we have from the premise ∀x∃yRxy)
then to use this fact, we can assume Rab, where b is some fresh name that we
have not used before. If we can then conclude some assumption C from that
assumption, where C also does not include the name b, then we can discharge
the assumption Rab and conclude C on the basis of the more general claim
∃yRay. And this is the form of the ∃E rule. So there is a sense in which the
statement Rab ‘follows from’ ∃yRay, not as a valid inference but in the sense
that its use as an assumption is discharged on the basis of the weaker formula
∃yRay.
So we have seen that the intuitive proof of our mathematicians’ everyday
reasoning corresponds fairly straightforwardly (after some rearranging) with a
formal natural deduction proof in which every step is either an introduction or
an elimination.
14 To state this precisely requires an accounting of when a variable in a formula is bound by a

quantifier and the substitution of a term for a variable so bound. The details can be found in any
good textbook, so we will not pause to give the precise definition here.
2.2 Inference Rules as Definitions

Let’s return to the idea of proofs as making explicit the connections between
the premises and the conclusion of an argument. In our original unformalised
proof, the basic moves were of two different kinds: they were ‘rewriting’ defi-
nitions of terms of art like ‘reflexive’, ‘transitive’ and so on. Disagreeing with
any of those steps in our reasoning would have simply been disagreeing about
what it is to be transitive or reflexive. Those definitional steps are straightfor-
ward enough. To accept the definition is to endorse those inferences. What can
we say for the other steps in the proof, which, if we are to accept the formali-
sation in the language of Gentzen/Prawitz natural deduction, are steps like ∀I,
→E, and so on. A natural thought is that these steps, too, are in some way dis-
tinctive. They certainly look as if they cannot be decomposed into any simpler
steps. If we are breaking inferences down into simpler steps, inference rules
like these, in which one component of a judgement is operated on at a time,
seem like the end point.15 For proof to not be a thoroughgoingly relative or
contextually determined notion, or a useless one, there must be some stopping
point at which the task of explication ends, and the primitive inference rules
of this form is at least suggestive of being the right place.16 One way that this
thought can go further is to take the parallel with definitional steps seriously: to
take the connective rules to define the concepts they govern. This is a natural
thought. After all, we do use rules like ∧I and ∧E as one way of specifying
what we mean when we use the logicians’ conjunction (in the same way that
we might use the truth conditions of a truth table for ∧). However, this view
brings with it some obvious questions. It’s one thing to introduce an abbrevia-
tive definition and to give some account of when it is a legitimate definition.
A non-circular definition of a connective may not always be helpful, but it will
always be a legitimate option. There is no bar to calling a relation R where
∀x∃yRxy directed. This poses no problems, provided we aren’t already using
that word for anything else. But our inference rules are not quite like abbre-
viative definitions, and as Arthur Prior pointed out (1960), not every pair of
introduction and elimination rules is legitimate. Prior’s argument is simple. If
a pair of introduction and elimination rules counts as a definition, then what is
to rule out defining a concept like tonk?
15 For those familiar with model theoretic semantics, inference rules of this form seem to be a
natural analogue to truth or satisfaction conditions for each connective, in which the concept
being interpreted is dominant and no other connective features essentially.
16 Lewis Carroll’s dialogue between Achilles and the Tortoise (1895) is a maddening and delightful
example of what can happen if there is no such stopping point.
A A tonk B
tonkI tonkE
A tonk B B
With the addition of ‘tonk’ to our vocabulary, we could reason from anywhere
(say, from the premise: I exist) to anywhere (to the conclusion: nothing exists)
by reasoning through the intermediate step (I exist tonk nothing exists) that
follows from the premise by a step of tonk introduction and from which the
conclusion follows by a step of tonk elimination. Clearly, this is too powerful
a tool to use in our reasoning toolkit. Prior took this to be a challenge to the idea
that inference rules like the introduction and elimination rules for conjunction
count as definitions, which may be freely added to our language whenever we
please. Rules of this general shape can be powerful – too powerful to be safely
added. Some kind of extra criterion must be imposed on our rules if we are
to take extending our vocabulary by way of these rules to be the kind of free
addition typically given by a stipulative definition.
Since it is such a crisply formulated problem, Prior’s ‘tonk’ has found no
small number of responses. One simple and straightforward answer was given
by Leslie Stevenson (1961), who argued that connectives like tonk are unsuita-
ble because they cannot be made consistent with any truth table. Such an answer
takes things out of the orbit of proofs and into the domain of models, so we will
defer its discussion to the next section. Nuel Belnap (1962) formulated a prop-
erly proof-theoretic reply when he argued that stipulative definitions – and the
traditional inference rules – share properties that rules like those for tonk lack.
They are conservatively extending and uniquely defining.
To explain these terms, let’s specify the set-up some more. We are consid-
ering the addition of some term T to some original language L to form a new
language L + T. We suppose that our language L comes equipped with a conse-
quence relation, formulated in terms of proof. We say that C is a consequence
of our premises P iff there is a proof from P to C. So, when we move from the
original language L to the larger L + T, there are not only new sentences that
can be formed in our new language but also new proofs that can be formulated
using our new vocabulary – our new term, T. In particular, it may be that we
have new proofs from premises in our old vocabulary (sentences from L) to a
conclusion in our old vocabulary (also from L), but it makes a detour through
our new vocabulary, in L + T. The addition of T to our new language is said
to be conservative if whenever there is an L + T-proof from premises in L to a
conclusion in L, there was originally an L-proof from the same premises to the
same conclusion.
Stipulative definitions – which add a new vocabulary item to our language
by specifying it as a shorthand for some phrase in the old vocabulary – are
conservative in just this sense, because if we had some proof in L + T, using
the new term T, we could simply rewrite that proof by replacing all the appeals
to the new term by uses of the phrase defining it it in the older vocabulary. The
result is a proof in the original language, L.
It is plausible that extensions by definition (whether by paraphrase, or any
other means of defining a term) should be conservative in this sense because
a non-conservative extension to our language is more than the addition of new
concepts to enable new things. A non-conservative extension is also a revision
of our view of what holds concerning the previous vocabulary. Such an addi-
tion to our vocabulary is a revision, since positions that were previously open
(granting the premises P and rejecting the conclusion C) are now deemed to be
closed.
Conservative extension in just this sense is one criterion that we could use
for distinguishing the behaviour of Prior’s ‘tonk’ given its rules, and the rules
used to define a connective like ‘∧’. Prior’s tonk is clearly not conservative
over a base language where the argument from some given premise p to some
given conclusion q had no proof, for once tonk is added, we have a short,
two-step proof from p to q through p tonk q. Can we show that the logical
connectives, like ∧, → and ¬, or the quantifiers, like ∀ and ∃, do not fall foul
of this criterion? We will see, in the next section, how there is a certain sense
in which these logical concepts (in certain contexts) pass Belnap’s test, but
before we consider that, we should consider the second of Belnap’s criteria:
uniqueness.
Consider a second pair of inference rules, designed to introduce another kind
of logical connective, one we can dub ‘gonk’.
A B A gonk B
gonkI gonkE
A gonk B B
These rules are quite conservative over our usual vocabulary, but they allow
for ‘wiggle room’. There is nothing in these rules that can distinguish gonk
from conjunction. If we replace ‘gonk’ in our language by ‘∧’ everywhere,
these rules would remain valid under this translation, so as far as these rules
are concerned, ‘gonk’ could be just another way to say ‘∧’. (The fact that these
rules do not say that we can infer from A gonk B to A does not mean that this
inference must be invalid.) Similarly there is nothing in these rules that can
distinguish A gonk B from B. If we replace every sentence in our language
of the form of A gonk B by alone, B and these rules remain valid under this
translation, too. Our rules for ‘gonk’ tell us only that A ∧ B entails A gonk B,
and A gonk B entails B, but exactly where between A∧B and B is left unsettled.
It follows that the rules for ‘gonk’ are defective in a different way to the
defects in the rules for ‘tonk’. The rules for a concept c are uniquely defining
if whenever we define two concepts c1 and c2 using those rules, then c1 and c2
are indistinguishable, using those rules, in the following sense. If c1 is used as a
premise, it could be replaced by c2 and vice versa. If c1 is used as a conclusion,
it could be replaced by c2 and vice versa. Our rules for gonk do not satisfy
this criterion. Suppose we use the ‘gonk’ rules twice to introduce connectives
gonk 1 and gonk 2 . If we had a proof from p gonk 1 q to p gonk 2 q using these
rules, then we could systematically rewrite that proof using the two translations
mentioned in the previous paragraph. Replace A gonk 1 B everywhere by B.
Replace A gonk 2 B everywhere by A ∧ B. The gonk rules (in their 1 or 2
guises) are valid under this interpretation, and this means that any proof from
p gonk 1 q to p gonk 2 q would become, under this translation, a proof from
q to p ∧ q. There is no such proof. So, our rules for ‘gonk’ do not live up to
the conditions of being a definition, in Belnap’s sense. It fails to be uniquely
defining.
Concepts introduced by explicit definition are unqiuely defined, provided
that there is no equivocation in the definition. So, again, uniqueness seems like
a good criterion to look for if we want to characterise definitions.
It seems, too, that the usual rules for connectives like ∧, →, ¬ and the quan-
tifiers, fare better than these defective cases. Is there a way to show that they do
and that the rules are truly well behaved? It is straightforward to show that the
rules are uniquely defining. Here is the argument in the case for conjunction.
A ∧1 B A ∧1 B A ∧2 B A ∧2 B
∧1 E ∧1 E ∧2 E ∧2 E
A B A B
∧2 I ∧1 I
A ∧2 B A ∧1 B
This shows that concepts introduced by means of the conjunction rules are inter-
translatable, at least at the level of use as a premise or a conclusion in a proof.17

How could we show that the rules for conjunction, the conditional, negation
or the quantifiers are conservative additions to any prior language? One way
to attempt to show that they are is by way of an interesting property possessed
by these rules. They allow for us to eliminate detours, in the following sense.
Consider a proof where we introduce a conjunction and then eliminate it imme-
diately afterwards. There is clearly a quicker way to get to that conclusion, by
17 Does this mean that they have the same meaning, tout court? Possibly not. It is plausible that
‘but’ satisfies the same inference rules as ‘and’ when used as a sentential connective. However,
the implicatures involved using a ‘but’ instead of an ‘and’ are plausibly thought to be a part of
the meaning of the word, and not purely determined by the pragmatics, though there is much
more to be said there.
refraining to introduce the conjunction in the first place. We get to the same
conclusion, and in fact, we need not have used the premises that were featured
in the proof of the other, unused conjunct.
Π1 Π2
A B Π1
∧I simplifies to
A∧B A
∧E
A
The same sort of move can be made if we introduce a conditional and then,
immediately thereafter, eliminate it.
[A]
Π1 Π2
B simplifies to A
→I Π2 Π1
A→B A B
→E
B
Here, instead of assuming the antecedent of the conditional, to prove the conse-
quent, we use the proof of the antecedent we used for the elimination step. We
insert that proof wherever we formerly assumed the antecedent, and the result
is a proof of the conclusion, from the same premises as before. We can do the
same sorts of things for other connectives and for the quantifiers.
These are the normalisation steps for proofs in natural deduction. Notice that
whenever we normalise a proof by performing one of these transformations, it
gets simpler in some sense, by removing formulas that are more complex than
those before and after them. These formulas are ‘local maxima’ in complexity,
and they are ‘flattened out’ in the process of normalisation. The proof does not
always shrink in this process, since when we normalise some →I/E pair, the
proof Π2 of the minor premise A may be large, and it may be substituted into
the proof Π1 at more than one place. Regardless, it can be shown18 that for any
proof at all using these rules, we can normalise it completely so that the result
contains no detours.19
What does this mean for conservative extension? Each little detour is a place
where we made some steps from a premise to a conclusion through a particular
concept (some use of a conjunction, conditional, quantifier, etc.) that did not
also occur in either the premises or the conclusion. In normalising that step,
18 Prawitz’s monograph (1965) contains the proof.

