The British Society for the Philosophy of Science

Mathematical Explanation and the Theory of Why-Questions

Author(s): David Sandborg
Source: The British Journal for the Philosophy of Science, Vol. 49, No. 4 (Dec., 1998), pp. 603-
624
Published by: Oxford University Press on behalf of The British Society for the Philosophy of Science
Stable URL: http://www.jstor.org/stable/688133 .
Accessed: 04/07/2011 00:36

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=oup. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Oxford University Press and The British Society for the Philosophy of Science are collaborating with JSTOR to
digitize, preserve and extend access to The British Journal for the Philosophy of Science.

http://www.jstor.org
 Brit. J. Phil. Sci. 49 (1998), 603-624

Mathematical Explanation and the

Theory of Why-Questions
David Sandborg
ABSTRACT
Van Fraassenand others have urged that judgements of explanationsare relative to
why-questions;explanationsshould be considered good in so far as they effectively
answer why-questions.In this paper,I evaluate van Fraassen'stheory with respect to
mathematical explanation. I show that his theory cannot recognize any proofs as
explanatory.I also present an example that contradictsthe main thesis of the why-
question approach-an explanation that appears explanatorydespite its inability to
answer the why-question that motivated it. This example shows how explanatory
judgements can be context-dependentwithoutbeing why-question-relative.
1 Introduction
2 Exposition of van Fraassen's theory
3 An example of mathematical explanation
4 Why-question analysis of Polya's explanation
5 Van Fraassen's theory and mathematical explanation
6 The why-question approach

1 Introduction
Philosophers of science have long described explanations as answers to why-
questions.1 The view that this is the most profitable way to study explanations
has gained recent prominence. For instance, Alan Garfinkel writes, 'Attending
to the questions rather than the answers and looking for the implicit question
hiding behind the answer are a useful device for analysing explanations and
understanding historical shifts' ([1981], p. 8). Recently, Jaakko Hintikka
and Ilpo Halonen have written on the subject [1995]. Bas van Fraassen, the
best-known supporter of this view, writes:
An explanationis not the same as a proposition,or an argument,or a list of
propositions;it is an answer. (Analogously,a son is not the same as a man,
even if all sons are men, and every man is a son.) An explanationis an
answerto a why-question.So, a theoryof explanationmustbe a theoryof
why-questions([1980], p. 134, my emphasis).
1 For instance, see Hempel ([1965], p. 334). Sylvain Bromberger[1962, 1966] also developed an
approachto explanationcentred on this insight about question-answering(though he did not
believe that why-questionsexhaustedexplanatoryrequests).

? OxfordUniversity Press 1998

604 David Sandborg

I shall call the approachto explanationsharedby these authorsthe why-

questionapproach.2Its most distinctiveclaim concernsexplanatoryevaluation:
explanationscan only be properlyevaluated with respect to why-questions.
Garfinkelcalls this thesis explanatoryrelativity.For him, an explanationmust
be evaluatedin the context of a contrastspace, specified in the why-question.
He says, 'I mean this as a claim abouthow to explainexplanations:the contrast
space is a basic presuppositionof the explanationcontext, an additionalpiece
of structurenecessary to explain how explanationsfunction' ([1981], p. 28).
Similarly,van Fraassensays, 'Since an explanationis an answer,it is evaluated
vis-a-vis a question, which is a request for information' ([1980], p. 156).
Following van Fraassen, Resnik and Kushner state this thesis strongly:
'[N]othing is an explanation simpliciter but only relative to the context
dependentwhy-question(s)that it answers' ([1987], p. 153). In this paper, I
will argue against this approach,focusing on van Fraassen'stheory of why-
questionsin particular.I will grantthatexplanationsdo answerwhy-questions
and explanatoryevaluations are context dependent,but claim that the why-
question approachnonetheless misses crucial aspects of certain explanatory
evaluations.
This essay will take the somewhatunusualapproachof consideringmathe-
matical explanations. Though van Fraassen did not have mathematics in
mind when he formulated his theory, an adequate why-question oriented
theory of explanations should address mathematical,as well as empirical,
why-questions. Analogous to the claim that scientific explanationsare only
distinguished by the fact that they invoke scientific information ([1980],
pp. 155-6), mathematicalexplanations should differ from other types only
in their subject-matter;not in any fundamentalway. Since why-questionscan
be and are asked and answeredabout mathematicalfacts (for instance, after
having been informed that 1-1 1+ -1+... converges to ?, it is certainly
reasonableto ask why this is so), as well as empiricalones, the why-question
approach should be adaptable to mathematical explanations. Resnik and
Kushner[1987], in responseto Steiner [1978], have also studiedmathematical
explanations, and though they do not fully endorse van Fraassen's theory,
they endorse the why-question approachin general. We shall see, however,
that mathematicalexplanations create special problems for van Fraassen's
why-questiontheory.
Aftera briefreview of van Fraassen's theoryof why-questions,I will present
an example of mathematicalexplanation,supplementingthose alreadygiven
by Steiner[1978], Resnik andKushner[1987], andKitcher[1983, 1989]. I will
2 Achinstein's illocutionarytheory of explanation
[1983] representsa similar approach,but not
restrictedto why-questions.Achinsteinfocuses on the act of explanation,notjust theproductof
such an act. An act of explanationis a response, so that an explanationmust be evaluatedwith
respect to the utteranceit respondsto.
 MathematicalExplanationand the Theoryof Why-Questions 605

then give two argumentsagainst van Fraassen'stheory of explanation:(1) the

methods he proposesto evaluate explanationsare trivializedin the context of
mathematicalexplanations;and (2) the structureof his theorymakes it difficult
to see how a proof could count as an explanation, though mathematical
explanations often do take that form. Finally, I will argue that, though a
strongertheory than van Fraassen'smight be able to avoid those two points,
no why-questiontheory can account for the way the explanationin the case
study is evaluated.

2 Exposition of van Fraassen's theory

Van Fraassen'stheoryof why-questionsis foundin Chapter5 of TheScientific
Image [1980].3 For him, why-questions consist of three items: a topic, a
contrastclass, and a relevance relation.When one asks, 'Why P?', the topic
is P. Thatis, the topic is whatwe ordinarilywould understandthe subjectof the
requestfor an explanationto be. The contrastclass, X, is a set of propositions:
{P1,P2, Pn}. The topic itself is some Pkfromthis class and all otherPi's are
.... to P. Thatis, when one wants to know why P is true,one asks why
alternatives
P is true ratherthan some otherpossibility. The contrastclass specifies the set
of possibilitiesthe questioneris willing to consider(includingP itself). Finally,
the relevance relation,R, is used to constrainadmissibleanswers,by specify-
ing what factors will count as explanatorilyrelevant. For instance, in many
situationsit would be inappropriateto answer 'Why does the blood circulate
throughthe body?' in terms of the heart's mechanical pumping action. This
question is often aimed instead at the function the circulationof the blood
serves in keeping an organismalive. A moreappropriateanswerto the question
in this sense is that the blood circulatesto provide nourishmentto the tissues.
The relevance relation is meant to distinguish between different senses of
such questions. In van Fraassen's words, the relevance relation encodes the
'respect-in-whicha reasonis requested'(p. 142). The only formalconstrainton
the relevance relation is that it obtains between proposedanswers and topic/
contrast-classpairs.
Van Fraassen'stheoryprovidestwo types of evaluationof answers.First, a
proposedexplanationmay or may not answerthe questionat all. For instance,
if the question 'Why does the blood circulatethroughthe body?' were intended
to invoke a functionalanswer,a purely mechanicalone would not answerthe

