Australasian Journal of Philosophy

ISSN: 0004-8402 (Print) 1471-6828 (Online) Journal homepage: http://www.tandfonline.com/loi/rajp20

How Do Logics Explain?

Gillman Payette & Nicole Wyatt

To cite this article: Gillman Payette & Nicole Wyatt (2017): How Do Logics Explain?, Australasian
Journal of Philosophy, DOI: 10.1080/00048402.2017.1342674

To link to this article: http://dx.doi.org/10.1080/00048402.2017.1342674

Published online: 22 Jun 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=rajp20

Download by: [The UC San Diego Library] Date: 23 June 2017, At: 09:33
AUSTRALASIAN JOURNAL OF PHILOSOPHY, 2017
https://doi.org/10.1080/00048402.2017.1342674

How Do Logics Explain?

Gillman Payettea and Nicole Wyatt b

a
University of British Columbia; bUniversity of Calgary

ABSTRACT
Anti-exceptionalists about logic maintain that it is continuous with the empirical
sciences. Taking anti-exceptionalism for granted, we argue that traditional approaches
to explanation are inadequate in the case of logic. We argue that Andrea Woody’s
functional analysis of explanation is a better fit with logical practice and accounts
better for the explanatory role of logical theories.

ARTICLE HISTORY Received 22 August 2016

KEYWORDS Logic; explanation in science; methodology of logic; philosophy of logic

1. Introduction
Logic is sometimes treated as an exceptional science. For exceptionalists, the truths of
logic, unlike the truths of biology, physics, or ecology, are supposed to be a priori, nec-
essary, or universal. In virtue of this, exceptionalists take the study of logic to be differ-
ent in kind from the so-called empirical sciences. In contrast, anti-exceptionalists about
logic maintain that logic is continuous with other sciences.1
As anti-exceptionalists, we maintain that, logic, qua the study of consequence, is just
like every other science: it is subject to the same methodological pressures as other sci-
ences, and the same epistemological questions. The issue with which we are concerned
is that of the role of explanation in logic. We focus on two questions: to what extent do
logical theories explain logical consequence, and how do they do so?
Taking anti-exceptionalism for granted, we consider various approaches to explana-
tion in philosophy of science. We argue that traditional approaches to explanation are
simply non-starters for logic. We then turn to Andrea Woody’s [2015] suggestion that
philosophers of science should seek to understand explanation via its role in scientific
practice. Applying this approach to logic, we show that the anti-exceptionalist can bet-
ter account for the role that logic plays in explaining the validity of arguments.

2. Anti-Exceptionalist Beginnings
Logical research can be purely formal, resulting in what Goldstein [1992] calls a
‘smooth’ logic. Our interest here is exclusively in ‘rough’ logics—logics that are devel-
oped, at least in part, for the purpose of understanding natural language reasoning

1
Quine is a prominent anti-exceptionalist [Quine and Ullian 1970; Quine 1986]. Other defenders include Maddy
[2002], Priest [2014], Russell [2015b], Hjortland [2017], and Williamson [forthcoming].
© 2017 Australasian Association of Philosophy
2 GILLMAN PAYETTE AND NICOLE WYATT

