You are on page 1of 13

Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.

Let a Thousand Flowers Bloom: A Tour of Logical


Pluralism
Roy T. Cook*
The University of Minnesota

Abstract
Logical pluralism is the view that there is more than one correct logic. In this article, I explore what
logical pluralism is, and what it entails, by: (i) distinguishing clearly between relativism about a par-
ticular domain and pluralism about that domain; (ii) distinguishing between a number of forms logi-
cal pluralism might take; (iii) attempting to distinguish between those versions of pluralism that are
clearly true and those that are might be controversial; and (iv) surveying three prominent attempts
to argue for logical pluralism and evaluating them along the criteria provided by (ii) and (iii).

1. Introduction
Logical pluralism is the view that there is more than one ‘correct’, ‘best’, or ‘legitimate’
logic, while logical monism is the view that there is only one such logic. Of course, this
simple description is of little help until we know what a logic is, what the study of logics
is meant to accomplish, and what counts as being ‘correct’, ‘best’, or ‘legitimate’.
Often the terms ‘pluralism’ and ‘relativism’ are used interchangeably, when in fact
there are sharp differences between a relativistic view of some phenomena and a pluralist
view of that same phenomena. Thus, the first task will be to sort out the differences
between pluralism and relativism. Once this has been accomplished, we shall survey dif-
ferent ways of spelling out a pluralist view regarding logic. Some ways of filling out the
details entail that pluralism is trivial (or, at least, that pluralism of these sorts can be seen
to be true with a little work), while others provide a richer, and more controversial,
understanding of the idea that there is a multiplicity of formal systems worthy of the hon-
orific ‘logic’. In particular, it turns out that many of the ways that one can obtain versions
of logical pluralism ‘on the cheap’, so to speak, involve making one’s account of ‘correct-
ness’ relative to various factors. This provides a focal point for the remainder of the arti-
cle, as we survey a number of attempts to provide a pluralist view of logic, because one
fruitful way for distinguishing between, and comparing, these variants of logical pluralism
is to determine whether, and where, they involve relativisms of various sorts.
The discussion will focus on three attempts to provide something like a pluralism
regarding logic: Rudolph Carnap’s logical tolerance, Greg Restall and JC Beall’s logical
pluralism, and Stewart Shapiro and Roy T. Cook’s logic-as-modeling view. We conclude
by noting a few outstanding issues.

2. Relativism Versus Pluralism1


One is a relativist about a particular phenomenon X if and only if one thinks that the
correct account of X is a function of some distinct set of facts Y. Thus, relativism about
X amounts to acceptance of the following schema:

ª 2010 The Author


Journal Compilation ª 2010 Blackwell Publishing Ltd
Let a Thousand Flowers Bloom 493

The correct account of X is relative to Y.


To take a simple example, most of us are relativists regarding etiquette, as the correct
account of etiquette – that is, the correct account of the conventional norms that consti-
tute polite behavior – seems to be a function of what culture one is living in.
One is a pluralist about a particular phenomenon X if and only if one thinks that there
is more than one correct account of X, and one is a monist about X otherwise.
Relativism about a phenomenon X and pluralism about X are often run together in
discussions. This is unfortunate, even if to some extent understandable, as the two are
independent – relativism, understood carefully, does not by itself imply pluralism, nor
does pluralism imply relativism.
The temptation to think that relativism implies pluralism is easily explained, because
typically relativism is only interesting in those cases where it also involves pluralism.
Nevertheless, relativism implies pluralism only if supplemented with the additional claim
that there is more than one relevant set of facts that can be plugged in for the
independent variable Y: Imagine a possible world which contains, forever and always, a
single uniform culture. In this world there is a single correct account of etiquette. Such
a situation would not be one where we denied that etiquette was relative to culture, but
it would be one where we would (or, at least should) be monists regarding etiquette
nevertheless.
Likewise, pluralism does not imply relativism. Intuitions to the contrary are easily
understood, as relativism about a particular phenomenon is surely the most straightfor-
ward route to pluralism. Nevertheless, we should not rule out the possibility of a plural-
ism that does not involve any sort of relativism – that is, a view that there is more than
one correct account of X even if we keep all other factors fixed. The possibility of this
sort of pluralism will be of great importance in what follows.
Nevertheless, as we have already noted, relativism is typically uninteresting except in
cases where it implies a corresponding pluralism. Thus, at the risk of burdening ourselves
with too much novel terminology, we shall introduce the following definitions: An
account of X is a dependent pluralism if it involves there being multiple ‘correct’ accounts
of X, but this multiplicity is the result of an underlying relativism. A pluralist account that
does not involve such an underlying relativism is a simple pluralism.
To sum up, being a pluralist about some phenomenon need not, in general, imply that
one is a relativist, nor vice versa. The next step is to work out how these notions interact
in the debate over the correct logic.