3 Van Fraassen'sgeneral approachis inspiredby Belnap and Steel's TheLogic of Questionsand

Answers [1976], though he introducessome additionalinnovations.Belnap and Steel proceed
from the dictum: 'Knowingwhat counts as an answeris equivalentto knowingthe question' (p.
35, attributedto Hamblin [1958]). However, as Bromberger [1962] points out, somebody
requestingan explanationmay not even be able to state one adequateanswer. Van Fraassen's
approachseems to turnthis dictumaround,by understandingthe answers(explanations)in terms
of the questionsthey are answersto.
606 David Sandborg

questionin its original sense. Second, even if a proposedexplanationanswers

the question in this sense, the answer may not be 'telling'. The proposed
answer that the circulationof the blood functionsto cool the tissues provides
the rightkindof informationto answerthe questionas intended,but is a poorer
explanationon factualgrounds.The theoryof whatcountsas an answerat all is
developedin the spiritof BelnapandSteel, butthe theoryof telling answershas
more in common with traditionalstudies of scientific explanation.
Answers are distinguishedfrom non-answersby the definitionof a direct
answer:
B is a directanswerto a questionQ = < Pk,X, R>exactlyif thereis some
propositionA such thatA bearsrelationR to <Pk, X> and B is the
proposition whichis trueexactlyif (Pk;and for all i7ik, notPi; andA)
is true(p. 144).
This definition also determineswhat, in Belnap and Steel's sense, the why-
question presupposes-what every direct answer to the question must affirm
as true. A why-question presupposes that its topic is true; that the other
members of the contrast-classare not true; and that at least one proposition
bearingthe relevance relationto the topic and contrast-classis true.If a why-
question has a false presupposition (for instance, if Pk is false), the best
response is not a direct answer, for no direct answer could be true. Instead,
the appropriateresponseis a correctiveanswer(such as 'Pkis not true!'). Van
Fraassentakes this accountof presuppositionto solve what he considersone of
the majorproblemsfor theories of explanation-accounting for rejectionsof
explanationrequests.
However, even if the presuppositionof a why-questionis true, there may
be no appropriateanswers,because none is telling. For instance, in the time-
honouredsyphilis-paresis example, thereis no telling answerwhy the mayor
of the town developed paresis, given that he had latent untreatedsyphilis.
There is nothing (so far as we can determine) that favours his developing
paresis among this contrastclass (p. 128).4 Van Fraassenneeds an evaluative
component for answers beyond the relevance relation, because even if an
answer gives the sort of informationthe questionerhas in mind (i.e. satisfies
the relevancerelation),and all of that informationis true,it may still not have
any bearingon the topic with respect to the rest of the contrastclass. As with
other forms of question, one wishes not to give false answers, but for why-
questions more is required. Van Fraassen introduces the theory of telling
answersto specify what more is needed. His theory containsthree elements.
First,a telling answermust be probablein light of our backgroundknowledge.
4 Van Fraassenstateson p. 146 thathe feels thatthe problemof rejectionsof explanationrequests
is solved without recourse to the theory of telling answers, but the syphilis-paresis example
indicatesthat some questionscan be rejected,not because they have no admissibleanswers,but
because they have no telling answers.
 MathematicalExplanationand the Theoryof Why-Questions 607

Second, it mustprobabilisticallyfavourthe topic over the othermembersof the

contrastclass relativeto backgroundknowledge.Third,it mustbe comparatively
betterin these regardsthanotherpotentialanswers.
The first of these criteria is straightforward,but the other two lead to
complications which van Fraassen took some pains to deal with. For our
purposes,the most importantproblemis that when one asks a why-question,
one generally knows that the topic is true, and other membersof the contrast
class false (indeed,the questionpresupposesthis). Therefore,in the light of our
full backgroundknowledge there is no way that any answercould favour the
topic. Forthe sake of evaluation,the topic mustbe probabilisticallyfavouredin
the light of some restrictedpartof the backgroundknowledge, K(Q). This in
addition to the topic, contrastclass, and relevance relation, is a part of the
context of the question.Van Fraassensaid nothingabouthow this subsetof the
background knowledge is selected, merely noting that the problem is not
unique to his theory.5
Van Fraassen mentioned several other problems concerning comparative
evaluationof answers.I mentionthem here brieflyfor completeness,but they
will not be importantin what follows. First, determining when an answer
favours the topic can be a subtle matter. Overly simple criteria can fail to
properlydelineateprobabilisticrelevance and some cases of favouringcan be
counterintuitive.6Second, there are additionalcomplicationsin evaluatingan
answerwith respectto otherpossible answers.Besides being betterat the first
two criteriathanotherdirect answersto a question,a telling answeralso must
not be screened off, or rendered probabilisticallyirrelevant, by any other
answer.
Because of these problems,van Fraassendid not put as much confidencein
the details of his theory of telling answersas he did in his characterizationof
direct answers, so we should not ultimately hold him to this theory. I shall
arguebelow thatit cannotapplyto mathematicalexplanation.But even further,
I will argue against van Fraassen'sentire pictureof the relationshipbetween
why-questionsandexplanatoryevaluations,andthatany theoryalong the same
lines will face serious challenges. Thus, I will spell out what I take to be the
philosophicalcore of his theory,strippedof all technicaldetails. Whateverthe
particulartheoreticaldetails, a theorywithin the why-questionapproachmust
follow these lines.

5 Salmon ([1989], p. 145) regardsthe problemof determiningthe appropriateK(Q) as equivalent

to the problemsleading to Hempel's requirementof maximal specificity and his own objective
homogeneity.
6 For instance, Hannson's initially plausible criterion-the probabilityof the topic given the
answer is higher than the average probability of members of the contrast class given the
answer-falls prey to simple counterexamples.(van Fraassen [1980], p. 128). On the other
hand,one mightthinkat least thatany answerthatdecreasedthe probabilityof the topic could not
favour it. However, if an answer decreases the probabilityof its likeliest competitorsin the
contrastclass even more, this may count as favouring(p. 148).
608 David Sandborg