19 The fact that the process of simplification of detours terminates and does not go on forever,
might seem obvious, but this requires a detailed argument. The simplification steps – in partic-
ular, the steps involving replacement of discharged formulas with entire subproofs, like that for
→I/→E—sometimes simplify a smaller proof involving a detour into a larger proof not involv-
ing that detour, but perhaps including more detours. To show that the process or reduction
terminates, you need to pay attention to the complexity of formulas.
we show that the appeal to that concept in that place is not required. What we
can show, in general, is that what happens locally at the site of each individual
step of normalisation obtains globally in the whole proof, once it has been nor-
malised. We can show that any normal proof has the SUBFORMULA PROPERTY in
this sense.20 Every formula in the proof is a subformula of one of the premises
or the conclusion of the proof. Normal proofs in this sense are analytic, they
proceed from the premises to the conclusion by way of the analysis into the
formulas’ components – no alien material is used.
The consequences for conservative extension can then follow. Suppose we
consider the rules for one of our connectives as a proposed definitional exten-
sion to our language, containing the other rules. We have seen that the rules for
negation are uniquely defining, so one of Belnap’s criteria is satisfied. But now
we can see that it is conservative, in the following sense: Consider a proof from
some premises to a conclusion in our new vocabulary, but where the premises
and the conclusion are in the old vocabulary, not using the newly introduced
concept. If we normalise the proof, the resulting proof is analytic, and so, since
the premises and conclusions do not feature the newly introduced concept, nei-
ther does any intermediate step in the proof. So, the proof does not use the
newly introduced rules, and so, it is a proof in the old language. Normalisation,
in this case, brings with it conservative extension.
So we have some degree of comfort that the logical concepts given by these
inference rules share features of concepts introduced by definition. They are
given an explicit semantics in a sharply defined way, and so, are useful wher-
ever it proves important to have sharply defined and agreed-upon rules. The
conservative extension argument shows that introducing such rules does no
violence to the pre-existing inferential practice, but adding new concepts can
give us new ways to reason with our original vocabulary.
2.3 Structural Rules

With this in hand, we can address philosophical issues in and around the theory
of proofs, including addressing the paradoxes we considered in Section 1. Our
argument puts pressure on the idea that the logical concepts are at fault in the
paradoxical arguments such as the Liar, since if the logical concepts are given
by definition, and if they are a free extension of the inferential practice in the
pre-logical vocabulary, then there is no point in pinning the blame on the logical
vocabulary – the problem must lie elsewhere. So consider these proofs. First,
20 This is an elegant result due to Dag Prawitz (1965).

a proof that a liar sentence (a sentence λ that says of itself that it is not true), is
contradictory:
λ = ⌜¬Tλ⌝ [Tλ]2
=E
T⌜¬Tλ⌝
TE
λ = ⌜¬Tλ⌝ [Tλ]1 ¬Tλ [Tλ]2
=E ¬E
T⌜¬Tλ⌝ ⊥
TE ¬I2
¬Tλ [Tλ]1 ¬Tλ
¬E TI
⊥ λ = ⌜¬Tλ⌝ T⌜¬Tλ⌝
¬I1 =E
¬Tλ Tλ
¬E
⊥
The new inferences in use are the truth rules TI/E, which make use of a
quotation term ⌜A⌝, for the formula A.
A T⌜A⌝
TI TE
T⌜A⌝ A
Our second example is the Curry paradox, introduced in footnote 5 on page 6.
Let ‘c’ abbreviate ‘{x : x ∈ x → p}’, a term for the class of all classes such that
if they are self-membered, then p. We have this proof:
[c ∈ c]2
∈E
[c ∈ c]1 c∈c→p [c ∈ c]2
∈E →E
c∈c→p [c ∈ c]1 p
→E →I2
p c∈c→p
→I1 ∈I
c∈c→p c∈c
→E
p
The novel principles here are ∈I/E, which state that t is in the class of all objects
x such that A(x) if and only if A(t).
A(t) t ∈ {x : A(t)}
∈I ∈E
t ∈ {x : A(t)} A(t)
These two paradoxical proofs have no rules in common. Regardless, they have
shared features. One such feature is the circularity or self-reference of the liar
sentence λ and the Curry class c. Perhaps the rules governing truth or class
membership should be modified in the presence of such circularity, or maybe
such circularity should even be banned. At the very least we should recognise
that the T and ∈ rules are not as conservative as rules for the connectives, quan-
tifiers and identity. We can prove ∃x∃y(x , y) using the T or ∈ rules, even
avoiding contraction. The connective, quantifier and identity rules alone do
not prove ∃x∃y(x , y), so neither the T nor ∈ rules are conservative over those
rules.
[p]1 [p]1
→I1 ∈I1
p→p p→p
TI ∈I
[⌜p → p⌝ = ⌜⊥⌝]2 T⌜p → p⌝ [{x : p → p} = {x : ⊥}]2 t ∈ {x : p → p}
=E =E
T⌜⊥⌝ t ∈ {x : ⊥}
TE ∈E
⊥ ⊥
¬I2 ¬I2
⌜p → p⌝ , ⌜⊥⌝ {x : p → p} , {x : ⊥}
∃I ∃I
∃y(⌜p → p⌝ , y) ∃y({x : p → p} , y)
∃I ∃I
∃x∃y(x , y) ∃x∃y(x , y)
Regardless, close analysis of the paradoxical proofs shows that despite the lack
of any shared inference rules, there are certain logical principles at work in both
proofs, and it is worth examining them more closely.
The most noticeable logical principle in common to both proofs is known as the
principle of CONTRACTION. In both proofs, two occurrences of the one assump-
tion are discharged at once. In the liar reasoning, the assumption Tλ is made
twice, to prove the contradiction ⊥. Both instances are discharged, to conlcude
¬Tλ. For the Curry paradox, the assumption c ∈ c is made twice, to prove p.
Both instances are discharged, to prove c ∈ c → p. If we impose a restric-
tion on discharging assumptions, to the effect that only one assumption may
be discharged at any time, then these paradoxical proofs would be ruled out
as incorrect. But why would such a restriction be plausible, and what kind of
restriction is it?
Contraction is one kind of structural rule. It is not a logical inference princi-
ple governing this or that connective, but rather, a principle governing proofs
as such, regardless of the content or structure of any of the statements in the
proof. Think of a proof from the premises A, B, C to conclusion D as a proof for
A, B, C D. This structure is called a sequent. Contraction is equivalent to the
idea that if we have a proof from for X, A, A B, then we also have a proof for
X, A B.21 Using duplicate discharge, we indeed see that whenever we have
a proof Π from X, A, A B then we can extend it into a proof for X, A B, as
shown here:
[A, A]1
Π
B
→I
A→B A
→E
B
21 Here, X is the collection of other premises used as the proof.

On the other hand, if whenever we have a proof for X, A, A B we can construct

a proof for X, A B, then we can get the effect discharging two occurrences
of A at once, contracting those two occurrences of A into one by moving from
the proof for X, A, A B to that for X, A B and then discharging the single
occurrence.
This principle, relating X, A, A B to X, A B involves statements as such
and has nothing to do with the particular content or structure of any of the
statements in question. As such, it is a purely structural rule, to do with the
structure of proofs as such, and not about the behaviour of this or that connec-
tive. It is just the kind of principle that might be lurking in the background,
behind paradoxical derivations like those we have seen.
Not only is the structural rule of CONTRACTION present in both of our para-
doxical derivations – we can show that if we prohibit it, then adding the T and
∈ rules to our proof rules for the connectives and quantifiers cannot result in a
proof of a contradiction. In fact, a more detailed analysis of proofs can show
that in the absence of contraction, there is no no proof of contradictions using
the logical rules and the T or ∈ rules. To do this, we can notice that proofs using
the T rules and the ∈ rules may be normalised, too, using these reductions:
Π Π
A A(t)
∈I Π
TI ⇝ Π t ∈ {x : A(x)} ⇝ A(t)
T⌜A⌝ A
TE ∈E
A A(t)
These reductions differ in one very important respect from the reductions for
connectives and quantifiers. The intermediate formulas (here T⌜A⌝ or t ∈ {x :
A(x)}) may be no more complex than the formulas on either side. In the case
where t ∈ {x : A(x)} is c ∈ c (from the Curry paradox) the formula is inferred
from c ∈ c → p, which is more complex than the introduced formula. Reducing
the proof does not involve cutting out a local maximum in complexity. The
reduction simplifies the proof by making the proof strictly smaller but not by
eliminating a local maximum in complexity.
The truth and membership rules, then, are well behaved from a proof-
theoretic perspective on one measure: normalising these detours shrinks the
proof. We’ve seen that in the presence of contraction, normalisation steps some-
times enlarge proofs, but the elegant result we can appeal to here is that if we
ban duplicate discharge, simplifying a detour always shrinks a proof, and so,
we can totally eliminate detours in proofs, even in the presence of the T and ∈
rules.22
22 In the presence of CONTRACTION, this fails. The paradoxical proofs, given on page 21 are not
normal. It is an enjoyable exercise to show that attempting to normalise them results in a cycle.
So, in the absence of contraction, we know that if there is a proof (even using
our T or ∈ rules) for X A, then there is a normal proof for X A. However, it is
straightforward to show that normal proofs satisfy the subformula property (as
mentioned on page 20). This has immediate and powerful consequences. There
can be no paradoxical derivations of arbitrary conclusions using these rules,
since there is no normal proof of an atomic formula p (or ⊥) from no prem-
ises. Since p and ⊥ have no subformulas at all, no introduction or elimination
rules could feature in any such proof satisfying the subformula property. So, the
addition of our truth or class rules cannot interact with our logical vocabulary
in this devastating way, if we reject CONTRACTION.

Once we isolate a structural rule like CONTRACTION, we notice that this dis-
cipline of proofs contains other structural rules. Another rule implicit in this
system of proofs is called WEAKENING:
XB
Weak
X, A B
We have passed from the stronger sequent X B to the weaker one. Believe it
or not, the rule of weakening is implicit in the conjunction rules ∧I/E. If I have
a proof of B, this quick detour through ∧I/E
Π
A B
∧I
A∧B
∧E
B
constructs a proof in which A is now present as an assumption, though it clearly

wasn’t needed to prove the conclusion B. With such a detour, we can construct
this proof
Π
[A]1 B
∧I
A∧B
∧E
B
→I1 ,
A→B
which leads from B to A → B, where the antecedent A might be totally uncon-

nected to the consequent B.The detour through conjunction makes the point
completely explicit, but it is not, strictly speaking, necessary. We could get the
Either of these proofs simplify to another proof that simplifies back into the original proof. This
process can continue indefinitely.
same effect by allowing for rules like →I discharge zero occurrences of an

assumption. The proof simplifies to
Π
B
→I1
A→B
Here, the →I step discharges zero occurrences of the assumption A. Such a

step is called vacuous discharge. So, contraction and weakening correspond to
duplicate and vacuous discharge, respectively. If we wish to avoid contraction
and weakening, then each discharging step discharges exactly one instance of
a formula.
The inference from B to A → B has been called a ‘fallacy of relevance’,
by those who expect the conditional A → B to record some kind of connection
between antecedent and consequent. Any such connection is lacking here. If we
wish the conditional to respect some kind of relevance, it is possible to discard
the structural rule of WEAKENING.
It is not hard to see that if we have a system of inference that attempts to keep
track of relevance, if when the premises in a proof are used in the deduction
of the conclusion, then we would like to resist WEAKENING. Another reason to
resist the rule of weakening is the kind of ‘default’ inference that allows us to
infer from ‘this is a bird’ to ‘this flies’, but if we add the premise ‘this is a pen-
guin’ then the inference is withdrawn. Typically, when considering deductive
inference, we generalise beyond default inferences to look for inferences that
have no counterexamples. The criterion of relevance, though, is less obviously
in tension with the canons of deductive validity.
Truly rejecting weakening involves modifying the conjunction rules, since
we have seen that they bring weakening in their wake.23 One simple modifica-
tion would be to replace the simple ∧E rules with this:24
[A]i [B]i
Π
A∧B C ′i
∧E
C
according to which, whenever we arrive at C in a proof in which both A and B

are assumed, we can instead discharge those assumptions and appeal instead to
23 So, you may notice that Prawitz’s result earlier, stating that the traditional rules for conjunc-
tion are conservative extensions over the base vocabulary assume WEAKENING over that base
vocabulary. If the language did not allow the inference q from p, q, then the addition of the
standard conjunction rules is no longer conservative, since we can reason from p, q to p ∧ q by
one step of ∧I and then to q by ∧E.
24 A rule of this form has come to be known as a general elimination rules. See von Plato (2001)
for a discussion of these rules.
A ∧ B. If we allow for vacuous discharge, then we can recover the effect of the
original ∧E rules:
A∧B [A]1 A∧B [B]1
∧E′1 ∧E′1
A B
However, without vacuous dischgarge, the new ∧E′ rule differs from its alter-
native, and it is a genuinely different treatment of conjunction that respects
relevance constraints.
There are two more structural rules that will be important in our discussions of
the paradoxes, proofs and models. WEAKENING and CONTRACTION concern the
behaviour of premise combination. The rules of IDENTITY and CUT focus on the
relationships between premises and conclusions:
XA Y, A B
A A Id Cut
X, Y B
Restricting the paradoxical arguments at the point of CONTRACTION is elegant

and uniform, but it comes with its own costs. For one, we appeal to duplicate
discharge everywhere in our reasoning, not only in paradoxical settings Rip-
ley (2015b). So, friends of contraction-free approaches to the paradoxes have
attempted to show how one might reincorporate contraction in safe non-
paradoxical settings, with limited success (Petersen, 2000, 2003). Another,
perhaps more fundamental concern, is philosophical. What does the single dis-
charge restriction mean? We are not used to accounting for our premises down
to the use, and it is not clear that such an accounting coheres with what we
are doing when we assert and deny. After all when we assert, one license we
grant is for others to re-assert that claim on the basis of our assertion (Brandom,
1983; Lackey, 2007). It would be strange if that right was not even extended to
ourselves, and the assumptions we make.
We can make one preliminary conclusion. When we look at paradoxical
proofs, we can see that CONTRACTION playing some kind of role. Exploring
proof systems that limit the use of contraction looks like one natural way to
respond to the paradoxes. However, incorporating this approach into a wider
account of what it is to assert, infer and reason – and giving some account of
when contraction fails – requires more work (Zardini, 2011).
We will turn our attention to other ways to respond to the paradoxes, inspired
by attention to the structure of the paradoxical proofs. Before we can dive into
those details, we need to address one further question about our inference rules,
and our assumptions about the structure of proofs in which those inference rules
are framed. And one way that this question could be asked starts with the rules
for negation.