3. Pluralism About Logic: A Map of the Terrain 2


For our purposes, we can define a formal logic (hereafter, simply ‘logic’) to be any pair
<L, > where L is a formal language (i.e., L consists of a nonempty set of primitive
symbols and a set of formation rules – typically recursive – that delineate which sequences
of primitive symbols are ‘well-formed’) and  (the formal consequence relation of the
logic) is a relation holding between a set of statements from L (the premises) and a single
statement from L (the conclusion).  can be given in terms of a system of proof, or a
formal semantics, or primitively – the pluralist debate hinges primarily on how many such
relations there might be, and not on how these relations are presented to us.
This definition is, strictly speaking, both too narrow and too wide: On the one hand,
it ignores logics, such as many formulated within the sequent calculus, that allow multiple
conclusions (see Gentzen 1965). On the other hand, it includes mathematical structures
with bizarre or random formal consequence relations (e.g., those that do not validate

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
494 Let a Thousand Flowers Bloom

{F}  F, those that are not transitive, etc.) that many logicians would fail to count as
potential candidates for being ‘correct’ or ‘acceptable’ logics in any sense. These simplifi-
cations do not affect the points to be made below, however.
With a definition of ‘logic’ to hand, we can now formulate the first, and weakest, vari-
ant of logical pluralism – one that we can call mathematical logical pluralism (MLP):
MLP: There is more than one logic.
Where or not MLP is true is a purely mathematical issue, as its truth-value hinges solely
on the existence of certain mathematical structures. Further, MLP is trivial: Given any
nonempty set of sequences of objects S, let 1 be all of }(S) · S, and let 2 be the
empty relation (}(A) is the powerset of A, and A · B is the Cartesian product of A and
B). Then both <S, 1> and <S, 2> are logics. Finally, MLP is orthogonal to the rela-
tivism ⁄ pluralism distinction discussed above, as MLP is a claim about the existence of
more than one logic, and entails nothing regarding the correctness, in any sense, of any
of these logics. Thus, this sort of pluralism is not what philosophers have in mind when
engaging in the logical pluralism ⁄ monism debate. Nevertheless, were MLP false, then
more substantial versions of pluralism would be nonstarters.
The next version of pluralism involves the idea that some logics (and, crucially, more
than one logic) are useful codifications of various phenomena that interest mathematicians
and philosophers. We can distinguish between two variants of this idea. The first we can
call Mathematical Application Pluralism (MAP):
MAP: There is more than one logic that can be fruitfully applied (in the general sense of
‘applied mathematics’).
Mathematical Application Pluralism amounts to the claim that there are at least two dis-
tinct logics that can be fruitfully applied to represent aspects of the world. The second
variant, which is of a more purely philosophical nature, we can call Philosophical Logic
Pluralism (PLP):
PLP: There is more than one logic that can be fruitfully applied to philosophically interesting
phenomena.
Philosophical Logic Pluralism is, in essence, a special case of MAP. In addition to claim-
ing that more than one logic is applicable to the world – that is, more than one logic is a
useful formal tool for the study of various phenomena – PLP asserts further that at least
two of these logics are useful tools for studying those parts of the world that are of pri-
marily philosophical interest.
Both MAP and PLP are pluralisms that result from a corresponding relativism – that is,
they are dependent pluralisms. To see this we need merely note that MAP and PLP are
(roughly) equivalent to:
MAP*: The correct logic (relative to the goals of applied mathematics) is relative to the phe-
nomenon that logic is meant to represent.
PLP*: The correct logic (relative to the goals of philosophical logic) is relative to the linguistic
phenomenon being represented.
Many readers with some mathematical and ⁄ or philosophical sophistication will conclude
that both MAP and PLP are true, and trivially so. After all, logics have been central to
the study of a number of phenomena, including many that have, at best, an indirect
connection to human reasoning such as electronic circuit design, database management,
and internet security. Such a reaction is only partially correct. While at this stage in the

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
Let a Thousand Flowers Bloom 495