As van Fraassenpoints out, a why-questionstated in English can often be

understoodin many different ways. The question 'Why did Adam eat the
apple?' seems perfectly straightforward,until we consider it with different
words emphasized: 'why did Adam eat the apple?' or 'Why did Adam eat
the apple?' or 'Why did Adam eat the apple?. (In van Fraassen's theory,
these readings of the question differ in their contrastclasses.) Each requests
a different sort of answer. The first requires that we say something about
Adam among all people that could have eaten the apple. The second
requires that we say something about the possible actions Adam could
have taken with the apple. The third requiresthat we say something about
the different things Adam could have eaten. If we say that Adam ate the
apple because he was hungry,we will not have successfully answered 'Why
did Adam eat the apple?' because he could have eaten something else and
satisfied his hunger.
Thus, if explanationsareregardedas answersto why-questions,they cannot
just be evaluatedwith respectto the ordinaryEnglish statementof a question,
which will often be ambiguous.Van Fraassen'slogic of why-questionsis an
attempt to specify unambiguously what a why-question evaluation will
involve. The natural-languagewhy-question is then identified with one of
these logical why-questions.This is where context steps in. It is the context
in which the natural-languagewhy-question is asked which is supposed to
determinewhich logical why-questionis being considered.In turn,this affects
how an answer will be evaluated. The two evaluative components of van
Fraassen'stheory,the relevancerelationandthe theoryof telling answers,both
require that contextual mattersbe specified before they can be applied; the
relevance relation depends directly on the context in which the question is
asked, and the telling-answerstheory depends on the contrastclass, and also
requires that K(Q) be specified. Therefore, for van Fraassen, explanatory
evaluationsare necessarily context-dependent.
This, thenis the role of contextin the why-questionapproachto explanation:
A why-questionis asked in naturallanguage. The context of utterancedeter-
mines which logical why-questionis intended.Explanationsarethenevaluated
with respectto this logical why-question.Contextplays a role in the evaluation
of explanations, but indirectly, through the specification of a logical why-
question. We can ask why the context could not figure more directly in
explanatoryevaluations,instead of throughthe mediation of why-questions.
My example will challenge the why-question approach on exactly this
point: an explanationis given that is judged explanatorythough it is a poor
answer to the logical why-questionthat provokedit. It is a better answer to
anotherlogical why-question,but that question cannot be formulatedin the
original context. In contexts in which such a question can be formulated,
however, the same explanationis not as effective. Thus, there appearsto be
 MathematicalExplanationand the Theoryof Why-Questions 609

something fundamentallywrong with the why-question theory's picture of

context-dependentevaluation.

3 An example of mathematical explanation

Let us now examine a case of mathematicalexplanation.Supplementingthose
already given by other authors,it gives some reason to believe that there is
something legitimately called 'explanation'in mathematics.After presenting
the example, I will consider some of the why-questionsthat this explanation
might be takento address.Thoughthis analysiswill be somewhatuseful, it will
turnout that van Fraassen'stheory fails to adequatelyaccount for mathema-
tical explanationin general,andthe explanationhere is not best regardedas the
answer to any why-questionthat could have precededit.
On p. 147 of Patternsof Plausible Inference([1968], Vol.2),7 GeorgePolya
gave an example of a proof which he findsprofoundlyunsatisfactory,thoughit
is well within accepted standards of rigour. After giving a presentation
intendedto reflect how the proof would be presentedin a textbook or journal
article,8he supplementedthe proof with additionalexplanatorymaterial.This
additionalreasoning, though unnecessaryfor proving the theorem,enhances
the proof's explanatoryclarity. Since the reasoning is somewhat technical
(thoughnot beyond the graspof a good advancedcalculus student),I will only
presenta sketch of these argumentshere.9
The theoremto be proved is: if the termsof the sequence a,, a2, a3, ... are
non-negativereal numbers,not all equal to 0, then

<e an.
>(ala2a3...an)1/n
n=1 n=1

The proof begins with the definition of an auxiliary sequence, c1, c2,
c3 .... by
theformulacl, c2, c3 .. c,, = (n + 1)n. Withthehelpof thissequence,
the theorem follows straightforwardlyby an unremarkableseries of inequal-
ities. The ci sequence is essential to the proof, but when it is first chosen, it
seems quite arbitrary.It is not clear, even afterone has studiedthe proof, why
an auxiliary sequence is introduced in the first place, much less why this
particular sequence is chosen. Polya called this a deus ex machina step,
since it seems to come out of nowhere, but helps complete the proof once
made. Because the readeris liable to be puzzled by the deus ex machina step,
Polya consideredthis presentationof the proof to be insufficient,thoughit is
7 This example also appearedin substantiallythe same form in his [1949], 'With, or Without,
Motivation?'
8 Substantiallythe same proof appearsin a 1924 articleby Polya in the Proceedings of the London
MathematicalSociety. The theorem also appearsas Theorem 334 in Hardy, Littlewood, and
Polya's [1934] Inequalities(with some, but not all, of the additionalexplanation).
9 The complete details can be found in Polya [1968] or Sandborg[1997].
610 David Sandborg

adequatelyrigorous.He suggested that the authorof the proof should inform

the readerwhy the step was taken,beyond the simple reasonthat 'it makes the
proof work'. He did the extra work in Patterns of Plausible Inference, and
claimed that these additions to the original proof make the derivationmore
understandable(p. 152).
In his explanation, Polya first gave some motivation for the theorem
itself-it is a lemma used to prove anothertheorem:if the series withpositive
terms converges, then the series
7n=lan (ala2a3 an1/n also con-
.

En=l on the left-hand side of

verges. Polya noted that the term (ala2a3...an)1/n
the equation is rather complicated. It is, however, also the left-hand side
of the well-known inequality between geometric and arithmetic means:
(ala2a3...an)1/'n Since the right-hand side of this inequality
a-a2+a3+..+an.
is simpler,it makes sense for Polya to base his strategyon this fact. However,if
applied naively, it leads to immediatetrouble:

1 +a2 +a3 + + an
n/n
Z(aia2a3...an)l"
n=l n=l 1

= ak
k=1 n=k n
The second step is a very naturalway to collect terms, but since the inner
sum is divergent,we have provednothing.Polya noted thatthe problemis that
since the series E l1 an converges by hypothesis,earliertermswill tend to be
much largerthanlaterterms(the sequenceal, a2, a3,... an, ... must eventually
decrease at least as fast as 1/n). Therefore, the two sides of the inequality
between the geometric and arithmetic means will tend to be somewhat
unequal.Polya recommendedtrying to balance the two sides by making the
termsin the inequalitymoreequal. One way to do this is to multiplyai by some
increasingfactor. This is the core of the idea behind the ci sequence.
However it is not yet clear what sequence one should use. Polya did not
begin by considering all possible auxiliary sequences. He first suggested
replacing the terms al, a2, a3, ... an, ... with lal, 2a2, 3a3, ... na, ..., but
quickly proposed a more general approach: lXaI,2Xa2,3Xa3... nXan... By
consideringthe more general case, he hoped to find a value for X that is most
favourablefor completingthe proof. His initial attemptleads quickly to a term
that cannot be calculated, but he was able to proceed by approximation,
eventuallywinding up with the term eXX-1 ak. If not for the approxima-
'=1 Even so, he was at least in a
tions, this would be very close to the desiredresult.
position to choose a value for X. Presumably,it is best to minimize eXX-', by
choosing X = 1. Polya's original firstguess of auxiliarysequence, ci = i, was
fairly good, but it now is more than a mere guess. Among the sequences
considered, it is most favourablefor Polya's purposes. However, even with
 MathematicalExplanationand the Theoryof Why-Questions 611