Consider our rules for negation. We have a simple proof from p to its double
negation, ¬¬p:
[¬p]1 p
¬E
⊥
¬I
¬¬p
Could we reverse the inference to find a proof from ¬¬p to p? It turns out that
the answer is: no, so far. There is no proof, using our natural deduction rules,
that will lead us from the premise ¬¬p to conclude p.
What does this say about the argument from ¬¬p to p? Is it valid? (We can
defer discussion of the question of whether it has a counterexample to the next
section.) If it is valid, what kind of proof could lead us from ¬¬p to p? Cer-
tainly the kind of truth tables familiar from elementary logic classes would tell
us that the argument is valid, so finding a proof would seem like a pressing
task. What are the fundamental features concerning negation that lead us from
¬¬p to p?
There are at least three different responses to this challenge. The first is to
refuse it. To agree with the verdict of our proof system and to say that the
argument is, in fact, invalid. This is the response of the intuitionist (Brou-
wer, 1913; Dummett, 1977; van Dalen, 1986). The inference rules defining
negation are strong enough to sustain the principles of intuitionistic logic and
no more.
The second approach is to add some principle to bridge the gap. For example,
adding the inference of double negation elimination by itself:
¬¬A
DNE
A
or something like this ‘reductio’ principle:
[¬A]i
Π
⊥
Reductioi
A
would suffice. Adding an extra rule like this comes at some cost to the idea that
the inference rules are definitions. For example, using any of these principles
we can prove Peirce’s Law:
[¬p]2 [p]1
¬E
⊥
⊥E
q
→I1
(p → q) → p p→q
→E
[¬p]2 p
¬E
⊥
¬Ii
¬¬p
DNE
p
→I2
((p → q) → p) → p
However, there is no proof of ((p → q) → p) → p in our original system.

So, if our aim is to have an account of the logical principles that count as
definitions, then it seems that either the definition for ‘→’ is incomplete without
supplementation or the addition of extra rules for negation is an overreach.
Nonetheless, all is not lost. An insight from Gerhard Gentzen can give us an
alternative way to understand the underlying framework of inference, so that
the existing inference rules suffice to define the logical concepts. Gentzen’s
insight is that when we put the rules in an appropriately wider context, there
will be new ways to combine those rules, resulting in proofs from ¬¬p to p, to
prove ((p → q) → p) → p and more.
To understand that insight of Gentzen’s, we must take a step back from nat-
ural deduction proofs to keep track of the score at each step of a proof. A
proof (so far, at least) always leads from some collection X (possibly empty) of
premises to some conclusion A. We can represent the ‘score’ in what we call a
sequent, a structure the form X A. As a proof is developed, the score changes.
The short proof that consists of the assumption p corresponds to the sequent
pp
in which the p is both an assumption and the conclusion. As you read the proof
from p to ¬¬p from top to bottom, you can see the score develop in parallel:
[¬p]1 p pp
¬E ¬L
⊥ p, ¬p
¬I ¬R
¬¬p p ¬¬p
In this ‘sequent derivation’ you can see two principles at work:
XA X, A ⊥
¬L ¬R
X, ¬A ⊥ X ¬A
If you can prove a contradiction from X and A, then we can prove ¬A from
X (‘blaming’ the contradiction on A). On the other hand, if we can prove A
from X, then X together with ¬A would be contradictory. For Gentzen, we can
represent contradictoriness in a sequent by leaving out the ⊥ formula. The rules

have this shape:
XA X, A
¬L ¬R
X, ¬A X ¬A
where a sequent of the form X indicates the fact that the assumptions X
are contradictory. The negation rules swap the formula from one side of the
sequent to the other. Gentzen’s insight was that one could make a very simple
modification to these sequents to allow properly classical proofs without adding
ad hoc extra principles. All we need to do is to liberate sequents from the shape
X A (or X ), to include more than one formula on the right hand side, too.
Then, the negation rules would take the shape as follows:
X A, Y X, A Y
¬L ¬R
X, ¬A Y X ¬A, Y
and our derivation of p¬¬p has a natural mirror image, a derivation of ¬¬pp.
pp pp
¬L ¬R
p, ¬p p, ¬p
¬R ¬L
p ¬¬p ¬¬p p
The rules for the conditional generalise in the following way, where the only
change from the original setting is that we allow for more than one formula on
the right hand side.
X A, Y X′, B Y′ X, A B, Y
→L →R
X, A → B, X′ Y, Y′ X A → B, Y
The extra ‘room to move’ allowed in a derivation with more than one formula
in the right hand side provides a derivation of Peirce’s Law, without any detour
through negation:25
pp
Weakening
p p, q
→R
p, p → q pp
→L
(p → q) → p p, p
Contraction
(p → q) → p p
→R
((p → q) → p) → p
There is no doubt that sequent derivations of this form are an elegant and pow-
erful representation of some kind of ‘provability’ in classical logic. However,
25 Notice that there are two explicit appeals to structural rules in this derivation. One step of weak-
ening adds a q in conclusion position, and another contraction in conclusion position transforms
two instances of p into one. These are naturally ‘dual’ to the structural rules on the left hand
side of the sequent.
the question must be asked: is this just a trick, with no real interpretation (Milne,
2002)? After all, sequents of the form X A have a real meaning given in terms
of proof. The derivation of a sequent of the form X A indicates that there is a
proof from X to A. What does the sequent p, ¬p represent? It cannot mean that
there is a proof of either p or of ¬p, since there is no such proof (Steinberger,
2011).
One possible response is to say that a derivation of X Y indicates that
it is out of bounds to assert each member of X and to deny each member of
Y (Restall, 2005). This makes sense of the derivation of p, ¬p, since indeed,
there seems to be a mistake involved in denying both p and ¬p—after all, our
grounds for denying p in the relevant sense, seem to be grounds for asserting
¬p. This matches the interpretation of p, ¬p , which tells us that it is out of
bounds to assert p and to assert ¬p—after all, our grounds for asserting ¬p, in
the relevant sense, seem to be grounds for denying p.
More could be said about the assertion/denial interpretation of ‘X Y’
sequents, but we should be aware of one important criticism that has been
repeatedly made: this interpretation must be supplemented with at least some
account of inference, if it is to have any connection with the original inter-
pretation of sequents, for which a sequent represents a proof from premises to
a conclusion. After all, we don’t want the merely negative result that a der-
ivation of X A tells us that it would be a mistake to assert each member
of X and to deny A, without also saying at least something concerning what
we could do, were we to grant each member of X. One thing that the proof
interpretation tells us is that we could conclude A. The assertion/denial interpre-
tation does not make this connection, at least, not without further interpretive
work.
Whatever we may say about that criticism, the point remains that the
interpretation of sequents in terms of constraints on assertion and denial are
independently compelling, and they may be explored independently of any
immediate connection to natural deduction proof. With that connection in mind,
let’s reconstruct our paradoxical derivations in a sequent setting. To do this, we
need to use not only sequent rules for our connectives, but also consider the
rules for ∈ and T. These have natural left/right forms in sequents, corresponding
to the elimination and introduction natural deduction rules:
X, A Y X A, Y
TL TR
X, T⌜A⌝ Y X T⌜A⌝, Y
X, A(t) Y X A(t), Y
∈L ∈R
X, t ∈ {x : A(t)} Y X t ∈ {x : A(t)}, Y
For Curry’s paradox, we start with this derivation of c ∈ c p.

c ∈ c→pc ∈ c→p c ∈ cc ∈ c pp
∈L →L
c ∈ cc ∈ c→p c ∈ c, c ∈ c → p p
Cut
c ∈ c, c ∈ c p
Contraction
c ∈ cp
(Notice that it uses a Cut.) If we call that derivation ∆, we can extend it as

follows:
∆
c ∈ cp
→R
c ∈ c→p ∆
∈R
c ∈ c c ∈ cp
Cut
p
For the liar paradox, the derivation has a similar structure. Start with this
derivation of λ = ⌜¬Tλ⌝ ¬Tλ.
Tλ Tλ ¬Tλ ¬Tλ
=L TL
λ = ⌜¬Tλ⌝, Tλ T⌜¬Tλ⌝ T⌜¬Tλ⌝ ¬Tλ
Cut
λ = ⌜¬Tλ⌝, Tλ ¬Tλ
¬R
λ = ⌜¬Tλ⌝ ¬Tλ, ¬Tλ
Contraction
λ = ⌜¬Tλ⌝ ¬Tλ
and we’ll call this ∆′. (Notice that this derivation uses a contraction on the right
of the sequent, in addition to a Cut.) We can extend the derivation as follows:
∆′
λ = ⌜¬Tλ⌝ ¬Tλ
TR
λ = ⌜¬Tλ⌝ T⌜¬Tλ⌝
=L
λ = ⌜¬Tλ⌝, λ = ⌜¬Tλ⌝ Tλ
Contraction
∆ ′ λ = ⌜¬Tλ⌝ Tλ
¬L
λ = ⌜¬Tλ⌝ ¬Tλ λ = ⌜¬Tλ⌝, ¬Tλ
Cut
λ = ⌜¬Tλ⌝, λ = ⌜¬Tλ⌝
Contraction
λ = ⌜¬Tλ⌝
Here, truth, membership, identity and the connectives are governed by left and
right rules instead of introduction and elimination rules. As before, we can see
the role of contraction, though here it is a separate inference rule. Two other
structural rules loom into focus. The more obvious is the CUT rule, and the
second is the IDENTITY axiom A A, appealed to at each leaf of the tree.
We have already seen that IDENTITY and CUT are structural rules, since they
act on any formula whatsoever, and make no appeal to the content of a formula.
In terms of the bounds, the IDENTITY rule states that it is out of bounds to assert
A and deny A. The CUT rule is a little more complex: read from top to bottom,
it tells us that if it is out of bounds to assert each member of X and deny A and
deny each member of Y, and it is also out of bounds to assert each member of X
and assert A and deny each member of Y, then the problem, as it were, lies with
asserting each member of X and denying each member of Y, rather than with
A. To read this contrapositively, it tells us that if it is in bounds to assert each
member of X and deny each member of Y, then either adding A as an assertion
is in bounds, or adding A as a denial is in bounds.
The two rules of CUT and IDENTITY have themselves faced the harsh glare of
judgement, since they – like contraction – are also exposed as common factors
in paradoxical proofs. Consider Cut: Why is this a suspect? If the Cut formula
is paradoxical, then we may well think that a position may be in bounds, but
adding that paradoxical statement an assertion may be out of bounds (after all,
if I assert Tλ then I should deny it as well) and adding it as a denial is also
out of bounds (after all, if I deny Tλ then this is grounds for asserting it). This
seems like a good reason to explore going without the Cut rule, at least in its
full generality.
Doing without Cut may be more straightforward than it might appear,
because it has been shown that as far as logic is concerned, there is no require-
ment to appeal to Cut in the derivation of a sequent (Gentzen, 1935a,b, 1969),
since any sequent X Y that be derived using Cut can be derived without it.
(Cut elimination, in the sequent calculus, corresponds closely to normalisation
in natural deduction.) So far, so good. However, Cut is everywhere as far as
logic is applied. We may have derivations of T A and of T A → B, where T
is some given background theory, and A and A → B are consequences of that
theory. In the absence of Cut, we cannot conclude that B also folows from T.
We cannot, in general, chain statements about the bounds together in this way.
In the absence of Cut, paradoxical concepts like naïve truth and class mem-
bership are safe, in the sense that they do not allow for the derivation of
paradoxical consequences, but there is an important sense in which they are
inert. We cannot get far, using them, in the absence of Cut. And not only the par-
adoxical concepts themselves, but also the whole vocabulary, unless we impose
Cut for other vocabulary items that are taken to be ‘safe’.
This problem becomes more acute when we raise the issue, stated previously,
concerning the connection between this approach to the bounds on assertion
and denial and our initial concern for proof. If we think of a proof as licensing an
inference, in the sense of meeting a justification request for an assertion, against
the context of other assertions taken for granted (so, a proof from X to A shows
you how to meet the justification request for an assertion of A, appealing only to
the premises X), then the issue of the validity or the invalidity of Cut becomes
the issue of the composition of proofs, or the chaining together of inferences.
To put the question more starkly: if we are to resist the paradoxical inferences
by objecting to the Cut steps in our previously mentioned derivations, then
the corresponding move when it comes to the natural deduction proofs seems
downright weird. We grant the proof from A to B and we grant the proof from B
to C but we resist chaining them together. This puts severe pressure on the usual
interpretation of natural deduction proofs, especially when we conceive of them
as the meeting of justification requests. If a proof from A to B shows how I can
justify the claim B by appeal to A, and a proof from B to C shows how I can jus-
tify the claim C by appeal to B, then where has the obvious justification of C on
the basis of A – by first justifying B, and then on the basis of that, justifying C –
broken down? Breaking the proof at the Cut step means that either we reject the
sense in which we can use proofs to accumulate commitments, making explicit
what was implicit in what we have initially granted, or we must find some other
way to make sense of this form of accumulation and composition of proofs.