(including reasoning in scientific contexts).2 Just as the disagreement between

advocates of classical logic and advocates of relevance logic is not concerned with
either system understood merely as a smooth logic, the disagreement between the anti-
exceptionalist and the exceptionalist is not concerned with the purely formal develop-
ment of logics. Logics, as we use the term in this paper, are theories of which natural
language arguments are correct.
Of course, the correctness of deductive arguments is not observed in the same way
as, or by the same mechanism through which, the number of bacteria and their sur-
rounding conditions are observed. The data of logic are the arguments that pass muster
in public intellectual life, broadly construed.3 One should not be misled by this into
thinking that, for the anti-exceptionalist, logic is a theory of our judgments of deductive
correctness and incorrectness. A theory of judgments is a psychological theory explain-
ing why we make this or that judgment. This almost certainly strikes you, as it does us,
as the wrong target for logical theory. Logics should be capable of explaining the cor-
rectness of arguments that do not seem correct, but are, for example, complicated theo-
rems of mathematics. Furthermore, it should explain the incorrectness of arguments
that seem correct but are not. Of course, our judgments will play some role in logical
practice, but they are not the focus. The observations that generate these data are fre-
quently shaped by our logical theories. But, for the anti-exceptionalist, this should be
no surprise, given the more general concerns that can be, and have been, raised about
the theory-ladenness of observation across the sciences.4
Scientific theories serve a variety of purposes. Theories of bacterial reproduction pre-
dict whether conditions are conducive to some kinds of bacteria reproducing. They can
also explain a set of observations of bacteria numbers.5 Keeping with the assumption of
anti-exceptionalism, logics will also have multiple uses. Our focus in this paper is to
account for just one of these uses: explanation. How is it that a logic provides an expla-
nation of argument correctness or incorrectness?
When we want an explanation of some event or observation E, we use or apply sci-
entific theories. Traditionally, philosophers of science have focused on giving success
or adequacy criteria for individual explanations. This approach underlies a number of
well-known models of explanation, such as the deductive-nomological (DN) model
and the unificationist model. Naturally, as anti-exceptionalists, we would expect these
models to apply to logic in so far as they apply to science generally.
From this perspective, one role for the philosopher of logic is to give an account of the
adequacy conditions for logical explanation. One challenge for any such account will be
the apparent resilience of logics in the face of counterexample. Counterexamples exist for
any proposed law of logic,6 and yet logical nihilists are a rare breed. Any account of ade-
quacy conditions for logical explanations must account for this phenomenon.
There is a complication when applying these traditional models of explanation to
logic. In the next section we explain that complication, and argue that, as a result,
focusing on the adequacy conditions for particular explanations is not helpful in under-
standing the explanatory power of logical theories. We then turn instead to functional

2
This is what Priest [2014] would refer to as the canonical application of logic.
3
Aberdein and Read [2009: 622] express a similar view of the empirical data for logic.
4
For an overview the theory-ladenness of observation, see Bogen [2014].
5
For another example, Odenbaugh [2005] gives a list of roles played by models in theoretical ecology.
6
See [Russell 2015a] for an overview.
 AUSTRALASIAN JOURNAL OF PHILOSOPHY 3

analyses of scientific explanation, as suggested by Andrea Woody [2015], for an appro-

priate model.

3. Traditional Models of Explanation

To make our point, we explore in some detail what goes awry with traditional models of
explanation when applied to logic. The problem with the traditional models is that they
end up in a circularity that doesn’t aid in understanding what was to be explained. The
DN model exemplifies this phenomenon.
According to the DN model, an explanation of some phenomenon E (the explanan-
dum) consists of a set of sentences fC1 ; . . . ; Cn g representing factual conditions
together with a set of sentences fL1 ; . . . ; Lm g representing laws, such that the argument

C1 ; . . . ; Cn
L1 ; . . . ; Lm
;E

is sound, and at least one of the laws Lj is essential to it. That is, the factual conditions
(Ci ) along with true laws (Lj ) entail E, and there is at least one law such that, if it were
removed, the argument would no longer be valid.
Turning to explanation in logic, what we are looking to explain is the validity/inval-
idity of some argument consisting of (fully interpreted) natural language sentences:
ðA; bÞ. The E in question, then, says either ‘ðA; bÞ is valid’ or ‘ðA; bÞ is invalid.’ Logical
theories explain the validity or invalidity of these arguments in terms of properties of
formal systems—at least, that is the contemporary practice, and similar practices go
back to Aristotle. In the past the formal systems were less mathematical, but they still
translated a natural language argument ðA; bÞ into a schematic, formal language repre-
sentation (G, ’), and assessed (G, ’) for validity in the precise terms of the logical the-
ory. Contemporary logical theories are just richer: formal language, proof theory, and
model theory. For an Aristotlean system, the logical theory would be a collection of
schematic syllogisms, but the process is the same.
To fit this methodology into the DN model, we interpret the Ci s—the conditions—
as stating how the sentences of the particular argument ðA; bÞ are translated into
(G, ’). There is no requirement that this translation be uniform. We may sometimes
understand a natural language ‘if … then …’ as a material conditional and in other
cases capture it in some other way. The laws (Lj s) are a specification of the formal
semantics/proof theory of the logic.
The logical theory implies either G ‘ ’ or G 0 ’ as a consequence, where ‘ means
‘follows according to the system’. That is, a valid argument can be given, starting from
the Lj s and ending with, say, G ‘ ’. Given that (G, ’) translates ðA; bÞ (which means
that the conditions Ci are correct) and that the logical theory (Lj ) is correct, it follows
that ðA; bÞ is valid. According to the DN model, this is exactly what is required for the
logical theory to explain the validity of ðA; bÞ.
In this model of explanation, we see a fit with the practice of formalization of argu-
ments, but there are tensions. The natural sciences might have trouble finding true laws
or troubles with the direction of proper explanation using the DN model, but they do
not have trouble with circularity. Natural scientists do not have to lean on the kind of
4 GILLMAN PAYETTE AND NICOLE WYATT