X = 1, he still came to a termhe could not calculate,and so could not complete

the proof.
At this point, Polya made an importantobservation,thoughhe was not very
explicit aboutits significance.He does not need to choose the ci sequenceto be
1, 2, 3, ... in orderto realize the advantageof choosing X.He could equallywell
use any sequence which is close to 1, 2, 3, ... in the limit (or any constant
multipleof such a sequence).Thus, any sequenceasymptoticallyequivalentto
1, 2, 3, ... will be as useful for balancingthe inequality,but some otherchoice
may lead to easier calculations.
To get an idea of what sort of sequence might aid in the calculation,Polya
next considered the completely general case, where ci is unrestricted.He
proceeded through a sequence of steps similar to that he used before, and
wound up with the term
1
>E n(CIC2C3...
Cn)ln
To proceed, he had to choose a ci sequence asymptoticallyproportionalto n,
andwhich also allows him to simplify this term.To do so, he resortedto a trick.
He noted that

1 n 1 n+1
n(n + 1)

and

Y1 n+ 1
n=-k
If CIC2C3...C, = (n + 1)n, the problematic sum takes this form, and he could
continuewith the proof.Indeed,the sequence so definedallows him to proceed
past the point at which he had been blocked, andcomplete the proof. Polya has
reconstructedthe original proof, but it is now easier to understandwhy the ci
sequence appeared.

4 Why-question analysis of Polya's explanation

How well does the theory of why-questions account for this example?
Certainly,the theoremandits initialproof seem to requiresome kindof further
explanation.This need is naturallyexpressed in terms of why-questions,and
the explanationsuperficiallyappearsto addressthese questions.Thus, a why-
question analysis initially appearsto be a promising way to understandthis
case.
Polya's dissatisfactionwith his originalproof can be expressednaturallyas
a why-question: 'Why is it appropriateto introduce the ci sequence in the
612 David Sandborg

proof?' This questioncan be readin two ways: (1) 'Why shoulda sequencebe
introduced into the proof?' and (2) 'Why, of all sequences, should this
particularone be chosen?' These two readingsof the questionaredistinguished
by differentcontrastclasses. Thus the situationthatpromptsPolya's explana-
tion seems quite amenable to a why-question analysis along van Fraassen's
lines.
Polya's explanationis meant to account for the ci sequence, so we would
expect that he answers one or both of the readings of the why-question. In
answer to (1) we might say that an auxiliary sequence was used in order to
replace a divergentseries by a convergentone. In answerto (2) we might say
that the particularsequence used was chosen for two reasons:it had a favour-
able growth behaviour, and it allows us to simplify a crucial term in the
derivation.Primafacie, we might say then that Polya's additionalexposition
is a successfulexplanationbecauseit providesa good answerto (1) and(2). We
will see below that the situationis not so simple. Though this why-question
analysis is useful in pointing out how Polya's explanation performs two
distinct functions, it does not correctly account for what makes Polya's
explanationgood. But, before considering the case study in more detail, let
us see how van Fraassen'stheory might handle mathematicalexplanationsin
general.

5 Van Fraassen's theory and mathematical explanation

In the following sections I discuss two kinds of difficulty with the why-
questionapproach.First,I discuss van Fraassen's theoryin the generalcontext
of mathematicalexplanation.I show thathis theoryof explanatoryevaluation
cannot account for mathematicalexplanations. In the next section, I look
at the Polya explanationin more detail from the general perspective of the
why-questionapproach.This will lead to a more sensitive diagnosis of why-
questiontheoriesof explanatoryevaluation,andpoint to a potentialalternative
approachto context-dependentexplanatoryevaluationsthatavoidsthe problems
uncoveredby attentionto mathematicalexplanation.
In this section I present two argumentsagainst van Fraassen's theory of
explanatoryevaluations.First,it cannotaccountfor mathematicalanswersto a
common type of why-question,in which the membersof the contrastclass are
mutually exclusive. Second, it cannot account for explanatorymathematical
proofs,becauseproofs arenot the sortof thing it recognizes as an answer,even
thoughproofsarecommonlyofferedas explanationsin mathematics(see Steiner
[1978] for numerousexamples of explanatoryand non-explanatoryproofs).
Though these argumentsare directedat van Fraassen'stheory of explanatory
evaluationin particular,it will become clear that any theory drawnup along
similarlines will encountersuch difficulties, as I shall show below.
 MathematicalExplanationand the Theoryof Why-Questions 613

Consider a why-questionfor which the contrastclass consists of mutually

exclusive members,such as 'why does 1 - + - +... convergeto , rather
thansome otherreal number?'Let us call such a questionan exclusive-contrast
question.10Under van Fraassen'stheory, any proof of the topic at all will be
a completely telling answer to such a question. The answer itself will be
judged as maximally probable, as it follows from accepted mathematical
propositions." The topic itself, having been proven, will have probability1.
All othermembersof the contrastclass, being incompatiblewith the topic, will
have probability0. No other answer can be more probable,favour the topic
better, or screen it off. Therefore, van Fraassen's theory of evaluation of
answers trivially recognizes any proof that establishes the truthof the topic
of the question as completely telling. Thus, at least for exclusive-contrast
questions,a proof must eitherbe explanatoryor not; thereis no middle ground.
But surelyin mathematicalcases, as in scientificones, some explanationsseem
betterthan others.If mathematicalexplanationscan be judged on more than a
binaryscale, the theoryof telling answersmustbe revised. Clearly,it could no
longer rely on the probabilitycalculus, which favours all true mathematical
propositionsequally. The Polya case suggests that such a revision might be
appropriate,because the judgements made there seem to form a continuum.
Polya's explanationdoes not appearto uniquely favour the ci sequence (for
reasons spelled out in the next section). A (hypothetical) explanation that
uniquely picked out the ci sequence would seem to be a more telling answer
to the why-question.
More importantlyfor our purposes,only the relevance relation can distin-
guish explanatoryfrom non-explanatoryanswers to exclusive-contrastwhy-
questions, since it is the only evaluative componentof van Fraassen'stheory
not left trivialized.This resemblesa point made in Kitcherand Salmon [1987].
They arguedthat for any pair of truepropositions,there is a way to 'explain'
one by the other, as judged by van Fraassen's theory. Their argumentalso
revolved aroundtrivializingthe theoryof telling answers,because the answers
they constructedwere maximallytelling-answers which, in conjunctionwith
10
Note that a contrast class need not in general consist of mutually exclusive members. For
instance, when we ask why the mayor has paresis, in contrastto the other members of the
countryclub, we accept thatmore thanone memberof the countryclub could have had paresis.
Indeed, if we were to ask 'why does 1 -?+1-4+ ..., ratherthan some other sequence,
converge to 1?', we would have to accept thatothersequencescould also convergeto the same
number.Moreover,in the Polya case, the question 'Why does this sequence serve to complete
the proof?' is not an exclusive-contrastquestion, since other sequences might serve as well as
the one actually used. Lipton [1991] discusses this issue further.However, exclusive-contrast
questions form an importantsubset of why-questions.Many of van Fraassen'sexamples use
exclusive contrastclasses (i.e. why did the sampleburngreen,as opposedto some othercolour);
and Garfinkeltoo considersexplanatorycontraststo be mutuallyexclusive.
"
Technically,it makes little sense to assign a probabilityto a proof. However,it is clearly in the
spiritof van Fraassen'stheoryto considerproofsas maximallyacceptableon this criterion-we
have completeconfidencein the proof.I will returnbelow to the questionof how well proofscan
be regardedas answersundervan Fraassen'stheory.
614 David Sandborg