history of logic, there can be little doubt regarding the truth of either MAP or PLP, we
should not be fooled into thinking that their truth is trivial. On the contrary, the utility
of logics in fields far removed from the study of reasoning would have been surprising a
mere century ago, when the initial, and greatest, strides in the study of logic were being
made (in other words, the relativism inherent in MAP and PLP might be obvious, but
the pluralism inherent in such views might not have been).
Nevertheless, there is little disagreement over MAP or PLP now. Thus, if logical plu-
ralism is to be a substantial and controversial thesis, something more must be intended.
That something more is the notion of logical consequence – that is, a logic is ‘correct’,
or ‘acceptable’, etc., if and only if it is a correct (or acceptable, etc.) codification of logi-
cal consequence. The idea that the philosophically primary (but obviously not only) goal
of logical theorizing is to provide a formal codification of logical consequence in natural
language traces back (at least) to the work of Alfred Tarski, who also provides us with a
useful first approximation of what logical consequence amounts to:
Consider any class D of sentences and a sentence F which follows from the sentences of this
class. From an intuitive standpoint it can never happen that both the class D consists only of
true sentences and the sentence F is false. Moreover, since we are concerned here with the
concept of logical, i.e. formal, consequence, and thus with a relation which is to be uniquely
determined by the form of the sentences between which it holds… the consequence relation
cannot be affected by replacing the designations of the objects referred to in these sentences by
the designations of any other objects. (1983:414–5)
In order to not beg any questions, we shall (except for where appropriate) speak of logi-
cal consequence generally, and not of the (singular, unique) logical consequence relation.
Thus, according to Tarski, logical consequence is a relation (or relations!) that holds
between statements (and collections of statements) of our actual language, and involves
two crucial ideas:
LC: A (natural language) statement F is a logical consequence of a set of (natural language)
statements D if and only if:
(Necessity) The simultaneous truth of every member of D guarantees the truth of F
and:
(Formality) This guarantee follows solely from the logical form of F and of the members of D.
In judging the correctness of a logic, on this understanding, we evaluate that logic in
terms of how well it ‘matches up’ with logical consequence. Given a logic <L, >, such
an evaluation is typically carried out along something like the following lines:

(1) Identify a subset LV of the primitive symbols of L – this is the logical vocabulary of
L.
(2) Construct a (partial) translation function T from LV to appropriate bits of natural
language – this projects the logical ⁄ nonlogical distinction in <L, > onto our natu-
ral language.
(3) Determine whether or not the following Correctness Principle (CP) holds:
CP: Given any recursive mapping I from L to statements in our natural language (i.e. an
interpretation) which agrees with T on LV, and given any statement F and set of statements
D from L:
I(F) is a logical consequence of I(D) if and only if D  F.
Put a bit more loosely, a logic (plus an identification and interpretation of logical vocabu-
lary) is correct if and only if, for any way of interpreting the nonlogical vocabulary, the

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
496 Let a Thousand Flowers Bloom

logic validates a particular argument if and only if the natural language statement corre-
sponding to the conclusion of that argument is a logical consequence of the natural lan-
guage statements corresponding to the premises of that argument.
Once we have all this in place, we can formulate another variant of logical pluralism,
which we shall call Logical Consequence Pluralism (LCP):
LCP: There is more than one logic such that CP holds for that logic.
Even this, however, does not get to the heart of the matter, as if there is one logic that
satisfies CP, then there is almost certainly more than one: Assume for the moment the
relatively common view that classical polyadic first-order logic with identity satisfies CP.
Then we can also satisfy CP by looking at the logics that result from restricting our atten-
tion to the propositional fragment of our first-order language, or the S1 fragment, or the
monadic fragment, etc. Each of these logics will also satisfy the correctness principle, as
each will also correctly codify logical consequence – they just codify logical consequence
for a more restricted portion of natural language. Thus, we can obtain logical pluralism
on the cheap merely by varying the formal language in question – that is, once we realize
that a correct account of logical consequence is relative to the formal language being
studied, and we note that there is more than one relevant language, pluralism follows
immediately.
In addition to varying the language, we can also obtain multiple logics satisfying CP by
varying which bits of primitive vocabulary are to count as logical. For example, if we
again assume that L is the language of classical polyadic first-order logic with identity,
there are distinct 1 and 2 such that <L, 1> satisfies CP if the identity relation sym-
bol in L is treated as primitive logical vocabulary and <L, 2> satisfies CP if the identity
symbol is treated as a nonlogical symbol.
Achille Varzi defends such a view in ‘On Logical Relativity’ (2002), arguing that the
correct account of logical consequence is relative to where one draws the logical ⁄ nonlog-
ical divide (and on how one interprets those expressions falling on the logical side of this
divide), and that, further, there is more than one legitimate way to draw such a line (and
make such an interpretation). Clearly, this account is, like the less controversial pluralisms
discussed above, a dependent pluralism: The correct logic is relative to which logical con-
sequence relation one is codifying, and logical consequence is itself relative to one’s
choice of where to draw the logical ⁄ nonlogical distinction. As, according to Varzi, there
is more than one legitimate way to draw this divide, there is more than one correct logic.
In short, Varzi’s pluralism about logic is a direct consequence of his prior pluralism
regarding the logical ⁄ nonlogical divide.
Although Varzi’s view is no doubt interesting and important, we shall set it aside for
the moment. The reason is simple: Although Varsi’s pluralism regarding the logical ⁄ non-
logical divide is controversial, the inference from pluralism regarding legitimate ways to
draw this distinction to pluralism about logics is not. Thus, it is natural to ask whether
and how one might be (and whether anyone has been) a pluralist regarding logic where
that pluralism does not stem from variation of the language or the logical ⁄ nonlogical
divide. We will call this sort of pluralism – that is, pluralism regarding the correct or best
logic once the language and logical ⁄ nonlogical divide are fixed – Substantial Logical Plural-
ism (SLP):
SLP: Given a formal language L and an identification of the logical vocabulary LV of L (and
projection T of LV onto natural language), there exist distinct 1 and 2 such that CP holds
for both <L, 1> and <L, 2>.