phenomenon that they are attempting to explain in order to offer an explanation using
the DN model. In chemistry, although the laws regarding 7-hydroxyphenoxazone chro-
mophore are essential to an explanation of why litmus paper turned red when dipped
in an acid, those laws are part of the explanans. They do not constitute the relationship
between explanans and explanandum. More generally, the adequacy conditions for
explanation (a valid argument) are not also the subject of the explanandum.
This circularity creates two problems. Normally, the ability of a theory to explain phe-
nomena provides additional support for the theory (how much support is debated). But
when we apply the DN model to logic, the adequacy of an explanation depends upon the
same kind of phenomenon that is to be explained: namely, the validity of an argument.
This raises doubts that explanatory power can provide any support for logical theories.
Moreover, it is unclear how we could establish the correctness of the conditions (the
translation) in an independent way. Given a translation, modern mathematized logic
allows us to establish validity and invalidity by rigorous but informal mathematical
proofs. However, for the overall argument which establishes the adequacy of the expla-
nation to be sound, the analyses of A and b by G and ’ must be correct, and the seman-
tics/proof-theory for the formal language must also be correct. But what provides
support for the correctness of the conditions and laws/semantics in this case is that the
explanation is successful. So, for the argument to be sound, and thus an adequate expla-
nation, it must already be an adequate explanation.
This same pattern of argument can be applied to various other accounts of explana-
tion in science, like Kitcher’s [1989] unification account. The basic problem that arises
in traditional accounts that focus on giving adequacy conditions for individual scien-
tific explanations is that they rely, at some point, on logical consequence to fill out the
relationship between explanans and explanandum. This is even the case for highly non-
traditional accounts, like those from Hindriks [2008] and Mizrahi [2012], which do not
require the laws to be true, or even, in the former case, predictive. Once the adequacy
condition relies on logical consequence, the possibility for circularity rears its ugly, but
round, head.
Alternative accounts of scientific explanation based in causal relations or mecha-
nisms strike us as non-starters: premises do not cause their conclusions. What is
needed is an account of explanation that can be applied in logic without circularity.
Woody [2003, 2013, 2015] offers an account of explanation that does not focus on ade-
quacy conditions, which we explore in the next section. Following this, we consider an
example of explanation in logic from Yalcin [2012] and we interpret it through
Woody’s methodological lens.

4. Woody’s Functional Approach

Traditional accounts focus on adequacy conditions for explanation, and share the
assumption that a satisfactory account of those conditions involves an analysis of the
relationship between the explanation and the explanandum. Accounts differ as to
whether that relationship should be understood inferentially, causally, counter-factu-
ally, or mechanistically.7 Problems with models of explanation focusing on adequacy