I indicated that the sequent derivation perspective on our paradoxical proofs

provides one other culprit. That culprit is not found at any step in the middle of
the derivation, but at the leaves. The Id steps in our proofs are the simplest start-
ing points: the atomic sequents from which all others follow. If we reject the Cut
steps because we took paradoxical claims like the liar to be both unassertible
and undeniable (a kind of gap), then the dual perspective, taking the paradoxi-
cal sentences to be both assertible and deniable (a kind of glut) would direct us
to have never started the derivation in the first place, for a sequent like A A
fails to hold in the case where A is paradoxical: it need not be out of bounds to
both assert the liar sentence and deny it. In that case, the paradoxical deriva-
tion does not get off the ground, because the identity sequent is not, in general,
valid. On this interpretation of the state of affairs, the Cut rule need not raise
any problems, since the issue is not that the liar sentence is too problematic to
assert or deny. It is rather that it is too accommodating to be ruled out. There
is no doubt that this interpretation of sequents allows for a distinctive way to
block the derivation. If we are not allowed to start, there is no way to derive our
problematic conclusion. Of course, in our sequent system, the identity axiom
is the only axiom, so we cannot prove anything, other than doubly conditional
conclusions. Our sequents are already, in some sense, conditional claims, and
now we cannot prove any sequent in a categorical sense, but we can prove them
conditionally, in the sense that (for example) if X, A B, Y were out of bounds,
so would X, ¬B¬A, Y. (Of course, we could posit particular axioms governing

particular vocabulary, taken to be safe, in the same way that the opponent of
Cut might impose particular instances of the Cut rule in just this sense.)26
If the issue of the interpretation of proofs was sharp in the case of rejecting
Cut, it is equally sharp, if not more so, in the case of rejecting Identity. To what
does the rejection of Id correspond in natural deduction proofs? It corresponds
to refusing to start in the first place. In the natural deduction setting, rejecting
the axiom of identity corresponds to rejecting the making of assumptions, for
the identity sequent A A corresponds to the natural deduction proof consisting
of the assumption of A, which is at the very same time the premise and the
conclusion of the proof.
What reason might we have for rejecting this principle? In the sequent set-
ting, we reject A A in the case where it might be in bounds to both accept
and to reject A, to assert it and to deny it at the very same time. In this case, we
may understand an atomic proof involving the single formula A as defective
because it does not constrain assertion and denial in the right way. If a proof
from P to C is to show that it is out of bounds to assert the premises and deny
the conclusion, then the so-called proof with premise A and conclusion A fails.
Since, on this view, it is in bounds to both assert A and to deny it, A should not
occur on its own as an atomic proof.
2.4 Other Proof Structures

There is much more that could be said about the application of proof-theoretical
techniques to the paradoxes, but before we finish this section, we will turn aside
to briefly consider proof-theoretical treatments of concepts beyond the famil-
iar logical connectives of conjunction, negation, the conditional and the like.
We have already seen the quantifiers, when we briefly discussed the role of the
universal quantifier introduction rule on page 14. You can see, when compar-
ing this inference rule to the connective rules, that quantifier rules involve an
analysis of sentences in our language in terms of predicates and singular terms,
and some notion of substitution of one term for another. The rules only make
sense relative to such an analysis. Proofs involving quantifiers involve a new
kind of structure, at the level of the formulas inside the proofs, when compared
to proofs involving the usual propositional connectives.27
26 See French (2016) for a further discussion of the issues surrounding the rejection of Identity in
sequent systems and the interpretation of the semantic paradoxes.
27 There are options to consider when it comes to quantifier rules like ∀I/E and ∃I/E. One choice
worth considering is whether our language allows for so-called undefined terms, and if so,
whether they can feature in inferences such as ∀E and ∃I. In a mathematical language in which
it is appropriate to make claims such as 01 is not defined, would it be appropriate to conclude:
Let’s briefly consider other concepts that are amenable to a logical analysis:
modal operators. The concepts of possibility and necessity have been subjected
to model-theoretic and proof-theoretic analysis in recent years. We will con-
sider models for possibility and necessity in the next section and attend briefly
to the issue of proofs for modality here. What could we do to prove □A, the
claim that A is necessary? What could we do to infer something from □A. One
option is to take our inspiration from the universal quantifier and take these to
be our rules:
X
Π □A
□E
A A
□I
□A
We have the proviso on □I: Each formula in X must start with a □. The elim-
ination rule is simple: From □A we can infer A. Conversely, if we can infer A
from premises that are themselves necessary, we can also conclude □A, that A
is necessary. For example, we can prove □(p ∧ q) from □p and □q, like this:
□p □q
□E □E
p q
∧I
p∧q
□I
□(p ∧ q)
These rules do define a modal logic28 However, we cannot extend the above-
mentioned proof into a proof from □p ∧ □q to □(p ∧ q) in the way you might
expect:
□p ∧ □q □p ∧ □q
∧E ∧E
□p □q
□E □E
p q
∧I
p∧q
□I
□(p ∧ q)
for the application of □I in this proof now no longer satisfies the constraint on
that rule. The formula □p ∧ □q does not feature a □ as its dominant operator.
∃x(x is not defined)? No (Feferman, 1995). This motivates a natural variation of the quantifier
natural deduction rules, which I discuss elsewhere (Restall, 2019).
28 It is the modal logic S4. If we allow active assumptions in a □I step to start with either a □ or a
¬□, this gives rise to S5.
We can construct a proof from this premise to that conclusion but only with
detours through other connectives:
[□p]1 [□q]2
□E □E
p q
∧I
p∧q
□I
□(p ∧ q) □p ∧ □q
→I1 ∧E
□p → □(p ∧ q) □p
→E
□(p ∧ q) □p ∧ □q
→I2 ∧E
□q → □(p ∧ q) □q
→E
□(p ∧ q)
So rules like these are not particularly pleasing, and in general, natural deduc-
tion and sequent calculus systems for modal logics are difficult to work
with.
One alternative proof-theoretic approach is inspired by the treatment of the
quantifiers. Different kinds of logical concepts involve different features of our
language and how we use that language. Quantifiers trade on terms and substi-
tution, features that play no role in proofs using just the standard propositional
connectives. Modal operators trade on something else. Let’s start with an elim-
ination rule for necessity. It is true that from □A we can infer A, but this rule
far from exhausts the power of the claim of necessity. If we have granted that
A is necessary, we can not only infer A ‘here’ (as it were), we could also infer
that it would hold, were things to be different. If we structure our dialogue (and
our proofs) so that we can allow hypothetical shifts like this, then we could
represent the power of the claim to necessity in an appropriately explicit way.
Let’s represent ‘contexts’ in our proofs by tagging each line with a label. Lines
with the same label are claims made in the same context. If the label changes,
we move to a different context, which may be represented in dialogue by words
like ‘suppose things had gone differently’, and the like. The rules for the stand-
ard connectives require the context to remain fixed. The rules for the necessity
operator allow for contexts to shift. They look remarkably similar to the quan-
tifiers, but instead of quantifying over objects, we generalise across contexts in
dialogue.29
X
Π
A · i □A · i
□I □E
□A · j A · j
29 See Read (2008, 2015) for more on proof systems of this form.
Here, in □E, we allow the context index to shift arbitrarily, from any index, to
any index. In □I, we also allow the context to shift arbitrarily, but only when
the index i tagging the formula A does not occur in the assumption set X. Using
these rules, we can represent the proof from □p ∧ □q to □(p ∧ q) more naturally:
□p ∧ □q · 1 □p ∧ □q · 1
∧E ∧E
□p · 1 □q · 1
□E □E
p · 2 q · 2
∧I
p∧q · 2
□I
□(p ∧ q) · 1
This reasoning corresponds to the informal rendering given later in the text,
where I have made explicit the context shifts as the reasoning proceeds.
[C1] Suppose it’s necessary that p and it’s necessary that q. Here’s why it’s
necessary that p and q. Suppose things go differently [C2]. Since [C1] it’s
necessary that p, we have [C2] p. Since [C1] it’s necessary that q, we have [C2]
q. So, [C2] we have both p and q, and since we made no special assumption
about how things go [C2 is arbitrary], we can conclude [C1] that it is necessary
that p and q.
The natural deduction proof is one way to make explicit and precise the kinds
of shifts we can do in everyday reasoning. This can be represented in a sequent
calculus as well (Poggiolesi, 2008, 2009, 2010; Poggiolesi & Restall, 2012;
Restall, 2012).
The similarity to the rules for quantifiers is striking, and it has led some to
ask the question: are modal operators really just quantifiers? If we think of
the context markers as representing items, then it is natural to think of them
as worlds, at which claims are true or false. If we were to model necessity as
a quantifier over worlds, then we could save on rules for logical vocabulary.
One consequence of this view is that the semantic structure of claims becomes
more complex. The claim of the form ‘Socrates is talking’, which was formerly
a predication Ts, and which might be asserted or denied in different contexts,
becomes a binary predication, Tsw (Socrates is talking in world w), which then
has the appropriate form over which the quantifier can operate. Is this the cor-
rect form to analyse our language and the structure of our claims about the
world? That’s one point of difference to be explored, concerning the scope and
significance of modal vocabulary.
Notice that this question is related to, but is distinct from, the question of the
ontology of possible worlds. You could reject the notion that possible worlds are
real existences and still think that modal operators are disguised quantifiers that
range over worlds. (You just need to give a story about quantifiers over worlds
that works for the metaphysically thin understanding of worlds that you have.)
More on this in the next section where we discuss possible worlds semantics,
so it is to this that we now turn.
3 Models
Proofs were our focus in the previous section. A proof is a certificate for the
validity of an argument. If I have a proof from premises to a conclusion, that cer-
tifies that the argument from the premises to the conclusion is valid. Models –
our focus for this section – help us view the other side of the coin. If you want
a certificate for the invalidity of an argument, you want a counterexample. A
counterexample to an argument is a model, which shows how to satisfy the
premises without at the same time satisfying the conclusion.
So, if we start with this role of models as counterexamples to arguments,
we see that at the very least, we want a model to specify when it satisfies a
formula (let’s write this ‘m ⊩ A’) and when it doesn’t (let’s write this ‘m ⊮ A’).
Then, we will say that the model m is a counterexample to the argument from
the premises X to the conclusion A just when m ⊩ B for each B ∈ X, and
m ⊮ A.
3.1 Two Truth Values