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
Let a Thousand Flowers Bloom 497

As we shall see, SLP is not necessarily pluralism without any relativism. Rather, SLP
merely rules out pluralisms that stem from the correct logic being a function of our
choice of, or interpretation of (vis-à-vis logicality) our language. As we shall see in our
examination of JC Beall and Greg Restall’s view, one can hold an account of logic that is
both a version of SLP and is a dependent pluralism.
Thus, a defense of logical pluralism, in its most ambitious form, amounts to arguing
that, given a fixed formal language and logical ⁄ nonlogical divide, there is more than one
logic that correctly codifies logical consequence in natural language. If a view about logic
is to be interesting or controversial qua logical pluralism, it needs to have something like
SLP as a part (of course, this does not mean that some ways of developing non-SLP ver-
sions of pluralism are not interesting, controversial, or important for other reasons, such
as Varzi’s pluralism regarding logical vocabulary!) In the sections to follow, we shall sur-
vey three well-known attempts to provide a pluralist account of logic, focusing in each
case on whether and how such an approach provides us with an instance of SLP.

4. Carnapian Tolerance3
The first stop on any tour of logical pluralism must be Rudolph Carnap’s notion of logi-
cal tolerance. In The Logical Syntax of Language (1959), Carnap argues that one can only
ask questions regarding the correctness of a logic relative to a particular language or lin-
guistic framework (for Carnap a framework might be more than merely a language, but
such exegetical issues need not detain us). Our choice of framework, however, is con-
strained only by pragmatic concerns regarding what our use of the language is meant to
accomplish and how well it carries out this task:
In logic, there are no morals. Everyone is at liberty to build his own logic, i.e., his own form
of language, as he wishes. All that is required of him is that, if he wishes to discuss it, he must
state his methods clearly, and give syntactical rules instead of philosophical arguments. (1959:52)
Thus, for Carnap, questions about the correct logic reduce to questions about the system
of language – the framework – that we have adopted. Once we have adopted a particular
linguistic framework we can ask, and answer, questions about the logic governing that
framework, but such questions are internal to the framework, and answers take the form
of ‘such and such is valid within our present system’ and not ‘such and such is valid per-
iod’. There is no ‘external’ perspective from which we can ask which logic is really cor-
rect, as any such question will have to be formulated within a language, and thus from
within the perspective of some framework or other. Thus, there are, for Carnap, no
primitive logical facts to be discovered purely through logical investigation (or through
any other investigational method) – the facts of logic, such as they are, are not external
to us, but are instead imposed by us in virtue of the linguistic frameworks that we adopt.
Carnap’s view certainly amounts to a form of logical pluralism. In particular, it is clear
that LCP holds within Carnap’s account (insofar as we can find a suitably ‘external’ per-
spective from which we can meaningfully formulate such global inter-framework theses
in the first place!). But this is a dependent pluralism, resulting from an underlying relativ-
ism – that is different logics result from varying the language in question (it is worth not-
ing that Carnap does not advocate pluralism within a framework – different linguistic
frameworks might be governed by different logics, but within a particular framework
there is a single logic that correctly codifies the (internal) logical consequence relation of
that framework). Thus, Carnap’s tolerance amounts to a version of logical pluralism, but
not a version of SLP.

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
498 Let a Thousand Flowers Bloom

There is, of course, a bit more to Carnap’s position than merely a defense of LCP but
not SLP: Carnap does not merely provide us with no reason for thinking that there are
multiple logics that correctly codify logical consequence in a particular natural language.
Instead, Carnap’s position is incompatible with SLP, as formulating this sort of substantial
pluralism requires that there be some ‘external’ position from which we can judge the
degree of fit between various formal logics and logical consequence within natural lan-
guage. Thus, Carnap’s position constitutes not a denial of SLP so much as a denial of its
very coherence.
Before moving on, it would be remiss to fail to mention a variant of Carnapian toler-
ance that has emerged much more recently. Stewart Shapiro, in Philosophy of Mathematics:
Structure and Ontology (1997), argues that:
… I urge a more eclectic attitude concerning logic, and that intuitionistic and classical systems
both be regarded as legitimate branches of mathematics …. The point here is that in most cases,
classical mathematicians recognize the intuitionist’s work as mathematics, even if they reject its
underlying restrictions. Intuitionists must also see at least a tight family resemblance between
their work and that of classical colleagues. Thus, I take it as a desideratum that a philosophy of
mathematics accommodate both and not reject either one as incoherent. If the theme of struc-
turalism is to be maintained, the structure of nonclassical systems ought to be countenanced.
(1997:190–1)
A full explication of the structuralist philosophy of mathematics underlying Shapiro’s logi-
cal tolerance would take us too far afield, but the bare bones of the approach proceeds
something like this: Mathematics is about abstract structures. A structure exists if and only
if there is some coherent description of it (where ‘coherent description’ is an informal
analog of a consistent set of axioms). A logic is legitimate if and only if it correctly codi-
fies the consequence relation on some structure. As intuitionistic and classical languages
generate distinct coherent descriptions, there are distinct structures (with their own logics)
corresponding to these descriptions, and there is thus more than one legitimate logic.
As a result, Shapiro’s notion of tolerance is, like Carnap’s, a dependent pluralism and,
again like Carnap, which logic is correct is relative to one’s choice of language. Shapiro
does introduce an additional step into the picture, as on his view logics are relative to
structures, but as structures depend on languages, this is, for our purposes, an unimportant
complication. Both views obtain pluralism as a result of the correct logic being relative to
the language one adopts, and as a result, neither view amounts to a version of SLP.