7
As Woody [2015] points out, although Kitcher’s [1989] account focuses not on individual explanations but on
the theoretical status of systems as a whole, the goal is still to answer the question of when a putative explanation
is adequate.
 AUSTRALASIAN JOURNAL OF PHILOSOPHY 5

conditions are not limited to the case of logic. As Woody [2015: 80] observes, tradi-
tional models of explanation all fail to account for the diversity and pluralism of expla-
nation in modern scientific communities. And, as Nancy Cartwright’s [1983] well-
known discussion of ceteris paribus laws in the physical sciences establishes, scientific
explanations in many cases turn on a number of convenient fictions—so-called laws
which we know to be false and for which we have more accurate refinements. Woody
suggests that philosophers would do well to consider other questions concerning expla-
nation, especially the question ‘What role does explanation play in science?’ This ques-
tion, suggests Woody, lends itself to a functional analysis, as understood by Hempel
[1965: 304–5]:
a functional analysis is invoked to explain … some recurrent activity or pattern of behaviour …
And the principal objective of the analysis is to exhibit the contribution which the behaviour
pattern makes to the preservation or the development of the individual or group in which it
occurs.

The functional perspective asks, ‘What role does explanatory discourse, and explana-
tory activity more generally, play in the practice of science?’ [Woody 2015: 80].
This perspective allows Woody to bypass the stalemate between traditional accounts
of the nature of explanation, focusing instead on the role that explanation plays in con-
stituting intelligibility for scientific disciplines and sub-disciplines [ibid.: 81]:
because explanatory discourse inculcates particular patterns of reasoning, it functions to sculpt
and subsequently perpetuate communal norms of intelligibility. In effect, explanations encode
the aims and values of particular scientific communities, telling practitioners what they should
want to know about the world and how they should reason to get there.

Cartwright also links the continuing relevance of scientific fictions to the function of
scientific explanation [1983: 47–8]:
[the] reasons have to do with the task of explaining. … we have to decide what kinds of factors
can be cited in explanations. One thing that ceteris paribus laws do is to express our explanatory
commitments. They tell what kinds of explanations are permitted.

We can see how this works by considering the ideal gas law. The law is an equation

pV ¼ nRT

relating three variables, pressure, volume, and temperature. If taken literally, it applies
only to ‘ideal gases’—gases in the limit of zero pressure—which is to say, to no gases at
all. Consequently, the law is either false, or, if the ceteris paribus clause is spelled out,
never applies. Further, the law is used even when the gases under study are not ones for
which it provides even a phenomenological approximation. What, then, is going on?
According to Woody, the ideal gas law serves to identify core features of gases. It
asserts that gases are composed of small, widely separated, particles exerting negligible
forces on each other. Against the background of this theoretical framework for gases,
the law is prescriptive rather than descriptive. It serves to focus the attention of chem-
ists on particular properties of gases—pressure, volume, and temperature. It ‘instructs
chemists in how to think about gases as gases’ [Woody 2013: 1574]. Even in cases in
which the law does not approximate gas behaviour, chemists conceptualize gases in
terms of deviations from this ideal [Woody 2015: 82]. Importantly for what is to come,
Woody also observes that the law does important conceptual work with respect to the
6 GILLMAN PAYETTE AND NICOLE WYATT

concept of temperature. Human perception of heat is subjective and inherently com-

parative, and accordingly cannot ground an objective concept. The ideal gas law pro-
vides for a partial quantitative definition of temperature in terms of its relationship to
volume [Woody 2013: 1575].
Two of Woody’s conclusions are centrally important to what follows. First, explana-
tory discourse is social and involves laws as idealizations that articulate, not universal
truths, but ‘inferential scaffolding’ that allows those in the relevant community to for-
mulate a common conceptual framework. Second, members of a given discipline or
sub-discipline share a set of exemplary explanations, which serve to set standards for
explanation for that community. Disciplines share a robust set of assumptions that
determine what sorts of information and/or argument can play a role in explanations
[Woody 2015: 81–5].