From here it is a very short jump to the idea of classical two-valued models,
which give rise to what we call ‘truth tables’. Select two items (most commonly
‘1’ and ‘0’) to represent truth and falsity, and for every atomic formula, we
specify the truth status of the formula by fiat. A model is specified by a choice
of truth value for each atomic formula. A model m is a function assigning a
truth value to each atomic formula. m(p) = 1 or 0, for each atom p. Then, given
this function assigning values to atoms, we can specify how the model assigns
values to the other propositional formulas in a natural way:
A B ⊥ ¬A A∧B A∨B A→B
0 0 0 1 0 0 1
0 1 0 1 0 1 1
1 0 0 0 0 1 0
1 1 0 0 1 1 1
The table indicates how the logical constants ⊥, ¬, ∧, ∨ and → interact with
two-valued models. The formula ⊥ receives the value 0 in every model. In any
model m, m(¬A) is the opposite value to m(A). m(A ∧ B) = 1 when and only
when m(A) = 1 and m(B) = 1 and so on.
So, a model m assigns values 1 or 0 to every formula. We will say that a

model m verifies a formula A (m ⊩ A) iff m(A) = 1, and so, m ⊮ A iff m(A) , 1,
which means here that m(A) = 0. The result is a classical two-valued logic. We
say that X ⊨TV A if and only if there is no model m that is a counterexample to
the argument X A.
So, now we have models and proofs as two different analyses of validity. The
soundness theorem states that if we ever have a proof from X to A, then indeed
X ⊨ A. The completeness theorem is its converse, and the most straightforward
way to demonstrate completeness is by proving the converse: if X ⊬ A then we
construct a model as a counterexample to the argument from X to A.30
How does this compare to the discussion of the last section? Let’s consider first
the treatment of the logical connectives. The meaning rules for the connectives
are given in terms of truth conditions, rather than introduction and elimination
rules. They are similar: very similar. There is a very short step from ∧I to the
left-to-right direction of the biconditional between m(A) = 1 and m(B) = 1 and
m(A ∧ B) = 1. However, one is stated in terms of truth-value assignments and
the other in terms of inference. The same parallel but difference holds for the
rules for the other connectives. We can see the similarity if we either understand
the inference rules as concerning truth (read ∧I with a sotto voce ‘is true’ after
each claim, rather than merely asserting them), or conversely, if you think of
the values 1 and 0 as ways of specifying ideal positions a reasoner might take,
concerning claims. There is clearly an interpretive distinction between how
we understand proofs and models (to be explored in the next section) but the
formal results concerning soundness and completeness should reassure us that
connections can be drawn between these two approaches.
The second differerence between the two presentations of the rules is that the
ambient background context in one case is the two-valued assignment system
that assigns one of two values to each formula, and in the other, the background
assumptions about proofs and how they can be composed. In the next section
we will explore what we might say about the connections between these two
understandings of the background context. For now, we will focus on what
insight we can gain concerning the paradoxes, when we pay attention to models.
Let’s start with the treatment of vague language and the sorites paradox. If
we look at the sorites paradoxical argument, introduced in Section 1, then we
can see that when we analyse the sorites argument (8, see page 7),
(8′) R1
R1 → R2
30 We will not go into the details of how to prove soundness and completeness here.
Random documents with unrelated
content Scribd suggests to you:
que, volviendo las espaldas a Europa, mirase resueltamente hacia el
Pacífico».[20]
Dejando para tratar más adelante en párrafos especiales de otras
cualidades que significan también factores importantes en la vida
económica digamos dos palabras sobre la educación.
La educación norteamericana merece sin duda casi todas las alabanzas
de que se la hace objeto y ya he dicho que se halla admirablemente
organizada para preparar al pueblo a la lucha económica. No es posible
entrar aquí en detalles completos sobre el particular, ni estaría tampoco en
situación de hacerlo. La educación vocacional, o sea la que tiende al
desarrollo de la eficiencia económica de los individuos, es materia de una
atención constante en todos los grados de la enseñanza. Grandes institutos
superiores están consagrados exclusivamente a este fin, como ser el
Instituto de Tecnología de Massachussets en Cambridge, que ya he
mencionado, un instituto para investigaciones científicas relacionadas con
la industria, anexo a la Universidad de Pittsburgo, los Colegios de
Agricultura y Minería de la Universidad de California, los Colegios de
Agricultura y Química Industrial de la Universidad de Wisconsin, el
Sheffield Colegio de la Universidad de Yale, y el Departamento de Física de
la Universidad de Princeton.
En otro capítulo he mencionado el Liceo Técnico de Oakland. Como
éste o si no tan completos, semejantes establecimientos se encuentran en las
principales ciudades para dar preparación técnica y comercial a la juventud
en la segunda enseñanza.
Fuera de lo que se hace directamente por medio de la instrucción
vocacional, son propios de la educación americana dos caracteres muy
favorables al fomento de la eficiencia económica, a saber los métodos
activos y prácticos y el sistema de ramos elegibles y no de plan fijo, como
existe entre nosotros. Cada alumno puede elegir para estudiar en el curso de
cuatro o seis años los ramos que más convengan a sus inclinaciones y
aptitudes intelectuales. Todos los alumnos deben estudiar una misma
cantidad de tiempo; pero no todos se hallan obligados a estudiar las mismas
cosas ni todas las cosas. Esta organización, junto con los métodos que he
mencionado, tiende poderosamente a estimular la iniciativa, el trabajo con
gusto, y, por consiguiente, la actividad y la eficiencia.
En relación con la iniciativa, no debemos silenciar la inventiva, como
una de las cualidades características del pueblo yanqui, que debe ocupar un
lugar muy importante entre las causas creadoras de su potencia económica.
Ya a mediados del siglo XIX, un comisionado de la Cámara de los
Comunes decía: «Debo expresar que la mayor parte de los nuevos inventos
introducidos últimamente en este país (Inglaterra) han venido de afuera; y
que, en especial casi todas las ideas relativas a maquinarias nuevas o al
perfeccionamiento de las existentes, han tenido su origen en Norte
América».
Cuando se habla del valor de la iniciativa individual, de cómo se la
desarrolla en los Estados Unidos y de que éste es un país de promisión para
ella, uno no puede dejar de acordarse al mismo tiempo de las
organizaciones que la limitan. Estas son los trusts, que nacieron poco
después de 1880,[21] como grandes combinaciones de compañías
industriales y comerciales encaminadas a alcanzar los siguientes principales
fines: suprimir la competencia; regularizar y obtener economías en la
producción; división de los mercados y mantenimiento de los precios. Estas
organizaciones gigantescas arrollan y aplastan sin dificultad a toda empresa
que asome en el mercado con pretensiones de hacerles competencia y no
dan lugar, es claro, a más iniciativas que a las que puedan desenvolver sus
asociados y empleados dentro de las férreas mallas de la institución misma.
En estas condiciones, no ha sido difícil que los trusts fueran culpables
muchas veces de los abusos del monopolio. Todo monopolio es odioso y los
trusts han dado bastante que hacer a gobiernos y legisladores en los últimos
treinta años. Se pensó primero en combatirlos levantando al frente de cada
trust otro que le hiciera competencia; pero este procedimiento no dió
resultados.
Uno de los trusts quebraba y las cosas seguían a veces peor que antes.
Dejarlos en completa libertad no era posible y menos aun eliminarlos
enteramente. No ha habido otro camino que someterlos a regularizaciones
establecidas por la ley, que es lo que se ha verificado.
Con anterioridad a la organización de los trusts y luego en combinación
con ellos, se ha operado una gran concentración de las riquezas en las
manos de unas pocas familias. De una estadística hecha sobre la propiedad
rural y urbana en los Estados Unidos en 1890, se ha sacado en claro que el
noventa y uno por ciento de las familias del país no poseen más del
veinticinco por ciento de la riqueza y que el nueve por ciento de las familias
tienen en sus manos el setenta y cinco por ciento de las riquezas.
A propósito de estos hechos, dice Bogart con cierta amargura: «Es
evidente que nos hemos alejado mucho de las condiciones democráticas de
una relativa igualdad económica que existía hace setenta y cinco o cien
años. Las fortunas colosales son un fenómeno del siglo XX, y muchas de
los Estados Unidos se han formado gracias a que los recursos naturales han
sido monopolizados por unos pocos que se han hecho ricos con el aumento
de la población y la consiguiente mayor demanda. Situados esos individuos
en puntos estratégicos del mundo industrial, han arrancado tributos a la
sociedad ni más ni menos como lo hacían los caballeros bandidos de otros
tiempos. Bajo esta circunstancia, el aumento de la riqueza no ha significado
ni una eficiente utilización de los recursos del continente, ni una mayor
ventaja social derivada de su uso. El aumento de la producción ha venido
acompañado de aumento de desigualdad en la distribución».
La distribución desigual va acompañada generalmente de una pobre
economía y de baja eficiencia en la obra de la producción. Mientras por un
lado se ha desarrollado en el país un espléndido tipo de organizador de
negocios, hay que notar por el otro las pérdidas sociales que resultan del
exceso de trabajo, de la labor de los niños, de las enfermedades
profesionales, de los accidentes y del vivir en arrabales y malas
habitaciones. Una mejor organización de la industria y de la distribución de
la riqueza evitaría muchos de estos males y elevaría el nivel general de la
eficiencia. Gran desigualdad significa también menos placer en el consumo
de la riqueza. Lo que se necesita es no solamente más y más riqueza, sino
que ésta sea de tal manera distribuída que el bienestar, la satisfacción
económica de la sociedad pueda ser llevada a su más alto grado. La suma
total de las satisfacciones económicas tiene que ser mayor en una sociedad
donde reine una justa igualdad de distribución que en otra donde el
millonario se codee con el vagabundo. Da prueba, por último, de cierta
inconsistencia, por no decir que se halla amenazada de un peligro, la
sociedad que es en el orden político democrática, y, al mismo tiempo, en el
económico plutocrática».[22]
Sin embargo, con motivo de la guerra europea tuvo el Gobierno que
entrar por el camino de tomar medidas consideradas por algunos como
propias de un estado socialista, como ser la apropiación y manejo de los
ferrocarriles, y de otras inspiradas en un amplio criterio de justicia social,
como la ley que ha venido a gravar la renta y llegado hasta imponer el
sesenta y cinco por ciento de contribución a las más grandes fortunas.
Aunque nunca haya sido el socialismo una entidad política
preponderante en los Estados Unidos, no por esto deja de inspirar cuidado.
«Hasta ahora no es muy extenso el número de adeptos oficiales al partido
socialista en nuestro país, dice N. Murray Butler, pero las doctrinas y
enseñanzas del socialismo se están difundiendo sistemática y
vehementemente entre nosotros. Muchas escuelas y colegios y muchos
púlpitos se dedican consciente e inconscientemente a tal labor. En las
elecciones de 1916, el partido socialista de los Estados Unidos obtuvo casi
exactamente el 3.3 por ciento de la votación total. Es probable que a causa
de haber adoptado oficialmente la política internacional de los bolchevistas
rusos se haya alienado el partido socialista la simpatía de sus primitivos
defensores, hasta el punto de que los votos a su favor se redujeran a menos
del dos por ciento. Por pequeño que parezca este número representa una
organización y actividad desproporcionada con su importancia. No es
posible equivocarse respecto de su programa. Califica abiertamente de
inmoral nuestra Constitución. Acusa de mercenarios, de pillos, de hombres
de inteligencia mediocre y de abogados de la clase capitalista a los padres
de la nación. La plataforma del partido socialista en 1912 demandaba
explícitamente no sólo la política comunista, sino la abolición del Senado
de los Estados Unidos y de la facultad del Presidente de imponer el veto; la
abolición de los tribunales federales, con excepción de la Corte Suprema y
la elección de todos los jueces por corto tiempo»[23].
Tampoco faltan los anarquistas, como lo prueban los numerosos
atentados dinamiteros llevados a cabo en 1919, en las principales ciudades
de la Unión, en contra de miembros de la judicatura de financieros y de
hombres públicos. Entre las asociaciones obreras es tenida por francamente
maximalista la de los I. W. W. (Industrial Workers of the World.
Trabajadores industriales del mundo).
Los detalles apuntados sobre los trusts y los anarquistas manifiestan que
el espléndido cuadro económico y social de los Estados Unidos no carece
de sombras. Es condición de los extranjeros dejarse deslumbrar por las
grandezas que encuentran fuera de su patria y exagerarlas para presentarlas
con la mejor intención como modelos. Esto les ocurre a muchos sud-
americanos, y no sería raro que en especial a los chilenos, con las
magnificencias de los Estados Unidos. El edificio económico de este país es
sin duda imponente y superior a cuanto se ha visto hasta ahora en la
historia, pero, como toda sociedad humana, sin excepción, la norte-
americana no puede sustraerse a la ley de que vayan apareciendo
desperfectos en ella, que engendran descontentos y reclaman reformas.
El espíritu democrático.—Falta de ceremonias y rudeza de
maneras.—En otro capítulo he hablado ya de las instituciones de la
democracia americana y en este momento no cabe sino recordar cómo su
espíritu se muestra en la manera de ser de los americanos.
Este es un pueblo muy poco afecto a las ceremonias. Recién llegado a
Berkeley, le pregunté a un profesor de la Universidad de California, cómo
debería tratar a un decano que me acababan de presentar y si era la
costumbre dirigirse a ese funcionario con su título más alto y llamarlo
Decano Jones o decirle simplemente profesor Jones. «Oh no, fué la
contestación», llámelo Mr. Jones y basta. Nosotros somos un pueblo sin
ceremonias. Cuando vaya a ver al Presidente de la República no se
preocupe de fórmulas de especial cortesía ni de nada semejante y dígale con
toda naturalidad «¿Cómo le va, Mr. Wilson?».
No es raro que a las personas a quienes se ha invitado a tomar el lunch o
a comer las conviden también los dueños de casa, después de los postres, a
lavar junto con ellos los platos en la repostería o en la cocina. Es este, sin
duda, un caso de falta de ceremonias; pero es igualmente un resultado de la
falta de servidumbre que proviene de la elevación creciente de las clases
sociales bajas y de las oportunidades que se les ofrecen de ocuparse en
mejores condiciones en las fábricas y talleres.
Los hombres no se sacan generalmente el sombrero para saludar a otros
hombres y ni los estudiantes proceden de otra manera con sus profesores.
Los comerciantes suelen no ser muy amables. Quieren despacharse
pronto y no se insinúan con el cliente. A veces dan ganas de quedarse
descalzo antes que comprarle botines a un sujeto que se le pone a uno por
delante y parece decirle con su actitud: «No se demore tanto, señor, apúrese,
elija pronto, pague y váyase». En cierta ocasión un joven sud-americano,
estaba comprando una maquinita para afeitarse y pedía explicaciones sobre
su mecanismo. «No pregunte tanto, señor, le contestó el vendedor, ejercite
su propio juicio».
De regreso de Washington a Nueva York nos quedamos un día en
Baltimore. Buscamos una pieza en uno de los mejores hoteles de la ciudad
y como nos pidieran seis dólares por ella al día, sin comidas, quisimos verla
antes de tomarla. Subimos cinco o seis pisos. No era muy buena; pero, en
fin, es por un día, nos dijimos mi esposa y yo. Era pleno verano; el día de
mucho calor y veníamos sofocados por el viaje en tren. Pedimos agua fresca
para beber y nos dispusimos a lavarnos. En esto suena el teléfono de la
pieza. Era el administrador que me llamaba desde la oficina.
—No ha dicho usted si se queda con la pieza.
—Si me quedo, le contesté.
—Baje entonces a inscribirse en el libro del establecimiento.
—En cuanto me lave bajaré.
Poco después entró el mozo con el agua fresca y me dijo:
—El administrador dice que baje a inscribirse, señor.
—Ya le he contestado que en cuanto me lave bajaré.
Al poco rato nuevamente el repiqueteo del teléfono. Otra vez el
administrador.
—Ya le he dicho, señor, que baje a registrarse, me observó en un tono
que no tenía nada de amable...
—Y yo le he respondido que en cuanto me lave bajaré repuse, y corté la
comunicación.
Después de unos pocos minutos golpean a la puerta.
—Adelante.
Nadie entra. Nuevos golpes.
—Adelante, exclamé con más fuerza.
Nadie entró tampoco. En mangas de camisa y sin cuello tuve que abrir
yo mismo la puerta. Era el administrador en persona que venía a insistir en
que bajara inmediatamente a inscribirme. Le repetí lo que ya le había dicho,
agregándole que no me explicaba su proceder que encontraba muy
descortés. Le dije también quien era (aunque esto en verdad no significaba
un argumento de mucho peso) y le presenté mi tarjeta.
—No tengo nada que ver con quien sea usted, me contestó; o baja
inmediatamente a registrarse o deja la pieza.
—No cabe la menor duda, señor, le dejo la pieza.
No era grato quedarse en una casa dirigida por semejante energúmeno.
Guardamos precipitadamente los objetos de tocador y ropa que ya habíamos
sacado de las maletas, arreglamos nuestras personas como pudimos y nos
marchamos. Me fuí pensando que seguramente los reglamentos de policía
exigen que todo pasajero se inscriba en el registro del hotel antes de
instalarse; pero que ese administrador era un majadero cuyo cerebro
estrecho no le consentía elasticidad para aplicarlos y llegaba hasta la
grosería en lo que él estimaba el cumplimiento de su deber.
No son escasas las veces, por otro lado, en que los empleados de tiendas
y almacenes, especialmente algunas niñas, dan prueba de una paciencia
estoica y de amabilidad sincera. Después que el cliente solicita los precios
de cuanta cosa se le ocurre, lo revuelve y lo mira todo, hace perder media
hora y no compra nada, la empleada lo despide con la más dulce sonrisa
diciendole:
«I see you again, come again», «hasta la vista, venga otra vez.».
Actividad.—Es un hecho reconocido que el clima de los Estados
Unidos, continental, variable y crudo como es, posee grandes virtudes
vigorizantes y estimulantes. Un viejo marino inglés decía que él se tomaba
una botella de vino en cada comida en Liverpool, y no podía, por efecto del
clima, tomar más de media botella en Nueva York.
En un párrafo anterior hemos visto cómo este clima ha transformado
favorablemente a los retoños de las razas europeas transplantadas al Nuevo
Mundo.
Aquí tenemos indicados dos de los antecedentes explicativos de la
actividad norte-americana: el clima y la raza. Habría que tomar en cuenta
también la acción de las instituciones democráticas. Ya hemos dicho en una
página reciente de qué manera éstas han obrado como focos de absorción
para atraer a las gentes de todo el mundo, seduciéndolas con sus
oportunidades de vida libre y próspera. El federalismo y la falta de
centralización son aspectos muy importantes de este mismo fenómeno. En
los Estados Unidos no hay una sola gran capital, un gran centro que sea
considerado por los habitantes del país como el único lugar posible para
vivir, y hacia el cual vuelvan todos la atención de sus espíritus dócilmente
imantados. No ocurre lo que en Francia y la Argentina donde la gente se
disloca por París y Buenos Aires, ni tampoco lo que pasa en nuestro
pequeño mundo de Chile, cuyos habitantes provincianos parecen sufrir de
una especie de desviación espiritual a fuerza de mirar tanto hacia Santiago y