an appropriatelyselected subsetof the backgroundknowledge,imply the topic

of the question and the negations of all other members of the contrastclass.
They were similarly able to trivialize the relevance relation. Because van
Fraassen put no formal constraintson the relevance relation other than its
relata, they had great leeway in constructing 'pathological' relevance
relations. Between any fact to be 'explained' and any true statement, a
relevance relationcan be found such that the statementis a maximally telling
answer to a why-questionwith that fact as its topic.
As an informal example, Kitcher and Salmon claim that van Fraassen's
theoryallows for an astrologicalexplanationof JohnF. Kennedy'sdeathon 22
November 1963. The topic is JFK's dying on thatparticulardate, the contrast
class consists of propositionsstating that JFK died on each particularday of
1963, or survivedthe year, and the relevance relationallows only answersin
termsof astralinfluenceto be consideredas explanations.This questioncan be
naturallyphrasedas: 'Why, in terms of astralinfluences, did JFK die on 22
November 1963, ratherthan some other date?' The answer that Kitcher and
Salmon proposeconsists in a truedescriptionof the configurationof starsand
planets at the time of JFK's birth, which, in conjunctionwith an appropriate
astrologicaltheory,would imply thatJFKdied when he did. This information
presumablywould answer the question as asked, and in a maximally telling
way. But surely the best answer to the question is a corrective one ('Astral
influences had nothing to do with JFK's death'). This example resembles
mathematicalexplanations in the way that it trivializes the telling-answers
consideration.Mathematicalexplanationsshow that the difficulty arises in a
large class of naturalcases, notjust for contrivedexamples.We cannotdismiss
a non-explanatoryproof as 'badmaths', as we might dismiss astrologyas 'bad
science'. The Kitcher-Salmon trivializationis a serious problemindeed.
The moral Kitcher and Salmon drew is that van Fraassen's theory needs
an additionalcondition on why-questions,namely that the relevance relation
used is a genuine relevance relation (ratherthan something gerrymandered).
However, thereare severalreasonswhy this does not seem to be an appropriate
solution. Van Fraassen'sdiscussion of explanatoryasymmetries([1980], pp.
130ff.) indicatesthathe favoursa somewhatunrestrictedrelevancerelation.In
situations where the traditionalasymmetriesof explanation are reversed, a
somewhatatypicalrelevancerelationmay be called for. I suspectthathe would
also resist such restrictionson the relevance relationbecause they would not
be logical constraints.Whetherastralinfluence is relevantto people's deaths
is not a logical matter,but a fact about our universe, and doesn't belong in
a theory of the logic of why-questions.Also, the reasons we do not consider
some proofsto be explanatorydon'tseem to haveto do with relevanceat all, or at
least not relevance that the questioner can specify in advance. An
earlyunintelligibleproofof a result(suchas one using 'bruteforce' calculational
 MathematicalExplanationand the Theoryof Why-Questions 615

techniques)may be obviously relevantto that result, even if considerednon-

explanatory,while anexplanatoryproofemployingmoreabstractmathematical
resourcesmay notbe so obviouslyrelevant;it maynot be immediatelyclearthat
the resources used are appropriateto the problem.Indeed, discovering these
resourcesmay be a large step towardssolving the problemin the firstplace. No
relevancerelationspecifiableat the outsetwouldbe able to distinguishbetween
explanatoryandnon-explanatoryproofs,because whatis relevantto the expla-
nandumis not known prior to the explanationitself. This argumentwill be
furtherdevelopedin the next section,wherea new way to view explanationswill
be proposedthatwill not be subjectto such trivialization.
Thus, van Fraassen's theory seems to have inadequateresources to avoid
trivializationin mathematicalcases. Neither the theory of telling answersnor
the relevance relation seems to do the trick; the theory of telling answers
recognizesany mathematicalproof as equally good in a wide rangeof realistic
cases, but a priori considerationsof relevancedo not seem to correctlydistin-
guish between explanatoryand non-explanatoryproofs. Thus, mathematical
explanationscall van Fraassen'stheoryof explanatoryevaluationinto question.
As it stands, there is anotherreason why van Fraassen's theory would be
difficult to apply to mathematical cases. He regards an explanation as a
presentationof descriptive informationrelevant to the topic: '[I]f you ask a
scientist to explain something to you, the informationhe gives you is not
differentin kind (anddoes not soundor look different)fromthe informationhe
gives you when you ask for a description'([1980], p. 155). This attitudeis also
implicit in the theoryof telling answers.An answeris evaluatedby considering
whatcontributionit makesto favouringthe topic when it is used to supplement
some subset of our backgroundknowledge, K(Q).12Thus, the answerpresents
a piece of crucial descriptiveinformationthat makes the topic more credible
than it would otherwise have been.
Supposewe attemptto extendthis theoryto mathematicalcases. The natural
thing to take as the backgroundknowledge K in a mathematicalcontext is
either the set of axioms of some mathematicaltheory, or their deductive
closure. The topic of the explanation (a theorem, for instance) will then
be either implied by or containedin K. Therefore,as with empiricalexplana-
tions, K mustbe furtherrestrictedto some suitableK(Q). The naturalway to do
so is to consider a smaller set of axioms, or its deductive closure. Then one
would expect an answerto be some propositionwhich, in conjunctionwith this
K(Q), implied the topic of the why-question.This could either be a missing
axiom, or some weaker statementthat still implies the theorem in question.
12 Note that the limitation of the
backgroundknowledge to K(Q) for the sake of explanatory
evaluation(p. 147) is an attemptto avoid a trivializationlike that noted above. In general, the
explanandumand the negation of all other members of the contrastclass will be completely
telling information.The above trivializationargumentsuggeststhatthereis a morefundamental
problemthat the invocationof K(Q) has failed to fix.
616 David Sandborg