5. Beall ⁄ Restall Pluralism


The second and most well-known stop on our excursion through the pluralist landscape
is the view defended by J.C. Beall and Greg Restall. In a number of jointly written arti-
cles (e.g., 2000, 2001) and a book (2006), Beall and Restall have argued for a view they
call merely ‘logical pluralism’, but which we shall for clarity’s sake refer to as Beall–
Restall pluralism. Beall–Restall pluralism begins with Tarski, as they adopt the following
principle which they argue captures Tarski’s (and the resulting Tarskian tradition’s) notion
of logical consequence:
V: A conclusion A follows from premises, S, if and only if any case in which each premise in
S is true is also a case in which A is true. (2000:476)
Beall and Restall point out that the logic one obtains from V depends on what one
thinks of as the relevant cases. Not anything will do, of course. Beall and Restall suggest

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
Let a Thousand Flowers Bloom 499

that a candidate logic must be necessary, normative, and formal (see 2006:14–23). Never-
theless, even given these restrictions, there are a number of different classes of cases that
can be plugged into V (including Tarkian models, constructions, and situations) resulting
in different logics (classical, constructive, and relevant logics, respectively).
Clearly, Beall–Restall pluralism is a version of logical pluralism, and in addition, and
unlike Carnap’s tolerance or Varzi’s relativism, the Beall–Restall account does not depend
on varying the language in question or shifting the line between logical and nonlogical
vocabulary. On their view we require a multitude of logics in order to codify all of the
different linguistic relations that are of interest to us as logicians:
We hold, of course, that classical logic is not universally applicable in the sense of dictating
what is true in each and every case. Not all situations and constructions are closed under logical
consequence. But that does not mean that classical validity is not validity. For it is still true that
if the premises of an argument are true, the conclusion must be true too. Of course, classical
consequence does not do every job required of deductive validity. Other logics are better suited
to some of these tasks. That is why we need a plurality of logics. (2001:15)
So far, so good – there are, no doubt, many interesting relations holding among state-
ments (or sets of statements) of our natural language, and the different logics which result
from substituting different classes of cases into V codify these different relations.
At first glance the reader might be fooled into thinking that this is all there is to the
Beall and Restall view – that is, that they are merely advocating a particularly well-
worked out variation of PLP along something like the following lines: V provides a use-
ful schema for generating a multitude of formal logics, corresponding to different ways of
specifying cases. Many of these logics will be useful tools for studying philosophically
interesting relationships between statements in natural languages, and which linguistic
relationship is usefully coded up by a particular logic will depend upon which type of
case is most relevant to that relationship. In addition, many of these logics will codify
relations that are worthy of the title ‘consequence’, and thus each of these logics should
be considered a ‘logic’ in some philosophically serious sense of the term.
Of course, if this is all Beall and Restall are up to, then we do not have a version of
SLP, as the claims of the previous paragraph are consistent with there being a unique
logic (corresponding to a particular choice of cases) that captures the unique logical con-
sequence relation in natural language.
Fortunately, Beall and Restall make it clear that their view amounts to much more
than this. They do emphasize that different logics are better or worse than others at cer-
tain tasks:
Obviously, since I endorse classical reasoning, I endorse relevant reasoning in the same sense –
relevantly valid arguments also never take us from truth to untruth. But I think that relevantly
valid arguments do more … Relevantly valid arguments never take us from what is true-in-a-situa-
tion to something not true-in-that-situation. (Restall 2002:428, emphasis added)
They deny, however, that this fact, combined with the fact that different logics codify
different relations, in any way implies that these logics are not all codifying logical conse-
quence. On the contrary, according to Beall–Restall pluralism there is more than one
logical consequence relation in natural language, and each distinct logic obtained by
substituting different classes of cases into V codifies a different consequence relation:
The mere fact that a single logic has a particular role to play (in a particular context) does not
mean that other things are not also logic. To cause problems for pluralism, one needs to show
that a given job is an essential characteristic of consequence; that is, one needs to show that a