5. Marbles, Models, Modus Tollens

How do explanations function in logic? What do they tell us about intelligibility in logic
and about the aims and values of the logical community? If explanation in logic is on a
continuum with other sciences, then these are the questions that the functionalist asks.
We need to pay the sorts of close attention to the actual practices of logicians that con-
temporary philosophers of science pay to the actual practices of scientists.
In this spirit we will take as a case study Seth Yalcin’s [2012] paper entitled ‘A Coun-
terexample to Modus Tollens’. The titular counterexample takes the following form
[ibid.: 1002]:
An urn contains 100 marbles: a mix of blue and red, big and small. The breakdown:

Big Small
Red 30 10
Blue 10 50

A marble is selected at random and placed under a cup. This is all the information given about
the situation. Against this background, the following claims about the marble under the cup are
licensed:
(P1) If the marble is big, then it’s likely red.
(P2) The marble is not likely red.
However, from these, the following conclusion does not intuitively follow:
(C1) The marble is not big.

A substantial portion of Yalcin’s paper is devoted to showing that this is a genuine case
of modus tollens failure, not to be explained away by creative descriptions of the logical
form of the argument. Our primary focus, however, is on his diagnosis of this failure.
Yalcin’s conclusion is that we should think of belief in sentences which, like (P1) and
(P2), involve probability claims, as giving certain odds to the outcomes. But, on this
way of thinking, believing (P1) and (P2) does not involve ruling some possibilities in or
out [ibid.: 1016]. This suggests, argues Yalcin, that we should model the semantic values
of sentences like probably-’ not in terms of possible worlds, but instead in terms of con-
straints on information states. In his paper’s appendix, Yalcin makes a start on this
 AUSTRALASIAN JOURNAL OF PHILOSOPHY 7

project by presenting two different formal semantics for a language including as opera-
tors the conditional, ‘might’, ‘probably’, negation, and conjunction. Both of the semantics
invalidate cases of modus tollens involving the probability operators of the language while
preserving the validity of other cases of modus tollens.
What does Yalcin’s paper tell us about explanation in logic? The language presented
in the appendix includes just enough vocabulary, and the model just enough structure,
to capture the arguments of concern. It is a criterion for the success of the models that
they also correctly capture other arguments involving just this stuff (namely, the vocab-
ulary and structure) and no more. But it is not a requirement that the formal semantics
capture all of the possible arguments involving this vocabulary and structure. Yalcin
identifies a number of other cases of modus tollens failure that are not fully accounted
for by his models [ibid.: 1002–3].
A simplistic approach to theory appraisal might assume that this is just piecemeal
theorizing. On that picture, the goal is a complete logical theory covering all of natural
language, and partial theories are justified in virtue of their role in getting us a complete
theory. Even setting aside questions about attainability, it is unlikely that such a theory
would serve any explanatory goals. The resulting formal language would be, perforce,
as complex and difficult to master as any natural language. Idealization and simplifica-
tion play an important role in the explanatory power of other sciences; so, too, in logic.8
It seems to us that in Yalcin’s paper both modus tollens and classical logic more gener-
ally (as well as the standard semantics that support them) are serving a similar explanatory
role to the ideal gas law, as described by Woody. Yalcin’s counterexample does not serve
to put modus tollens or the logics containing it into disrepute, as someone who takes truth
as an important feature of explanations might expect. On the other hand, the counterex-
ample is not to be explained away—as noted, Yalcin spends much of the paper arguing
against reinterpreting the counterexample. What he does do is to offer an account of the
‘problem’ cases in terms of deviations from the standard semantics of the conditional.
Generally, Yalcin’s treatment conforms to the expectation that logical explanation
will be of the same kind as is found in classical logic. To echo Cartwright [1983], the
model appeals to the right sorts of factors, in that it uses the right kinds of machinery
for the construction of the formal language and the model theoretic semantics. More
specifically, Yalcin offers an analysis of his counterexamples on which logics endorsing
modus tollens can be seen as assuming the absence of probability operators. This is
analogous to the assumption that gas particles are not subject to forces other than colli-
sions with other particles of gas in the ideal-gas law. He then offers alternatives to clas-
sical semantics for the conditional that incorporate the familiar machinery of model
theory. To use a phrase of Woody’s, inferential scaffolding for Yalcin’s account is pro-
vided by the standard structure of a formal language and its interpretation in the formal
semantics.9
Taking stock, we have given the role of logical theories explaining validity a func-
tional account, by using Woody’s approach. As in other disciplines, we find that certain
logical theories can retain a privileged status, despite their known inaccuracy, because