Van Fraassen's theory thus suggests that mathematicalexplanationswould

have to be analysesof the preconditionsrequiredfor a theoremto be true,such
as showing that a theoremdependedon the axiom of choice.
Van Fraassenis not alone in regardinganswersto why-questionsin this way.
Thoughthe model of why-questionspresentedby HintikkaandHalonen[1995]
is significantlydifferentfromvan Fraassen's, it also turnsout thattheyimplicitly
regardexplanationsas providingadditionallogical information.This is clear
when examiningthe assumptionsbehindtheirmodel; thatP(b) can be derived
from (T&A),but not T alone, where T is an 'initialpremise',A the answer,and
P(b) the explanandum.They consider the assumption that P(b) cannot be
derived from T to rule out one possible kind of trivialization(p. 648), but it
appearsthatmany explanatoryproofs are 'trivial'in exactly thatway; they are
alternateways to derive a theoremfrom the same set of premises.Undertheir
model, as under van Fraassen's, mathematicalexplanations would take the
form of analyses of axioms requiredfor theorems.
Though such analysis certainlygoes on in-and is an importantpartof-
mathematics,13 it is neithera clearlyexplanatoryactivity,nordoes ourexample
(or manyotherswhich could be mentioned)takethis form at all. Polya does not
single out any propositionthat, in conjunctionwith some other part of back-
ground mathematics,leads to the result in question. Further,van Fraassen's
pictureof explanationmakesit mysteriouswhy one shouldever offer a proof as
an explanation.14A proof does not fill in any missing information,but instead
drawsout consequencesfrom previouslygiven propositions.Ideally,a proof is
taken to add no informationat all. Yet some proofs are said to explain their
conclusions.
Thus it seems that van Fraassen'stheoryof explanatoryevaluationwill not
easily be extended to mathematicalcases. Indeed, suitably modified they
highlight the difficulty of extending many theories of explanationto mathe-
maticalexplanations.For any such extension will have to recognize that some
proofs explain their conclusions, while others don't. But any theory that uses
probabilistictools to evaluate explanations is liable to regard all proofs as
explanatory,for proofs will tend to attainmaximal scores in such evaluations.
On the otherhand, any theorythattakes explanationsto increaseour proposi-
tional knowledge will tend not to regardany proofs as explanatory,since they
do not do this, but at best display the consequences of what we have already
accepted as given.
13 Such analyses fall underthe
14
heading of 'reversemathematics'.
I do not wish to claim thatall mathematicalexplanationsareproofs. Indeed,the exampleshows
thatmathematicalexplanationis not the same thingas proof.Proofis neithernecessary(Polya's
originalproof is not explanatory),nor sufficient (Polya's explanationis not itself a proof) for
explanation.None the less, proofs are often vehicles for mathematicalexplanation.For the
purposes of my argument,it is enough to observe that some answers to mathematicalwhy-
questions are proofs, and van Fraassen'stheorycannot account for answersof that form.
 MathematicalExplanationand the Theoryof Why-Questions 617

6 The why-question approach

Rather than pursue the above reasoning further,I now turn to an argument
against any theory, not just van Fraassen'sformulation,that reconstructsthe
idea of 'explanatoryrelativity'in termsof why-questions.The problemis that
some context-dependentfactors that go into our evaluation of explanations
cannot be understoodin terms of why-questions. Though the why-question
approachis rightin recognizingthe importanceof context in such evaluations,
the correctaccountof the context-dependentevaluationsin the Polya example
cannot properly be called a why-question account. The resolution to this
argumentwill also point the way to an account of explanation that is not
subjectto the objections in the previous section.
My argumentresemblesone given by Paul Humphreys([1989], pp. 137-8).
I will show that the Polya explanation representsa concrete example of a
problemHumphreysraised abstractly.The mathematicalexample more shar-
ply delineates the problem, and demonstratesthat it is pervasive to the why-
question approach,extending far beyond the scope of the singular causal
explanationsto which Humphreysrestrictshis attention.
Consider again the situation after Polya has presented his initial proof.
Though we agree it is successful as a proof, certain aspects of the proof
remain puzzling. Polya's explanation largely dissolves this puzzlement; it
appearsto be a largely satisfactoryexplanation.Accordingto the why-question
theory,this explanatorysuccess must be due to Polya's having answeredsome
why-question.Indeed, our initial puzzlement seems to be well expressed in
terms of a why-question, namely 'Why is it appropriateto introducethe ci
sequence in the proof?' Therefore,it seems naturalto suppose that Polya's
explanationwould satisfactorilyanswerthis question.
As we saw above, this why-questioncan be given two differentreadings.
Polya seems to addressboth readingsin his explanation.In responseto reading
(1), he observesthatan auxiliarysequencesharpensa crucialinequalityused in
the proof. The observationmay be a bit vague, but it is not unsatisfactory.
However, this observationis only the beginning of Polya's explanation.The
rest appearsaimed at reading (2). Moreover,if he had simply stopped at this
point, the explanationwould not have been as satisfying.Therefore,it appears
that much of the virtue of Polya's explanationmust be in answeringreading
(2).
However, Polya's explanation,on close examination,is not a very success-
ful response to this reading of our initial why-question. It shows how the
growthrateof the ci sequencemakesit favourablefor completingthe proof,but
this is far from enough to show thatthis sequenceis the appropriateone to use.
There are many other sequences with the same growth rate, some seemingly
more natural.The c, sequence is chosen in part because it allows a crucial
618 David Sandborg

simplificationof the term

'

But this choice still feels like a 'trick',even afterPolya's expandedexposition.

En(IC2C3..Cn)l/n
Besides the raw fact thatthe chosen sequenceenablesthis simplification,there
is no reason for using that sequence instead of anotherwith similar growth
behaviour.We areshown thata particularci sequenceworks;we arenot shown
why. The explanationthus does not tellingly distinguishbetween the sequence
used in the proof and a large class of others,andcannotbe considereda telling
answerto the original why-question.15
Thus, we appearto have a good explanationthat doesn't provide a good
answer to its motivating why-question. This presents a serious problem for
the why-questionapproachbecause it seems to imply that the explanationis
evaluated on other grounds than having answered why-questions.But there
are several options open to the why-questionadvocate. First, the claim that
Polya does not satisfactorilyaddressthe (second reading of the) motivating
why-question can be questioned. Second, the argumentrelies on our intui-
tive judgement that Polya's explanation is actually satisfactory.Perhaps it
is really not as effective as we initially thought. Finally, we might claim
that the explanation is effective because it answers some other why-
question, not this particularone. Let us address these points in turn. The
last one will be particularlyimportant.It will turn out that Polya's explana-
tion indeed does answer a different why-question satisfactorily,but we will
also see that its virtue as an explanation is not primarilyin answering this
why-question.
The argument that Polya's exposition does not tellingly answer the
(second reading of the) initial why-question may be felt as a bit too
quick. Perhaps a more sophisticateduse of van Fraassen's telling answers
theory will allow us to reject its conclusion. Indeed, van Fraassen's theory
has some resources which one might think would allow him to claim this.
He allows an explanation to be judged as telling not only by how well it
favours the topic over the other members of the contrastclass, but also by
how well it fares in comparisonto other possible answers. One could then
claim that although Polya's explanation doesn't fully spell out why this
particular sequence helps complete the proof, no other answer would be
able to do any better. Though we can't be certain of this judgement without
considering all possible answers to the initial why-question, it is plausible.
We suspect that the ci sequence takes such a seemingly unusual form
15 HereI do not necessarilyassumethe particulartheoryof telling answersvan Fraassenhas given,
but presumewe can makeintuitivejudgementsabouttellingnessnot subjectto the trivialization
argumentgiven in the previous section.
 MathematicalExplanationand the Theoryof Why-Questions 619