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
500 Let a Thousand Flowers Bloom

given application is required of any admissible instance of GTT. As above (and throughout), we
see no reason for imposing such a constraint. (Beall and Restall 2006:99)
(GTT is just the version of V used in Beall and Greg 2006)
Put in the terms introduced earlier, Beall–Restall pluralism is, like Varzi’s view and
Carnap’s tolerance, a dependent pluralism, but unlike Varzi or Carnap, they do not
obtain their pluralism by varying the language in question or the logical ⁄ nonlogical
divide. Instead, the notion of logical consequence has a certain ‘unsettledness’ that ‘affords
different, legitimate precise codifications of the core notion’ (Beall and Restall 2006:104)
and, as a result, the correct logic is relative to which logical consequence relation in natu-
ral language one intends to codify.

6. Logic-As-Modeling
The final destination on our journey through the pluralist terrain is Stewart Shapiro and
Roy Cook’s logic-as-modeling view (the development of the logic-as-modeling view is
independent of Shapiro’s logical tolerance discussed earlier). The logic-as-modeling view
is summarized nicely by Shapiro in Vagueness in Context:
The present claim is that a formal language is a mathematical model of a natural language, in
roughly the same sense as, say, a Turing machine is a model of calculation, a collection of point
masses is a model of a system of physical objects, and the Bohr construction is a model of an
atom. In other words, a formal language displays certain features of natural languages, or ideal-
izations thereof, while simplifying other features. (2006:49)
On the Shapiro ⁄ Cook view, a logic is not meant to be a perfect representation of the lin-
guistic phenomenon being studied. Instead, a logic is a model of that linguistic phenome-
non, and as a result, all of the advantages and limitations that are present in modeling
elsewhere (such as in the empirical sciences) should be expected to reappear in the study
of logic.
The primary application of the logic-as-modeling view has been in the literature on
vagueness (e.g., Cook 2002; Shapiro 2006): The position was, to a large extent, formu-
lated in order to answer worries about how logics formulated in terms of precise mathe-
matical tools could be of use in the study of vague expressions which are, by their very
nature, imprecise (this is the problem of inappropriate precision – see Cook 2002 for discus-
sion). Nevertheless, the view also suggests a means of developing a novel version of logi-
cal pluralism.
The crucial insight of the logic-as-modeling view for our purposes is that there can be
multiple, incompatible, competing models of the same phenomenon:
… with mathematical models generally, there is typically no question of ‘getting it exactly
right’. For a given purpose, there may be bad models – models that are clearly incorrect – and
there may be good models, but it is unlikely that one can speak of the one and only correct
model. There is almost always a gap between a model and what it is a model of. (Shapiro
2006:50)
Thus, when modeling a particular phenomenon, we may arrive at different, and even
incompatible, models by adopting different goals or simplifying or idealizing in different
ways.
As a result, the logic-as-modeling viewpoint immediately provides a version of plural-
ism, albeit a relativistic PLP version: Different logics, viewed as models of various linguis-
tic phenomenon, are correct relative to different theoretical goals, or relative to different

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
Let a Thousand Flowers Bloom 501

ways of simplifying, idealizing, or precisifying the phenomena in question. As noted at