de suspirar por él. En los Estados Unidos, Washington, la capital federal, no
puede competir, en cuanto centro de atracción, con los grandes emporios del
comercio y de la industria; y el hecho de que Nueva York sea una metrópoli
colosal de siete millones de habitantes, no quita que muchas ciudades como
Chicago, Filadelfia, Baltimore, Boston, San Francisco, Los Angeles,
Buffalo, San Luis, y varias otras, constituyan centros importantes,
autónomos en cuanto a sus elementos de cultura y con caracteres distintivos
y propios. Muchas de estas condiciones se encuentran aún en pequeñas
ciudades universitarias como New Haven, Berkeley, Ithaca, Princeton, Palo
Alto, etc. De aquí resulta en general la existencia por doquiera de un
ambiente favorable a la actividad porque el norteamericano encuentra en
cualquier parte los elementos de una vida completa y hace dar a su
iniciativa donde le toca el mayor rendimiento posible, sin que necesite
esperar algo de un centro más o menos remoto.
Al lado de la influencia externa de las instituciones en el desarrollo de
las iniciativas individuales es menester no olvidar la de la educación, que
tiende a la formación de personalidades independientes y vigorosas. La
educación, obrando de consuno en muchos casos con cierta religiosidad
tradicional y honda, llega a sugerir un concepto serio de la vida, apartado de
frivolidades y que busca la realización del destino humano en el
cumplimiento del deber y en las satisfacciones que ofrece una existencia
interior armónica y tranquila.
La actividad norteamericana lleva generalmente como perfume propio
una sana alegría, según hemos tenido ocasión de hacerlo ver más de una vez
en estas notas.
Patriotismo.—Los norteamericanos han dado múltiples pruebas de
patriotismo en el curso de su historia y los ideales nacionales han sido
resortes poderosos en muchas de sus empresas de carácter político y
económico, como ser la lucha contra la esclavitud, la colonización del Oeste
y la ocupación de Oregón hacia 1842. Después de haber fijado los límites
entre los Estados Unidos y Canadá, hasta los montes Roqueños, había
quedado en suspenso la decisión relativa a las tierras del Oeste, o sea
principalmente Oregón. No inspirando confianza la actitud de Inglaterra al
respecto, los norte-americanos se lanzaron a ocupar de hecho ese Estado, y
gracias a su empuje quedó incorporado a la Unión.
En los días en que estos apuntes fueron tomados las manifestaciones de
patriotismo habían pasado a ser detalles de la vida cotidiana. A donde
volviera uno la vista encontraba la bandera estrellada flameando sola o en
compañía de las de las naciones aliadas. En las ventanas de las casas se la
veía frecuentemente al lado de la pequeña banderola en que se indicaba con
estrellitas cuantos miembros de la familia habían ido a la guerra. A veces
las estrellitas eran negras, lo que significaba que ellos ya no volverían más.
En las oficinas públicas y privadas y en los almacenes, grandes banderas, a
menudo con centenares de estrellitas, decían la contribución de sangre que
el personal había enviado al Ejército de la Patria. Todo espectáculo
empezaba invariablemente con el himno nacional, ejecutado por la
orquesta, a veces cantado, y que la concurrencia escuchaba con
recogimiento de pie.
La cooperación de todas las clases sociales en esos momentos supremos
era manifiesta. Ya he indicado en otro capítulo de qué manera se levantaron
los primeros empréstitos de la libertad en San Francisco. Una vez terminada
la guerra le llegó el turno al último de ellos, al de la victoria, destinado a
saldar la situación. Fué como un levantamiento general de las poblaciones
en un movimiento arrebatador de festividad patria y de civismo
intensificado por los mil recursos de la propaganda americana. Las amplias
calles de altísimos edificios ricamente embanderados y los faroles
adornados con los colores nacionales que terminaban en panoplias de
flámulas y gallardetes parecían estar de gala para recibir a un ejército
victorioso.
La operación se inició con un discurso pronunciado por un alto
funcionario administrativo a mediodía en las gradas del edificio del Tesoro
en Nueva York. Los acordes marciales de una banda militar detuvieron por
un instante la inmensa corriente de gente que a esas horas llenaba Wall
Street y las inmediaciones de la Bolsa, y el orador, bajo un sol de verano y
al lado de la estatua de Washington, expuso las razones de suscribir con
creces el empréstito de la victoria.
Luego millares de oradores continuaron insistiendo en todas partes sobre
el mismo asunto durante una semana. Tinglados y tabladillos se levantaron
en las principales plazas y en las más concurridas boca-calles de todas las
avenidas donde pequeñas bandas de músicos tocaban para atraer a la gente
a oir la oración patriótica de los propagandistas del empréstito. Los
guardianes que dirigen el tráfico callejero llevaban en la mano un disco con
una V, que se levantaba sobre la muchedumbre cada vez que ellos paraban o
hacían continuar el movimiento, para hacer presente la victoria y la
necesidad que había de pagarla. Agentes del empréstito recorrían las
grandes arterias en autos embanderados y tocando cencerros. Desde garitas
levantadas en las esquinas, damas ofrecían bonos a todos los transeuntes; y
vendedores de todas clases, hombres, mujeres y niños, lo asaltaban a uno en
las aceras y en los hoteles. En todos los teatros, hacia la mitad del
espectáculo un orador pronunciaba un discurso sobre el empréstito y luego
empleadas de la empresa iban a vender bonos entre la concurrencia. A
veces el orador mostraba al público un casco prusiano que decía haber sido
tomado a un oficial del Kaiser y lo ofrecía como premio al que comprara
una suma alzada en bonos. No pasaba mucho rato sin que el casco fuera
entregado a algún señor de la platea o de un palco. Se había organizado una
sugestión inmensa, irresistible para saldar la victoria; las ciudades
competían unas con otras a fin de sobrepasar la cuota que se les había
señalado y de esta suerte se suscribieron muchísimos más millones de los
fijados por el Gobierno.
Moralidad (la familia y el divorcio.)—Me atrevo a apuntar como
cualidades concomitantes de la actividad norteamericana en términos
generales a la moralidad y la religiosidad. No se me oculta que si es
arriesgado hablar de la psicología de un pueblo, no lo es menos hacerlo de
la moralidad. Nunca faltan razones y datos para sostener en este terreno las
tesis más opuestas. En contra de la moralidad americana se recuerdan los
estupendos golpes de mano que suelen dar los criminales de aquel país. A
no pocos he oído quejarse de la falta de honradez y seriedad de los
comerciantes yanquis. No obstante algunos incidentes apuntados en estas
notas, que pudieran pesar en contrario, mi impresión general, formada con
detalles que también he referido y con el trato de gente principalmente
universitaria, es que los norte-americanos son por lo común buenos y
honrados. Llegaría a decir que en la apreciación que este pueblo hace de las
facultades y valores humanos prefiere la bondad a la habilidad.
Se suele afirmar también que en los Estados Unidos la familia casi no
existe, que se halla en plena disolución. En verdad no existe con los
caracteres tradicionales con que se suele presentar en algunos pueblos de
Chile donde se ven casos de que vivan sólidamente unidas en un mismo
hogar tres generaciones, de abuelos a nietos, incluyendo un respetable
número de yernos y de nueras. Se encuentra, sí, la familia en el sentido
estricto del término, es decir, como el grupo formado por un matrimonio y
sus hijos o un matrimonio solo. Nicolás Murray Butler habla en el estudio
ya citado de que en los Estados Unidos hay más de diez y ocho millones de
viviendas ocupadas aproximadamente por veintiún millones de familias y
de que seis millones de familias poseen propiedades sin gravamen y otros
tres millones son propietarios de bienes sujetos a hipotecas.
También es cierto, que esta familia en sentido estricto no tiene en los
Estados Unidos los caracteres de inmutabilidad y solidez que la distinguen
en Chile; pero quedaría todavía por averiguar si las condiciones de acá son
mejores que las de allá.
Se observa que el número de los divorcios ha ido en proporción
creciente de año en año después de la guerra separatista, como se ve en las
siguientes columnas:
Divorc
Años
ios
1867 9.937
1877 14.800
1886 25.535
1896 42.937
1906 72.062
124.00
1916
0
Sería razonar de una manera muy incompleta limitarse a inferir de estas
cifras que la inmoralidad aumenta de un modo alarmante en los Estados
Unidos y que la familia norte-americana se halla amenazada de muerte.
El número de los divorcios ha venido siendo mayor de año en año, no
porque los hombres y las mujeres se hayan vuelto más malos, sino porque
la evolución social, después de la mencionada guerra, ha puesto en acción
fuerzas nuevas que no han podido dejar de modificar la constitución del
hogar y a cuya influencia no han podido sustraerse los individuos. El
enorme florecimiento económico y la transformación industrial que se
notan desde aquel momento histórico, han hecho perder gradualmente a la
familia su antiguo carácter de centro de producción, han abierto vastos
horizontes de trabajo a la mujer, permitiéndole de esta suerte alcanzar su
independencia económica, y han aumentado al mismo tiempo las exigencias
de la vida.
Dentro de las condiciones anteriores al momento histórico indicado, el
hombre y la mujer habían podido soportarse mutuamente muchos defectos e
incompatibilidades espirituales y morales porque el hogar les ofrecía
ventajas económicas indispensables para la vida y difíciles de alcanzar de
otra manera. La mujer especialmente, sin preparación para desempeñar
ningún oficio o empleo fuera de la casa, no concebía la existencia sin los
medios que le proporcionaba el marido y se encontraba sometida a éste por
una estrecha dependencia económica.
Desaparecidas esas ventajas y arrancadas esas raíces materiales por
decirlo así, los cónyuges hanse tornado más exigentes en cuanto a
condiciones de armonía y de felicidad interior y menos tolerantes para
aguantar fricciones llevaderas en otro tiempo. Se considera ahora que es
menos inmoral la disolución que la permanencia de los matrimonios sin
amor.
Comprendido en este proceso de que hablamos se observa triunfante el
reconocimiento amplio de los fueros de la individualidad humana. La moral
dogmática ha podido sostener la indisolubilidad del matrimonio y condenar
el divorcio; y, más o menos de acuerdo con ella, ciertas doctrinas
sociológicas y jurídicas afirman que la célula social es la familia y no el
individuo y que éste debe resignarse a vivir férreamente encerrado entre las
mallas sin escapatorias de las instituciones matrimoniales y familiares. Pero
estos conceptos no se avienen con el espíritu norte-americano. Como dice
un profesor de sociología de la Universidad de Pennsilvania, «a la par que
se han hecho más intensos los sentimientos de nacionalidad y de conciencia
social se ha reconocido mayor valor a la personalidad del ciudadano.
Creadas sobre una base de franco utilitarismo, las formas e instituciones
sociales de cualquiera clase que sean no existen para llenar un fin en sí, sino
para el bien de los miembros de la colectividad. Libres de muchas de las
tradiciones, propias de los gobiernos monárquicos o despóticos y relativas
al carácter sagrado de algunas instituciones, los americanos no temen
ningún desastre por el hecho de llevar a cabo los cambios requeridos por la
expansión de la vida social. Los cambios no significan para nosotros
muestras de desintegración social, sino que los consideramos más bien
como la condición de un progreso sostenido. Miramos a la familia desde el
mismo punto de vista que a todas las demás instituciones sociales y no goza
de ninguna protección especial o tabú que pueda excusarla de ser sometida
a la prueba del utilitarismo. Ella, la familia, como todas las demás, debe
servir a los fines de su existencia o sufrir las transformaciones necesarias.
Desde que no es compatible con los ideales americanos de justicia y de
libertad que la familia sea tenida por más sagrada que el individuo, el
remedio debe buscarse en la modificación de la primera antes que en el
sacrificio del último».[24]
El mismo autor estudiando el problema desde un punto de vista moral,
dice: «Nuestras modificadas ideas éticas nos hacen sentir hoy día que el
matrimonio ha sido hecho para el hombre y no el hombre para el
matrimonio, y que el valor moral de
éste depende de la mutua felicidad que asegure a los desposados.
Cuando esta condición no se realiza ni se puede realizar, los más altos
intereses, tanto del individuo como del estado reclaman que se rompa ese
matrimonio perjudicial y que se coloque a los cónyuges en situación de
encontrar la felicidad que les ha faltado. De esta suerte un nuevo
humanitarismo ha venido a ocupar en religión y en ética el lugar que antes
llenaban las reglas teóricas de la ortodoxia de generaciones pasadas.
Descansa él sobre una moralidad práctica que estima el valor de las
instituciones en proporción al servicio que hacen en la formación del
carácter y en la producción del bienestar humano».[25]
Con estas palabras queda señalada la actitud americana ante el divorcio y
desvanecido el cargo de inmoralidad que por este motivo se pudiera
formular.
Religiosidad.—El Estado norte-americano no tiene religión oficial, y,
por lo mismo, florecen aquí las religiones como plantas espontáneas en un
terreno libre de las limitaciones planeadas por los hombres y donde obran
con toda libertad las fuerzas naturales fecundantes. En los Estados Unidos
prosperan más de cincuenta religiones y sectas, que cuentan con más de
cuarenta y dos millones de adeptos.
Es característico de las religiones en Norte América que orienten su
actividad en un sentido humano de bien y de servicio social y que sean
amplias, hospitalarias y tolerantes.
Estos rasgos han sido descritos de la manera siguiente por el católico
inglés William Barry: «Los americanos en otro tiempo creían con pavor en
la total depravación del hombre, de la cual sólo un pequeño número de
agraciados sería redimido. Ahora creen que el hombre es por naturaleza
bueno, por destino perfecto, y completamente capaz de salvarse a sí mismo.
En una especie de América ideal señalan el motivo fundamental de esta
vida más humana, a la cual deben tender sin cesar. La república, la
comunidad es su fin, los negocios el camino para el cielo, el progreso su
deber, la libre competencia su método. Rechazan el misterio, la obediencia
y la renunciación de la propia personalidad; pero admiran la disciplina que
condena y proscribe lo contrario a la dignidad del hombre, o, deferentes a
una idea delicada, practican la temperancia. Forman los americanos una
casta de héroes más que de ascetas. Para ellos la divinidad es un amigo y no
un señor omnipotente o un hado misterioso. Su creencia en la naturaleza
humana como algo que posee valor intrínseco que debe ser perfeccionado
significa la libre aceptación de una idea divina que el hombre tiene la
obligación de realizar. De esta suerte civilización y religión no son más que
diferentes fases de una misma gloria».
En muchas universidades privadas hay templos. En el de Princeton se
celebran servicios religiosos diariamente, pero ahí ofician los ministros de
todas las religiones, alternando, por consiguiente, el católico con los de las
diversas sectas protestantes o con un predicador de los Jóvenes Cristianos.
Los estudiantes no son obligados a asistir al templo sino una vez cada
quince días.
La misma índole de hogar cristiano, sin confesión determinada, abierto
al más dilatado espíritu religioso, es propia del templo de la universidad de
Leland Stanford Junior. Muchas tardes se dan aquí recitales de órgano, sin
ninguna ceremonia o acto del culto, y hasta el alma más descreída puede
disfrutar de una hora de reposo y recogimiento íntimo, y sentirse llevada en
alas del arte a las regiones de una pura religiosidad.
En medio del más agitado torbellino mundano, en Broadway, la
gigantesca arteria de Nueva York, en la puerta de un templo se lee la
siguiente dulce invitación: «No importa lo que pienses, no importa lo que
creas; entra; esta es la casa de Dios abierta para todos; entra a descansar». Y
aunque uno no entre, agradece la invitación, y siente que una sensación de
placidez discurre por todo su ser.
El ambiente norte-americano de libertad y la falta de una religión de
Estado son favorables al desarrollo del sentimiento de religiosidad. No
ocurre aquí lo que en aquellos países hispano-americanos, donde el
catolicismo goza de la preeminencia de religión privilegiada, llega
generalmente hasta las urnas electorales a defender sus privilegios, y queda
degradado por ambiciones de predominio social y político y por las
enconadas pasiones que se suscitan en su contra. No encontrando además
los espíritus por lo común en la América latina otra manera de ser religiosos
que la que les ofrece el credo católico, y no pudiendo comulgar con éste por
razones científicas, filosóficas o políticas, caen en el indiferentismo con
perjuicio a veces de su completa vida espiritual y moral y de su paz interior.
En los Estados Unidos, en cambio, se puede ser religioso de muchos modos
y el sentimiento y la fantasía encuentran más posibilidades que en otras
partes de llenar el vacío angustioso que los misterios de lo desconocido
suelen abrir en algunas almas.
Idealismo.—Grandes ideales han movido a los norteamericanos en
momentos culminantes de su historia, como ser la unión de las trece
colonias que obtuvieron su independencia de Inglaterra en 1783, la defensa
de la autonomía del continente americano, encarnada en la doctrina de
Monroe, la lucha por la abolición de la esclavitud, y el amor a la
democracia, que constituye un leitmotiv en la evolución social y la política
de este pueblo.
La fuerza de un resorte idealista se observa hoy día en la acción de una
élite formada por millares de personas de ambos sexos que comprenden
desde sacerdotes de todas las religiones y profesores hasta los millonarios.
No hay problema de la colectividad que no traten de resolver, no hay dolor
que no vayan a aliviar, y los beneficios de su voluntad filantrópica y de sus
millones pasan las fronteras del país y llegan hasta los huérfanos de Francia
y Bélgica o van a tender una mano salvadora al infeliz y secularmente
atormentado pueblo de la Armenia.
Es un idealismo práctico de carácter social. Su acción se muestra de una
manera estupenda en toda obra de cultura, y en el sostenimiento de
bibliotecas y de institutos de educación. Existen en el país, debidas casi en
su totalidad a la iniciativa privada, más de tres mil bibliotecas con un total
de setenta y cinco millones de volúmenes. Se gastan al año en el
mantenimiento de escuelas para el pueblo en toda la nación cerca de
seiscientos cincuenta millones de dólares. La renta de las universidades se
ha elevado, en virtud de las donaciones y legados que han recibido, de diez
y siete millones de dólares que era en 1892 a ciento veinte millones en
1915.
Es este idealismo tan propio del alma americana que es el que informa
las doctrinas de sus más genuinos filósofos y pensadores. Como hemos
recordado en un capítulo anterior, Emerson fué un moralista y William
James ha preconizado en su pragmatismo la prioridad de la acción sobre el
pensamiento y la busca de la verdad, no por la verdad misma, sino en
cuanto puede servir para la vida. Ideas más o menos semejantes sustenta el
filósofo más representativo del momento actual, Juan Dewey; y, aun
pensadores que no han comulgado con el pragmatismo, como el sociólogo
Lester F. Ward, han señalado a la existencia del hombre la finalidad del
«meliorismo» o sea el mejoramiento de la vida, lo que equivale a llegar por
distinto camino al mismo objetivo.
Esta filosofía, que puede presentarse como no satisfactoria e incompleta
para inteligencias especulativas y ansiosas de ahondar en las últimas
razones de las cosas, ha probado ser bastante para espíritus ante todo
activos, como los americanos, y, ya hemos visto en el párrafo anterior, que
hasta la religión misma ha recibido el sello de esa orientación distintiva del
alma nacional.
CAPITULO VII
CARTAS DEL DOCTOR N.—CONCLUSION