because it allows one to simplify the term

1
n=k
n(c1C2C3...Cn)l/n
This feature of the sequence may not be amenable to any furtherprincipled
analysis;it would then be a sortof mathematicalcoincidence.ThoughPolya's
answerdoesn't seem too good on its own, it may be the best we could expect.
Even if this reasoningis correct,it simply does not make the explanationa
good answer. That there is no better alternativedoesn't add to the positive
qualities of the explanation,yet Polya's explanation strikes one as good by
what it accomplishes, not by comparisonto unspecifiedalternativeattempts.
The comparativeaspectof van Fraassen's theoryof telling answersis best used
to rule out seemingly good explanations that are dominatedby some other
explanation(such as probabilisticexplanationsthatare 'screenedoff' by other,
better explanations).If an explanationwe judge to be good is dominatedby
anotherone, van Fraassen'stheory will correctlylead us to reject it. It is less
clear that by the same principle we should reconsider an initially negative
judgement of explanatoryquality in light of the fact that there is no better
explanationavailable. That something is the best available answer does not
make it a good answer. To say otherwise is to resort to the force majeure
argumentthat van Fraassenhimself rejects in Laws and Symmetry([1989],
pp. 144-5). Therefore,it appearsthatwe must accept thatPolya's explanation
does not successfully answerits motivatingwhy-question.
The why-questionadvocatemightthen say thatthis shows Polya's exposition
is not a good explanationafterall. Thatseems jarringin light of our admittedly
pretheoreticalbut clear intuitionsabout the example; and more motivatedby
an a priori adherenceto the why-questiontheory thanby attentionto the case
study. Certainly,the explanationdoes manage to clear up a lot of confusion
about the ci sequence. Though one cannot be said to have received a fully
illuminatinganswerto the question, 'Whyis it appropriateto use this particular
sequencein this proof?' one is farless puzzled aboutthe sequencethanone was
beforehand.Furthermore,we are not confined to pretheoreticintuitivejudge-
ments; we can say at least something positive about what this explanation
contributes.A key partof the explanationseems to be the use of growthrates.
The concept of growthrate did not explicitly arise in the originalproof, but it
seems to be the best way to understandthe selection of the ci sequence.
Therefore, our initial puzzlement about the proof seems to stem from the
fact that it did not highlight an explanatorilycrucial propertyof the sequence,
andthe virtueof the explanationseems to be thatit does highlightthis property.
This is an importantobservation,which we will returnto below.
Given that Polya's explanationis a good one, but that it doesn't answerthe
motivating why-question,there appearsto be one other response available to
620 David Sandborg

the why-question theorist-to seek a different why-question that Polya's

explanation does answer. The focus on growth rates points us to such a
question: 'Why is it appropriateto introduce a sequence asymptotically
proportionalto 1, 2, 3, ... into the proof?' The contrast class has shifted.
Rather than sequences, it now contrasts different growth rates. Polya's
supplementaryexposition successfully shows why the asymptotic growth
rate selected is the best candidate to solve the problem. The augmented
exposition points to a unique growthrate.
Thus, the why-questionadvocatecan claim thatPolya's explanationis good
not because it answersthe initial why-question,but because it answersa new
question. It would appearthat this situationis a variationof rejection of the
why-question as described by van Fraassen.Instead of simply rejecting the
initial why-question with a corrective answer, Polya goes on to address a
somehow more appropriatewhy-question. In this case we actually have a
satisfactory explanation,even in a case where the initial why-question had
a false presupposition.
But for this analysis of the situation not to be ad hoc, the why-question
theory must be supplemented somehow. Once the initial why-question is
rejected, we cannot expect to answer just any other why-question and still
have a successful explanation.For one thing, this would put the why-question
approach in danger if a trivial why-question can be constructed for any
purported'explanation',as Kitcherand Salmon have done for van Fraassen's
particularformulation.Secondly, such a responseappearsto minimize the role
of context in the evaluation of explanations, since the apparent context,
specified by the initial why-question, would no longer be an importantpart
of the evaluation. If we are to account for why Polya's explanation was
effective in this situation,we must understandhow it respondsto our initial
state of puzzlement.It seems thereforethatthe why-questiontheorystill owes
us a storyof how Polya's explanationrelatesto the initialwhy-question(which
is presumablythe way the why-question theory would find that puzzlement
best expressed), even if it does not answer it.
But even if the why-questionapproachcan answerthis challenge,it will still
miss an importantway in which Polya's explanationresponds to our initial
state of puzzlement after we have first seen the original proof. If the above
responsewere correct,the problemwith the initialquestionwas thatit made an
incorrectassumption-that therewas a telling reasonto favourthe ci sequence
over some other with the same growth rate. The questionercould have and
should have asked a differentquestioninstead. Once the questioneris 'put on
the right track', so to speak, the explanation can proceed. However, to
formulatethis new question, van Fraassen'slogical apparatusdemandsthat
the contrast class consist of statements of the form: 'The growth rate of
sequence ci is the most appropriatefor an auxiliary sequence that will lead
 MathematicalExplanationand the Theoryof Why-Questions 621

to a successful proof.' If the concept of growth rate is not available to the

questionerin the first place, we cannot do this. The explanation,if it gets its
virtueby answeringa why-questionat all, does so by answeringa questionthat
couldn't even be asked priorto the explanation.
Suppose, however, that the questionercould formulatethe question 'Why
is it appropriateto introduce a sequence asymptotically proportional to
1, 2, 3, ... into the proof?' The ability to formulatethis question presupposes
some familiarity with mathematicalanalysis; at least enough familiarity to
understand(a) what a growth rate is, and (b) that it might be importantto
choosing the ci sequence. A person able to ask this question is less likely
to be puzzled in the first place by the ci sequence than someone more naive
in mathematics. The augmented exposition is therefore less likely to be
explanatorilyeffective for a person in a position to ask the question than for
one who cannot. Whether the concept of growth rate is available to the
questioneris a crucialpartof the context in which the need for an explanation
arises. But since the explanationanswersthe same why-questionregardlessof
whetherthe questionercould formulateit or not, the why-questionapproach
cannot distinguishbetween the sophisticatedand naive questioners.The why-
questioncorrectlyacknowledgesthe role of contextin evaluatingexplanations,
but misses an importantaspect of that context-the conceptualresourcesthe
questionerhas availableto analyse the situation.Much of what makes Polya's
exposition successful is thatit shows thatgrowthrateis important;not because
it picks out a particulargrowth rate rather than another. Our reactions to
Polya's explanationarenot best understoodas stemmingfromhow he answers
why-questions,butfromthe conceptualresourcesfor analysingthe situationhe
brings to our attention.
Since I mentioned the contrast class above, I might be accused of still
attacking only van Fraassen's theory ratherthan the general why-question
approach. However, the argument doesn't depend upon the details of his
theory at all, but only on the model of context-dependentevaluation that is
characteristicof the why-question approach.The key point is that a why-
question is taken to implicitly fix the way an answermust regardits topic. In
van Fraassen's theory,this is done by specifying a contrastclass andrelevance
relation, which must thereforebe antecedently specifiable in the context in
which the question is considered. But any why-question-basedevaluation
should at least be able to distinguish between asking about sequences and
asking about growth rates. It must then allow an answer in terms of growth
rates to be judged explanatoryby a questionerwho couldn't even talk about
growth rates until the explanation had been given.16 For the why-question