the outset, however, such a pluralism is not controversial.
The logic-as-modeling view does hold out the possibility of a more substantial version
of pluralism, however – one satisfying SLP. Once we realize that logics are merely mod-
els of the consequence relation in natural language, we might wonder whether there
could be two logics such that (i) the logics apply to the same language, and involve the
same interpretation of the logical ⁄ nonlogical divide; (ii) the logics are incompatible – that
is, they validate different sets of inferences; (iii) the logics are, all things considered,
equally good models of logical consequence in natural language; and (iv) there is no third
logic such that this logic is a better model of logical consequence than the two competing
logics.
On the face of it, there seem to be no a priori reasons for ruling out such a scenario:
If Cook and Shapiro are right about logics being models, then the existence of such pair-
wise maximally good logics would seem to be an open question. Unfortunately, no posi-
tive arguments have, as of yet, been given for the existence of such a plurality of logics
within the logic-as-modeling framework.
In ‘Intuitionism Reconsidered’ (2005), however, Cook does present an argument for a
weaker claim – one that illustrates how such an argument for pluralism might proceed.
After surveying the intuitionist’s arguments against classical logic, which are based on the
idea that certain classical logical truths such as excluded middle are, from the intuitionist’s
perspective, epistemically too strong, the intuitionist concludes that intuitionistic logic is
the one, true, correct logic. Cook points out, however, that this move is too quick:
There are continuum-many logics strictly between intuitionstic and classical logic (these
are called intermediate logics), and the intuitionist’s objections to classical logic look less and
less plausible when applied to weaker and weaker intermediate logics. Thus, even if the
intuitionistic worries about classical logic are correct, they don’t generalize to all logics
stronger than intuitionistic logic. After carefully working through the consequences of
this situation for the intuitionist, Cook concludes that:
The intuitionist’s position at this point is rather subtle. The appropriate intuitionist attitude
towards … logical monism, is analogous to the position he recommends with regard to
excluded middle. The intuitionist, on pain of contradiction, cannot advise the classical logician
to accept the negation of excluded middle – he can merely suggest that his rival abstain from
asserting the principle in question. Analogously, the intuitionist need not explicitly accept that
there is some intermediate logic that is just as legitimate as his own favored codification of logi-
cal consequence. He merely ought to refuse to assert that there is no such logic (i.e. refrain
from claiming that his logic is the sole correct one). The intuitionist need not explicitly accept
[logical pluralism], but he has good reason for not asserting [logical monism]. (2005:409)
If Cook is right, then, according to their own principles, the intuitionist has no grounds
for a monistic account of logic – that is, he has no means for ruling out all other logics as
illegitimate. Being careful about the logical complexities of intuitionistic reasoning, this
position is weaker than an acceptance of pluralism (as the intuitionist does not have to
accept ‘monism or pluralism’ any more than he needs to accept any other instance of
excluded middle).
While falling short of a full defense of pluralism of the SLP variety, this argument consti-
tutes at least a promising start for a version of logical pluralism based on the logic-as-model-
ing framework – one that is both a version of SLP and is, unlike the accounts explored
earlier, is a simple pluralism that involves no relativism of any sort. On this view, logics
are models of logical consequence in natural language. Unlike the Beall–Restall view,

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
502 Let a Thousand Flowers Bloom

however, here we have no need for multiple consequence relations in natural language, as
pluralism, on the logic-as-modeling view, would be a result of the inexact fit between the
unique (but imprecise) relation in natural language and a number of precise logics.

7. Conclusion
We began by carefully distinguishing between relativism and pluralism with respect to
logic and sketching a number of different versions of logical pluralism while paying spe-
cial attention to whether and how these variants of pluralism involve various sorts of rela-
tivism. The two primary contenders – Beall–Restall pluralism and the pluralism inherent
in Shapiro and Cook’s logic-as-modeling view – are easily distinguished in terms of
whether or not they involve the choice of correct logic being relative to a choice of
which natural language consequence relation is being codified. We shall conclude by
sketching a final variant of logical pluralism which has, as of yet, been neither described
nor defended, and then mention a few open questions that demand further scrutiny.
We shall call this final version of logical pluralism rabid logical pluralism (RLP). RLP
is obtained by combining Beall–Restall pluralism with Shapiro and Cook’s logic-as-
modeling view. On this view, there is be more than one logical consequence relation
holding of natural language, corresponding to different ways of settling the ‘unsettledness’
of the intuitive notion of consequence, just as Beall and Restall claim. Further, if formal
logics are models, then each of these different natural language relations (or, at least, some
of them) could be codified by multiple maximally good logics – this is the logic-as-
modeling aspect of the account. As a result, RLP is doubly pluralistic, as, on the one
hand, there are different relations in natural language to be codified but, on the other
hand, even if we fix our attention on one of these natural language relations, there might
be more than one formal logic that ‘correctly’ or ‘best’ codifies it.
The purpose here is not to defend RLP, or even to suggest that it is plausible. Instead,
the importance of RLP is that it is an internally coherent view, and thus demonstrates
that Beall–Restall pluralism and the pluralism suggested by the logic-as-modeling view
are not, strictly speaking, competing accounts of the right way to be a logical pluralist.
Rather, these two accounts amount to distinct, but compatible, ways of being a pluralist.
Finally, we move to the open problems: This brief summary of different sorts of plural-
isms regarding logic has left a number of important issues untouched. One such issue is
determining the relations between the various versions of pluralism about logic and corre-
sponding pluralisms about other semantic notions such as truth and meaning. Perhaps
most pressing, however, is determining which logic is to govern reasoning about logics,
if one accepts that there is more than one correct, or legitimate logic. For example, sup-
pose that both intuitionistic and classical logic are correct codifications of the logical con-
sequence relation. Then how should we reason about these logics (and, in particular,
about the connections between them)? Should we restrict ourselves to intuitionistic logic,
applying a ‘better safe than sorry’ methodology, or should we allow ourselves the more
powerful methods of classical reasoning, figuring that if classical reasoning is correct, then
it cannot lead us into error? Answering such questions will be critical for any successful
defense of logical pluralism. But such issues must wait for another time.