Primera Carta.—Los norteamericanos en la gran guerra.—Wilson.—La
Sociedad de las Naciones.—La humanidad futura.—Segunda Carta.—
Del amor y de la libertad.—Tercera Carta.—Algunas diferencias de la
psicología y de las instituciones de los norte y sudamericanos.—
Conclusion.—Apreciaciones sobre Europa y Estados Unidos.—
Desigualdades sociales en Norte y Sud América.—Los millonarios de acá
y de allá.
A principios de 1919 llegó a Boston, procedente de Europa, el doctor N.
Venía a conocer de cerca la famosa Escuela de Medicina de Harvard y a
recorrer un poco el este del país.
Me escribió algunas cartas que, por tener cierta relación con los asuntos
de este libro, no carecer de interés psicológico y ser de un personaje ya
conocido del lector, creo oportuno reproducir:
PRIMERA CARTA
Boston, Febrero de 1919.
Ya me tiene usted, mi querido amigo, en tierra americana después de
unos cuantos meses de estada en Europa. Mi primer saludo es para usted,
deseando que éste se pueda expresar pronto en un abrazo. Vengo a hacer
algunos estudios en la Universidad de Harvard y después veremos. Pero,
¡qué fría y de clima tan riguroso es esta grave ciudad de Boston! Sin
embargo, la calefacción anda aquí mucho mejor que en París.
¡Qué provechosa ha sido para mí esta temporada en Europa y cuánto he
enriquecido mi espíritu! Fuera de lo que atañe a los progresos de mi
profesión, de que he podido imponerme, he visto la desolación de la Europa
después de esta guerra bárbara y respirado en el austero ambiente de
heroísmo que ella esparció junto con el dolor. He presenciado el justificado
delirio patriótico del pueblo francés y de los aliados en París el día de la
firma del armisticio. Vivimos a no dudarlo en un momento histórico
trascendental y que lo es muy especialmente para el pueblo norte-
americano. Hace un siglo el mundo se encontraba en una situación algo
parecida en cuanto, tras las largas guerras napoleónicas, agotado ya, llegaba
por fin a una paz cualquiera. Pero, ¡qué de diferencias por lo demás! Los
vencedores de entonces no tenían otra mira que los intereses dinásticos de
monarcas absolutos; los de hoy han llevado en sus banderas los lemas de la
libertad y democracia. Entonces la Europa se consideraba a sí misma como
el único continente civilizado y los Estados Unidos permanecieron sin
tomar parte en el conflicto como un niño sano y vigoroso que hace sus
tareas tranquilamente en su casa sin prestar atención a los ruidos de la calle.
Ahora los americanos, no sólo han vaciado sus arcas en favor de los aliados,
y los han provisto de armas, de pertrechos de guerra y de víveres, sino que
sus ejércitos han cruzado el océano para salvar a la Europa y decidir muy a
tiempo la victoria. En galardón, los americanos han recibido de los
vencedores homenajes de glorificación.
El Presidente Wilson ha sido objeto de apoteosis en las capitales
europeas que talvez nunca se han tributado en forma análoga a otros jefes
de Estado. Es una gloria para la nación que representa, pero, lo es asimismo
para él, porque esas manifestaciones han sido en parte principal coronas
discernidas a las virtudes relevantes de su personalidad. Tengo una alta idea
del valor moral, del idealismo, del patriotismo y de la inteligencia de
Wilson. En sus discursos se ha mostrado siempre franco, avanzado, valiente
y demócrata sincero, y creo que en sus actos ha sido consecuente con sus
discursos. Su genio significa la síntesis feliz que resulta de haber fundido en
una sola alma las cualidades de un gran universitario y de un estadista.
Como adalid de los grandes ideales de la humanidad, la democracia y la
sociedad de las naciones—a la cual él ha sabido darle la forma más perfecta
que haya tenido hasta ahora—su figura ha alcanzado proporciones
eminentes.
Sin embargo, no todos juzgan a Wilson de la misma manera. El
corresponsal del Times de Londres en París, decía en estos días: «Wilson
anda con diferentes baterías de pequeño calibre moral; pero sus grandes
cañones son asuntos de dólares. Más de un conflicto debe haber entre su
idealismo personal y los intereses financieros que él representa. Recibe
muchas instrucciones de Wall Street y de Chicago». Vaya a averiguar usted
lo que haya de cierto en esto.
¡La Sociedad de las Naciones! No faltan espíritus escépticos que no
crean en ella. Yo, a quien usted tiene motivos para considerar uno de esos,
sustento, sin embargo, la más confiada esperanza en su porvenir. Creo aún
que, como la especie humana tiene al parecer más de dos millones de años
de vida por delante y su historia se complicará mucho si seguimos
dividiéndola en la forma en que lo hemos hecho hasta ahora, será menester
englobarla en la simple distinción de dos grandes épocas, la de la
humanidad guerrera y la de la humanidad pacífica. En el punto de
separación de una y otra, nuestros días ocuparán un lugar preeminente
como precursores y preparadores de la nueva vida. Entonces habrá
triunfado la justicia social y habrá llegado la gran era del arte y hasta de la
metafísica. No sujetos ya los hombres, como ocurre ahora, a la
preocupación casi exclusiva de luchar por el sustento, por la propiedad, por
el dinero, dejará de ser la creación artística la obra atormentada de unos
pocos, las lucubraciones sobre problemas ultra-científicos no serán sólo
afanes de raros soñadores, y el espíritu humano explorará en vuelos
atrevidos, pero serenos y resignados a veces, las regiones superiores de lo
bello y de lo divino. Para aligerar la carga de la memoria histórica se
simplificarán los períodos de la vida de la humanidad guerrera, como ha
ocurrido más o menos con la prehistoria, y se hundirán para siempre en el
limbo de un piadoso olvido los héroes de la espada, de la política y de la
tiranía. En el panteón de los hombres sólo se elevarán altares a los héroes
del espíritu, a los reformadores, a los filósofos, a los artistas, a los hombres
de ciencia, a los grandes fundadores de religiones.
En fin, amigo mío, esto no parece carta ya; acepte un abrazo cordial de
su amigo, escríbame y cuénteme de su vida.
SEGUNDA CARTA
Boston, Marzo 1919.
Mil gracias por su saludo de bienvenida.
Me pregunta usted cuál es el estado de mi corazón. Ay! amigo mío, no
obstante el interés que he tenido y tengo por muchas cosas, mi alma se
encuentra más o menos como usted la conoció hace algunos meses. La
imagen de esa mujer, de esa amiguita querida, como solía llamarla, me ha
seguido a todas partes. La tengo metida en el alma y cuando no duermo,
ella ocupa mi conciencia. En los bulevares, en el teatro, en alguna recepción
he mirado a la distancia trajes parecidos a los que ella solía llevar y ya creía
verla surgir de repente; he sentido el perfume que le era propio y he vuelto
la cabeza esperando encontrarla no lejos de mí. Es una obsesión.
A veces me entra una ternura sin motivo inmediato; es como si se me
abrieran suavemente las entrañas y me dan ganas de llorar. ¡Qué dulce debe
ser una vida en que, después del trabajo diario, se encuentre un pecho en
qué reclinar la cabeza fatigada y hacer que esa ternura que se ha abierto en
uno no lo ahogue, y corra, corra como una fuente refrescante, bienhechora y
blanda! Otras veces, viendo la inutilidad de mis sueños, me desespero. Si
amanece un día hermoso, el azul del cielo me causa enojo. ¿Para qué brillas
sol, digo, si no he de ver con tu luz a la mujer que amo? Si azota la lluvia y
bajo un cielo gris estalla la tempestad me es indiferente: son tan grandes mi
pesar y mi desesperanza, que se me ha formado una coraza contra todos los
males que puedan venir de afuera. Con resignación pienso que, a pesar de
mi soltería, éste será mi último amor, que viviré con él y lo llevaré conmigo,
como una herida siempre abierta a cuyo dolor me iré acostumbrando.
En este mundo de ajetreo sin fin no encuentro sino una cosa que lo
justifique: el amor. Ahondo, ahondo y no veo a qué resultado definitivo
pueda conducir nuestra actividad. No es que el amor signifique una
explicación de la vida, no; el amor es el imperativo de la vida y una fuerza
que nos toma de las entrañas, la fuerza más poderosa para querer la
existencia. El amor y la actividad son las mejores alas para hacer el tránsito
por este mundo; pero no explican nada. La vida humana se presenta (y así
será aunque dure millones de años), como una franja de luz entre dos
noches infinitas. El amor y la actividad, he dicho; mas cuando el amor es
desgraciado, la laxitud que se apodera de uno como un fluido enervante, no
permite ser activo.
¿Y la libertad de los hombres? No digo nada de la mía; ya irá viendo
usted que mi pecho es un escenario dantesco en que hay rachas de infierno,
de purgatorio y de cielo, y que en medio de ellas—no quiero hablar de mi
libre albedrío—mi voluntad no es más que una pobre ave encogida,
entumecida, sin fuerzas, con los resortes de sus alas relajados. Pero me
consuelo pensando en cómo suele andar la libertad de los demás, ah! sin
exceptuar a los más graves y campanudos. Que se hable de amor y de
mujeres delante de sabios, políticos, filósofos, financieros, es decir, ante la