16 How any why-questiontheorycould do this is unclear,rememberingBelnapand Steel's dictum

that to understanda questionis to know what would count as an answer.
622 David Sandborg

theory,an explanationmustrespondto a question,which implies a fixed way of

looking at the topic. But our initial stateof puzzlementmay be due to not even
knowing how best to regardthe topic. An explanationcan gain most of its
virtue by responding to this state of affairs-showing us an effective way
to understandthe subject-matter-rather than through any particularwhy-
questionsit happensto answer.In so far as asking a why-questionfixes a way
of looking at the explanandumand demandsan explanationin those terms,the
why-questionapproachwill be subjectto this problem.But this seems to be the
very point of 'explanatoryrelativity', a central feature of the why-question
approach.Thus it is hardto see how the why-questionapproachcan address
this problemsuccessfully without being watereddown beyond recognition.
This problemis highlightedby attentionto mathematicalexplanations,but
is not restrictedto them. A physical explanationmay also introducea com-
pletely new way of looking at its topic. For instance,IsaacNewton providedan
explanationfor the movements of the planets,but not one which answersthe
question posed by his predecessors;they demandedmechanicalexplanations
withoutreferenceto actionat a distance.Newton's answerwouldhave failed to
resolve the question the Cartesianshad in mind. Indeed, it would have been
impossible to specify a Newtonian answer as an appropriateanswer to a
question posed before the Principia; the pertinentconcepts couldn't yet be
given to indicate that kind of answerwas appropriate.
These considerationssuggest an alternateway to look at theoreticalexpla-
nations than the why-question approach.An explanationmay be significant
because it deploys relevantconceptualresourcesnot previouslyavailable.This
still allows for context-dependentexplanatory evaluations, since what is
considered 'previously available' will depend upon the context in which the
need for explanationarises;but the role thatcontextplays is differentfromthat
recognized in the why-question approach.Even though only programmatic,
this pictureof explanation-evaluationhas a numberof virtues.In particular,it
avoids both problems with van Fraassen's theory of explanatoryevaluation
mentioned in the previous section. First, it avoids trivialization, because
perfect 'favouring'of the explanandum,as so often happensin mathematical
cases, is no longer sufficient for explanation.When an explanationdoes not
look at the explanandumin a new way, it need not be consideredexplanatory.
Second, our picturemakes clear why one might offer a proof of a result as an
explanationof it. Thougha proof may not give us access to new propositional
information(certainlynot if we presumelogical omniscience), it may invoke
new conceptualresourcesthat were not previously available.It is this, rather
than establishingpropositionalfacts, that makes such a proof explanatory.
The furtherstudyof mathematicalcases will be importantto developingthis
new approachto explanation.First,mathematicalcases must be consideredin
order to avoid the trivialization argumentsraised in the previous section.
 MathematicalExplanationand the Theoryof Why-Questions 623

Second, the role of new conceptual resources,ratherthan new propositional

knowledge,will be most clearlyseen in mathematicalexamples,wheretypically
no new propositionalknowledge is generated.I thereforeclaim that in order
to improve our understandingof explanation, we must seriously consider
mathematical,as well as empirical,explanations.
We can now respondto van Fraassen'sshort argument,'An explanationis
an answerto a why-question.So, a theory of explanationmust be a theory of
why-questions.'Even thoughan explanationmay be offered as an answerto a
why-question,it may not be in virtueof its being the answerto a why-question
thatit is explanatory.Thougha theoryof why-questionsmay aid the theoryof
explanation,the theory of explanationmust go beyond it.

Acknowledgements
I would like to thank Wesley Salmon, David Rudge, and the anonymous
reviewers for this journal for their helpful comments on earlier versions of
this paper. I would also like to thank KennethMandersfor his criticism and
guidance throughoutthis project.
Departmentof History and Philosophy of Science
Universityof Pittsburgh
Pittsburgh,PA 15260
USA

References
NewYork,OxfordUniversityPress.
Achinstein,P. [1983]:TheNatureof Explanation,
Belnap,N. andSteel, T. [1976]: TheLogic of QuestionsandAnswers,New Haven,Yale
UniversityPress.
Bromberger,S. [1962]:'AnApproachto Explanation',in R. J. Butler(ed.),Analytical
Philosophy,SecondSeries,Oxford,BasilBlackwell,pp.72-105.
Bromberger, in R. G. Colodny(ed.),Mindand Cosmos:
S. [1966]:'Why-questions',
Essays in contemporaryscience andphilosophy,Pittsburgh,Universityof Pittsburgh
Press,pp. 86-111.
Garfinkel,A. [1981]: Formsof Explanation:Rethinkingthe Questionsin Social Theory,
New Haven,YaleUniversityPress.
Hamblin,C. L. [1958]: 'Questions',AustralianJ. Phil., 36, pp. 159-68.
Hardy,G. H., Littlewood,J.E., andPolya,G. [1934]:Inequalities,Cambridge,Cambridge
UniversityPress.
Hempel, C. G. [1965]: 'Aspects of Scientific Explanation', in Aspects of Scientific
Explanationand OtherEssays in the Philosophy of Science, New York, Free Press,
pp.331-496.
for Why-Questions',
Hintikka,J. andHalonen,I. [1995]:'SemanticsandPragmatics
J. Phil., 92, pp. 636-57.
624 David Sandborg

Humphreys,P. [1989]: The Chancesof Explanation:Causal Explanationin the Social,

Medical, and Physical Sciences, Princeton,PrincetonUniversityPress.
Kitcher, P. [1983]: The Nature of Mathematical Knowledge, New York, Oxford
University Press.
Kitcher,P. [1989]: 'ExplanatoryUnificationand the CausalStructureof the World', in
Kitcherand Salmon [1989], pp. 410-505.
Kitcher, P. and Salmon, W. [1987]: 'Van Fraassen on Explanation', J. Phil., 84,
pp. 315-30.
Kitcher,P. and Salmon, W. (eds) [1989]: ScientificExplanation:MinnesotaStudiesin
the Philosophy of Science, Vol. XIII,Minneapolis,Universityof MinnesotaPress.
Lipton, P. [1991]: 'ContrastiveExplanationand Causal Triangulation',Phil. Sci., 58,
pp. 687-97
Polya, G. [1949]: 'With, or Without,Motivation?',AmericanMathematicalMonthly,
56, pp. 684-91.
Polya, G. [1968]: Mathematicsand Plausible Reasoning(2 vols), second printing(first
printing 1954). Princeton,PrincetonUniversityPress.
Resnik, M. D. and Kushner,D. [1987]: 'Explanation,Independence,and Realism in
Mathematics',Brit. J. Phil. Sci., 38, pp. 141-58.
Salmon, W. [1989]: 'FourDecades of Scientific Explanation',in Kitcherand Salmon
[1989], pp. 3-219.
Sandborg,D. [1997]: 'Explanationand MathematicalPractice', doctoraldissertation,
Universityof Pittsburgh.
Steiner,M. [1978]: 'MathematicalExplanation',Phil. Stud., 34, pp. 135-51.
Van Fraassen,B. [1980]: The ScientificImage. Oxford,ClarendonPress.
Van Fraassen,B. [1989]: Laws and Symmetry.Oxford,ClarendonPress.