8
See Odenbaugh [2005] for a rich discussion of the function of models that do not display any fit with the phe-
nomena they purport to model. According to Odenbaugh, a principal function of such models in ecology is to
serve as a baseline against which actual systems can be understood in terms of deviations.
9
We believe that Yalcin’s approach also has similarities to the how-possibly models [Brandon 1990] used in ecol-
ogy [Odenbaugh 2005]. We intend to explore this relationship elsewhere.
8 GILLMAN PAYETTE AND NICOLE WYATT

of the role that they play in establishing standards for intelligibility for logic. When
explanation in logic is viewed from the functional perspective, there is no need to
assume the existence of a univocal semantics for the ‘logical’ words of natural lan-
guages. Logics are used to explain particular inferences, not to account for the ‘mean-
ings’ of natural language vocabulary.10 We also have seen that ‘how-possibly’ models
play a similar role in logical explanation to that which they play in other model-centric
disciplines. Since the functional analysis of explanation avoids adequacy conditions
altogether, there is no point at which explanations rely for their success on deductive
consequence, and no point where the circularity that troubled traditional models of
explanation can creep in. We are now in a position to defend our anti-exceptionalism.

6. Anti-Exceptionalist Endings
We’ve argued that if logic is non-exceptional and thus explanatory, then accepting the
functional picture of explaining—wherein theories articulate norms of intelligibility—
makes good sense of logical practice. We will now consider and reject two exceptional-
ist accounts that deny that logic aims at explanation: one that maintains that logics
serve no explanatory function [Szab o 2012], and another that takes logics to aim merely
at explication. We start with the latter.
Explications are meant to replace vague, intuitive notions with precise ones that
advance our understanding. This is seen as an alternative to explanation [Dutilh
Novaes and Reck 2017]. However, explication is common in science, so offering expli-
cations does not make logic exceptional. Furthermore, from the functionalist perspec-
tive, explication is an aspect of explanation—recall the case of temperature in section
4.11 The ideal gas law allows us to replace the comparative notion of temperature with
an objective explication. Explications are successful when they give rise to ‘rich explana-
tory practices’ that advance our epistemic aims [Woody 2015: 86]. So, explication isn’t
a competitor to explanation; it is part of explanation.
Szabo argues, contra Davidson, that we cannot hope to arrive at an explanation for
logical relationships or our knowledge of them by providing a compositional semantics
for natural language: ‘Inferences can be subtle, and audiences can be dull. Either way,
validity is sometimes in need of explanation’ [2012: 125]. Traditionally, claims Szabo,
we ‘explain’ validity by disregarding the parts of the sentences that do nothing for truth
preservation, by replacing them with schematic letters. He calls that ‘explanation by
abstraction’. He continues [ibid.: 125, emphasis ours]:
Explanation by abstraction is just a step away from no explanation at all. It is roughly akin to
saying ‘The validity of this inference is self-explanatory—you will see it for yourself as soon as I
remove the irrelevant details that obscure your insight.’ It is an attractive view that the limits of
logic are set by the scope of adequate explanation by abstraction. Logical validity is epistemically
fundamental; to explain it all we can do is clear the dust and hope that validity will shine
through. Other validities are not self-evident; to explain them we have to appeal to necessary
truths.

10
See Jennings [1994] for an exploration of flexibility in speaker interpretations of logical words. Thanks are due
to an anonymous referee for pressing us to make explicit this aspect of the account and for reminding us of the
relevance of Jennings’ work.
11
A fuller account of the relationship between explication and the functionalist account is warranted, but here is
not the place.
 AUSTRALASIAN JOURNAL OF PHILOSOPHY 9

So, while an explanation of the truth of ‘This is red, so this is coloured’ needs to appeal
to some necessary truth, ‘This is red, so it is red or green’ doesn’t appeal to a necessary
truth for explanation. In fact, there is no need for explanation at all in this case. That is
what makes it a logical truth. He ends by saying [ibid.: 125–6]:
I endorse this view. I will not call it a theory, for obvious reasons: short of a serious account of
what it is for a validity to be self-explanatory, it explains nothing. It certainly does not tell us
why logical validities are valid or how we know that they are. …To say that something is self-
explanatory does not mean that we can provide a real explanation for it that relies on nothing
beyond itself. Rather, it means that no substantive explanation is needed.