Proofs and Models in Philosophical Logic Greg Restall All Chapters Instant Download

Uploaded by

Copyright:

Available Formats

Proofs and Models in Philosophical Logic Greg Restall All Chapters Instant Download

Uploaded by

Document Information

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

Proofs and Models in Philosophical Logic Greg Restall All Chapters Instant Download

Uploaded by

Copyright:

Available Formats

Full download ebooks at ebookmeta.

Proofs and Models in Philosophical Logic Greg

Download more ebook from https://ebookmeta.com

Logic and Language Models for Computer Science Fourth

Cavell s Must We Mean What We Say at 50 Cambridge

Seven Myths of the Spanish Conquest Matthew Restall

Towards Scientific Metaphysics Volume 2 Benedykt

Proofs & Theories 1994th Edition Louise Gluck

Program Proofs 1st Edition K. Rustan M. Leino

Mathematical Stories I – Graphs, Games and Proofs: For

Cassirer and Heidegger in Davos: The Philosophical

Proofs and Models

ISSN 2516-418X (online)

Cambridge University Press is part of the University of Cambridge.

Elements in Philosophy and Logic

Author for correspondence: Greg Restall, gr69@st-andrews.ac.uk

Abstract: This Element is an introduction to recent work on proofs and

Keywords: proofs, models, logic, semantics, paradox

© Greg Restall 2022

1.1 Proofs and Models

this is to focus on the production and the evaluation of argumentation, reasoning

(1) R is transitive. It’s symmetric. It’s directed. It must be reflexive, too!

(2) Take an object a. Since R is directed, there is some object b where R

be a very different problem than not understanding a large leap in thought. It

We can fill in this reasoning into a proof in the following way:

Suppose, on the other hand, we tried a similar argument, with necessity in

This argument is less convincing. We can propose a model as a counterexample.

Example 1: The liar paradox

(7) Sentence (7) is not true.

Example 2: The sorites paradox

If we wish to find a counterexample to the validity of the sorites argument, we

1.3 The Plan

2.1 Proof Structures

straightforwardly.10 A proof from a set X of premises to a conclusion A is a

The →E rule is straightforward. The →I rule needs some explaining. It says

to now conclude p → r from the premises p → q and q → r. This is a very

s=t A(s) s=t A(t)

Figure 1 Natural deduction rules

on a single logical concept, a quantifier or a connective. The rules for the

14 To state this precisely requires an accounting of when a variable in a formula is bound by a

2.2 Inference Rules as Definitions

18 Prawitz’s monograph (1965) contains the proof.

2.3 Structural Rules

20 This is an elegant result due to Dag Prawitz (1965).

21 Here, X is the collection of other premises used as the proof.

On the other hand, if whenever we have a proof for X, A, A  B we can construct

constructs a proof in which A is now present as an assumption, though it clearly

which leads from B to A → B, where the antecedent A might be totally uncon-

same effect by allowing for rules like →I discharge zero occurrences of an

Here, the →I step discharges zero occurrences of the assumption A. Such a

according to which, whenever we arrive at C in a proof in which both A and B

Restricting the paradoxical arguments at the point of CONTRACTION is elegant

or something like this ‘reductio’ principle:

However, there is no proof of ((p → q) → p) → p in our original system.

represent contradictoriness in a sequent by leaving out the ⊥ formula. The rules

For Curry’s paradox, we start with this derivation of c ∈ c  p.

(Notice that it uses a Cut.) If we call that derivation ∆, we can extend it as

I indicated that the sequent derivation perspective on our paradoxical proofs

so would X, ¬B¬A, Y. (Of course, we could posit particular axioms governing

2.4 Other Proof Structures

On the other hand, if whenever we have a proof for X, A, A B we can construct

For Curry’s paradox, we start with this derivation of c ∈ c p.

so would X, ¬B¬A, Y. (Of course, we could posit particular axioms governing