Short Biography
Roy T. Cook works in the philosophy mathematics, the philosophy of logic, and the
history of analytic philosophy, with additional interests in the history of early modern

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
Let a Thousand Flowers Bloom 503

philosophy and in the esthetics of popular art (in particular, cinema and comics). He has
authored or co-authored papers appearing in a number of journals including Analysis,
Dialectica, The Journal of Philosophical Logic, The Journal of Symbolic Logic, Mind, The Monist,
The Notre Dame Journal of Formal Logic, Philosophia Mathematica, Synthese, and Studia Leibni-
tiana, as well as contributing to a number of anthologies including Issues on Vagueness,
New Waves in Philosophy of Mathematics, The Oxford Handbook of Philosophy of Mathematics
and Logic, and Revenge of the Liar. He edited The Arché Papers on the Mathematics of Abstrac-
tion (Springer 2007), which collected together important papers on the technical aspects
of Neo-Fregean philosophy of mathematics, and is the author of A Dictionary of Philosoph-
ical Logic (Edinburgh University Press, forthcoming). Roy also serves as the philosophy of
mathematics editor for the Internet Encyclopedia of Philosophy. His current research is
focused on three topics: First, the further development of a Neo-Fregean, ‘logicist’ phi-
losophy of mathematics. Second, an historical investigation into Gottlob Frege’s original,
failed logicist project. Third, the further development of his solution to the Liar paradox
and related semantic puzzles based on the idea that language is indefinitely extensible.
Roy currently teaches at the University of Minnesota, and is an Associate Fellow of
Arché: The Philosophical Research Centre for Logic, Language, Metaphysics, and Episte-
mology at the University of St Andrews in St Andrews, Scotland. Prior to this he was a
visiting professor at Villanova University and a postdoctoral fellow at the Arché Research
Centre. He holds BA in both Philosophy and Political Science from Virginia Polytechnic
Institute and State University (‘Virginia Tech’) and a PhD in philosophy from the Ohio
State University.

Acknowledgments
I would like to thank Stewart Shapiro and Hasok Chang for interesting conversations on
matters related to this article, Greg Restall for extremely useful comments on an early
draft, and the members of the University of Minnesota Undergraduate Philosophy Club for
comments on talk based on these ideas.

Notes
* Correspondence: Roy T Cook, Department of Philosophy, University of Minnesota, 819 Heller Hall, 271 19th
Ave S Minneapolis, MN 55455 USA. Email: cookx432@umn.edu.

1
This section owes much to Chris Swoyer’s excellent essay on relativism in the Stanford Encyclopedia of Philoso-
phy (2003).
2
Parts of this section have been adapted from the discussion of logical consequence and logical pluralism in Cook
(2005).
3
This section owes much to Greg Restall’s ‘Carnap’s Tolerance, Meaning, and Logical Pluralism’ (2002).

Works Cited
Beall, J.C. and Restall, Greg. ‘Logical Pluralism.’ Australasian Journal of Philosophy 78 (2000): 475–93.
—— and ——. ‘Defending Logical Pluralism.’ Logical Consequences: Rival Approaches Papers from the 1999 Society for
Exact Philosophy Conference. Eds. Bryson Brown and John Woods. Stanmore: Hermes, 2001. 1–22.
—— and ——. Logical Pluralism. Oxford: Oxford UP, 2006.
Carnap, Rudolph. The Logical Syntax of Language. Paterson, NJ: Littlefield, Adams, & Co., 1959.
Cook, Roy. ‘Vagueness and Mathematical Precision.’ Mind 111 (2002): 227–47.
——. ‘Intuitionism Reconsidered.’ The Oxford Handbook Of Philosophy Of Mathematics And Logic. Ed. Stewart Shap-
iro. Oxford: Oxford UP, 2005. 387–411.

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd
504 Let a Thousand Flowers Bloom

Gentzen, Gerhard. ‘Investigations into Logical Deduction II.’ American Philosophical Quarterly 2 (1965): 288–306.
Restall, Greg. ‘Carnap’s Tolerance, Language Change and Logical Pluralism.’ Journal of Philosophy 99 (2002):
426–43.
Shapiro, Stewart. Philosophy of Mathematics: Structure and Ontology. Oxford: Oxford UP, 1997.
——. Vagueness in Context. Oxford: Oxford UP, 2006.
Swoyer, Chris. ‘Relativism.’ Stanford Encyclopedia of Philosophy. 2003 <http://plato.stanford.edu/entries/relativism/>.
Last accessed date December 20, 2009
Tarski, Alfred. ‘On the Concept of Logical Consequence.’ Logic, Semantics, Metamathematics. 2nd ed. Ed. John Corc-
oran. Indianapolis, IN: Hackett, 1983. 409–20.
Varzi, Achille. ‘On Logical Relativity.’ Philosophical Issues 10 (2002): 197–219.

ª 2010 The Author Philosophy Compass 5/6 (2010): 492–504, 10.1111/j.1747-9991.2010.00286.x


Journal Compilation ª 2010 Blackwell Publishing Ltd

You might also like