So, logical validity doesn’t need explanation. That is, once the dust clears one has no
reason to ask ‘why is that true?’ or ‘why does that follow?’ The problem for Szabo’s
view is that we do just that.
First, there are plenty of logical validities that aren’t self-evident, even when given
their formal representation. It would be uncharitable to say that Szabo was unaware of
that, but it does raise a problem. Perhaps the complex logical validities can have an
explanation in terms of the ‘obvious ones’ (via proof), but then the complex validities
lose their epistemic fundamentality. But we wouldn’t say that they are any less logical
for that. The obvious reply is that some of the validities are in need of explanation. We
can accept that.
Second, people often ask for explanations of logical validities of the fundamental/
simple type that Szab o uses as examples. Moreover, the explanation by abstraction
method starts by eliminating anything irrelevant to truth preservation. Isn’t truth pres-
ervation explaining validity, then? To accept Szab o’s view of some logical truths as epi-
stemically basic is to rule out truth preservation as relevant to logical validity.12
Third, if we accepted certain validities as epistemically basic, then we would need to
explain the sub-discipline of the philosophy of logic which regularly asks for explana-
tions of just the sort for which they shouldn’t need to ask. Are all of us philosophers of
logic just being pig-headed? We are inclined to say ‘no’, because there are putative
counterexamples to all basic principles (cf. Russell [2015a]). We will not rehearse the
examples here because, even if they are not genuine counterexamples, that itself
requires explanation. Responding with ‘The inference/validity is epistemically basic’ is
a dogmatic and insufficient defence.
In contrast, Woody’s functional approach to the role of explanatory discourse allows
for a more satisfactory explanation of the apparent ‘epistemic basicness’ of (some) logi-
cal validities. The process of explanation by abstraction embodies the norms that gov-
ern explanations of logical validity. The things relevant to the determination of validity
are encoded in the process of abstraction, which are partly determined by the aims and
values of the logical community, and give the communal norms of intelligibility for the
discipline. From this anti-exceptionalist perspective, explanations by abstraction are
genuine explanations, but neither they nor the abstractions themselves—in the form of
the formula and argument schemata—need to bear the foundationalist weight that
Szabo’s account places on them. Even when such abstract ‘laws’ do not predict or
describe validity, they are part of the scaffolding on which we give explanations for why

12
Field [2009, 2015] also holds that truth preservation shouldn’t be used in the definition of validity, and that log-
ical validity is basic. For Field, the conceptual role that validity plays is a normative one regarding belief
maintenance.
10 GILLMAN PAYETTE AND NICOLE WYATT

they do not hold. They need not be epistemically basic, or even valid, to play an explan-
atory role, because they can still contribute to our understanding in the cases where
they fail. Logical truths and schematic validities are epiphenomena of the norms of
explanatory practice in logic.
In short, an anti-exceptionalist approach to logic, and thus to the explanatory power
of logical theories, solves a problem that challenges the exceptionalist view. For the
exceptionalist, the epistemic basicness of validity is brought into question both by the
proliferation of logics and by the apparent absence of exceptionless logical laws. But
this problem simply evaporates, from the anti-exceptionalist perspective offered here.13

Funding
Gillman Payette would like to thank the Social Sciences and Humanities Research Council of Canada
for supporting this research through a Banting Postdoctoral Fellowship.

ORCID
Nicole Wyatt http://orcid.org/0000-0001-7212-1117

References
Aberdein, A. and S. Read 2009. The Philosophy of Alternative Logics, in The Development of Modern
Logic, ed. L. Haaparanta, Oxford: Oxford University Press: 613–723.
Bogen, J. 2014. Theory and Observation in Science, The Stanford Encyclopedia of Philosophy (Summer
2014 ed.), ed. E. N. Zalta, URL = https://plato.stanford.edu/entries/science-theory-observation
Brandon, R.N. 1990. Adaptation and Environment, Princeton: Princeton University Press.
Cartwright, N. 1983. How the Laws of Physics Lie, Oxford: Clarendon Press.
Dutilh Novaes, C. and E. Reck 2017. Carnapian Explication, Formalisms as Cognitive Tools, and the
Paradox of Adequate Formalization, Synthese 194/1: 195–215.
Field, H. 2009. What is the Normative Role of Logic? Proceedings of the Aristotelian Society S83: 251–68.
Field, H. 2015. What is Logical Validity? in Foundations of Logical Consequence, ed. C.R. Caret and O.
T. Hjortland, Oxford: Oxford University Press: 33–70.
Goldstein, L. 1992. Smooth and Rough Logic, Philosophical Investigations 15/2: 93–110.
Hempel, C.G. 1965. The Logic of Functional Analysis, in his Aspects of Scientific Explanation and Other
Essays in the Philosophy of Science, New York: The Free Press: 297–330.
Hindriks, F. 2008. False Models as Explanatory Engines, Philosophy of the Social Sciences 38/3: 334–60.
Hjortland, O.T. 2017. Anti-Exceptionalism about Logic, Philosophical Studies 174/3: 631–58.
Jennings, R.E. 1994. The Genealogy of Disjunction, New York: Oxford University Press.
Kitcher, P. 1989. Explanatory Unification and the Causal Structure of the World, in Scientific Explana-
tion, ed. P. Kitcher and W. Salmon, Minneapolis: University of Minnesota Press: 410–505.
Maddy, P. 2002. A Naturalistic Look at Logic, Proceedings and Addresses of the American Philosophical
Association 76/2: 61–90.
Mizrahi, M. 2012. Idealizations and Scientific Understanding, Philosophical Studies 160/2: 237–52.
Odenbaugh, J. 2005. Idealized, Inaccurate but Successful: A Pragmatic Approach to Evaluating Models
in Theoretical Ecology, Biology and Philosophy 20/2-3: 231–55.
Priest, G. 2014. Revising Logic, in The Metaphysics of Logic, ed. P. Rush, Cambridge: Cambridge Uni-
versity Press: 211–23.
Quine, W.V.O. and J.S. Ullian 1970. The Web of Belief, New York: Random House.

13
We, the authors, would like to thank the anonymous referees for their helpful comments and suggestions. We
would also like to recognize Gillian Russell, Teresa Kouri, Samuel Khoo, and attendees of the 2016 meeting of the
SEP in Miami for their helpful discussion of our ideas.
 AUSTRALASIAN JOURNAL OF PHILOSOPHY 11

Quine, W.V.O. 1986. Philosophy of Logic, 2nd edn, Cambridge, MA: Harvard University Press.
Russell, G. 2015a. Could There Be no Logic? Society for Exact Philosophy (Talk).
Russell, G. 2015b. The Justification of the Basic Laws of Logic, Journal of Philosophical Logic 44/6: 793–
803.
Szabo, Z.G. 2012. Against Logical Form, in Donald Davidson on Truth, Meaning, and the Mental, ed.
G. Preyer, Oxford: Oxford University Press: 105–26.
Williamson, T. forthcoming. Semantic Paradoxes and Abductive Methdology, in The Relevance of the
Liar, ed. B. Armour-Garb, Oxford: Oxford University Press.
Woody, A.I. 2003. On Explanatory Practice and Disciplinary Identity, Annals of the New York Acad-
emy of Sciences 988/1: 22–9.
Woody, A.I. 2013. How Is the Ideal Gas Law Explanatory? Science and Education 22/7: 1563–80.
Woody, A.I. 2015. Re-Orienting Discussions of Scientific Explanation: A Functional Perspective, Stud-
ies in History and Philosophy of Science Part A 52: 79–87.
Yalcin, S. 2012. A Counterexample to Modus Tollens, Journal of Philosophical Logic 41/6: 1001–24.