A Desperate Fix
Author(s): Bob Hale
Source: Analysis, Vol. 55, No. 2 (Apr., 1995), pp. 74-81
Published by: Oxford University Press on behalf of The Analysis Committee
Stable URL: http://www.jstor.org/stable/3328903 .
Accessed: 31/01/2011 12:12
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 [email protected].
Oxford University Press and The Analysis Committee are collaborating with JSTOR to digitize, preserve and
extend access to Analysis.
http://www.jstor.org
A desperatefix
BOBHALE
In the words of its author,Gideon Rosen, modal fictionalismis 'an effort
to reconcile two thoughts' namely that 'talk of possible worlds in the
discussionof modalityis clearlyilluminating;and yet for many of us it is
also palpablya matterof make-believe'.The reconciliationis to be effected
by taking such talk as
talkaboutthe contentof a fiction.Wherethemodalrealistproposes
to analysea modalstatementP by meansof a non-modalstatement
aboutpossibleworlds,P*,thefictionalist
proposestheparasiticparaof
a
to
the
hypothesis pluralityof worlds(PW),
phrase:'According
P*'. ([5], pp. 67-68).
In my earlierpaperin this issue [2], I posed a simpledilemmafor this view,
and in formulatingit, availed myself of one of the paraphrasesof statements formed with the fictionalist'sprefix which Rosen provided ([4],
p. 344), namely: 'If PW were true, it would be true that P*'. For simplicity,
the dilemmafocused on the atheisticfictionalistwho wished to rejectPW
as false, and asked whetherPW was supposedto be necessarilyfalse, or
merelycontingentlyso. On the firsthorn,it appearsthat all such statements
will turn out vacuouslytrue. On the second,the problemis to see how the
fictionalistcan understandthe involvedclaim that PW might be true - he
cannotapplyhis preferredanalysis,sincethat would give us a statement('If
PW were true, therewould be a world at which PW is true')which would
be true, even if PW were necessarilyfalse; but if he invokes some other
accountof this modal claim which does not offend againsthis ontological
conscience,then it becomesobscurewhy that alternativeaccount should
not be appliedacrossthe board, with the resultthat fictionalismloses its
point. In his reply[5], Rosentakes on my dilemmaand, since he thinksthe
fictionalisthas no option but to regardthe hypothesisof possible worlds
as necessarilyfalse, seeksto bluntthe firsthorn.The principalobjectof this
note is to explainwhy I think he has made no impressionon it. As a coda,
I shall add some remarksexplainingwhy one attemptto blunt the second
horn, which I electednot to discussin my originalnote, comes to nothing.
It is - and was all along - obvious that the first (necessaryfalsehood)
horn might be bluntedin one of two ways:
(a) by denyingthat the truth of 'If PW were true, it would be true
that P*' is sufficientfor that of 'Accordingto PW,P*', and hence
rejectingthe formeras a satisfactoryparaphraseof the latter;
55.2, April 1995, pp. 74-81. ? Bob Hale
ANALYSIS
A DESPERATE FIX
75
(b) by rejectingany standardaccount of subjunctiveor counterfactual conditionals,underwhich those with impossibleantecedents
are vacuouslytrue.
Rosencanvassesboth strategies,and- althoughsuccessfulimplementation
of eitherseparatelywoulddo - apparently
endorsesboth.I shallfirsttry
to showthatstrategy(a)is hopeless,so thathe hasno optionbutto plump
for (b), andthenarguethathe oughtto be a good dealless enthusiastic
aboutthatoptionthanhe is.
Strategy(a)In [4], Rosenwrote:
Wemightbegin[i.e.to explainthe fictionalist
prefix]by offeringany
of thefollowingglosses:If PWweretruethenP wouldbe true;If we
supposePW,P follows;It wouldbeimpossible
forPWto betruewithout P beingtrueas well.Thesearenot perfectparaphrases.
Nonethe
to
a
fair
each
seems
indication
of
what
we mean
less,
give
preliminary
in every
whenwe use the fictionalist's
And
the
is
that
trouble
prefix.
casethe key phrasein an overtlymodallocution.Thissuggeststhat
the fictionalist's
deviceshoulditselfbe classedas a modaloperator.
([4],p. 344)
Thisadmittedly
of the suggestedglossesis followed
guardedendorsement
by overfourpagesof discussionof objectionswhichdependupontaking
them seriously.In [5], however,after airing some misgivings(to be
discussedbelow)aboutthe standardtreatmentof counterfactual
condiRosen
writes:
tionals,
Moreimportantly,
the explanationof the fictionalist's
in
story-prefix
termsof a counterfactual
was neverpartof the fictionalist's
official
view.Rather,it wasrejectedas inadequate
forpreciselythesereasons.
the prefixis primitive.([5],p. 70)
Officially,
It is hardnot to feelthatthegroundhasshiftedsomewhat,butI shallnot
pressthe point.Letus focus,instead,on the closingofficialcommunique.
It seemsclearthatit is simplynot an optionfor the fictionalistto declare
hisprefixa primitive,andleaveit at that.Evenif nothingrankingas a definitionin anystrictsenseis needed,someelucidation
is calledfor- without
it, thereis just no theoryto discuss.It was, presumably,
sensitivityto
preciselythispointwhichled Rosento proposehis variousglossesin the
firstplace.If he now thinkstheyplayhimfalse,heoughtto putsomething
moreto his likingin theirplace,particularly
as they are apt to strike
as
the
essential
idea.
competentEnglishspeakers capturing
We know two things about 'According to PW, ...' - (1) it has to be
non-factive, and (2) if my dilemma is to be avoided, it can't be para-
76
BOB HALE
phrased as a strong (strict or counterfactual)conditional or entailment
with 'PW is true' as hypothesis. For whilst such paraphrasespreservethe
requisitelack of factivity,theirotherlogicalpropertiesrenderthem unsuitable for Rosen'spurposes.What other non-factiveoperatorsmight better
serve?
The most obvious suggestion- given the backgroundthought that we
are dealingwith some kind of fictionaloperator- is to gloss our prefixas
'PW says that ...'. This has the requisite non-factivity; and while 'PW says
that A' presumablydoes entail 'Ifwhat PW says were true, it would be that
A', it is - at least on one naturalreadingof these sentences,false that the
latterentailsthe former.(Forexample,the necessarytruthof A sufficesfor
the truthof 'If what PWsaysweretrue,it would be that A', but not for that
of 'PW says that A'). The snag is, that cleavingto this readinggives something far too restrictivefor the fictionalist'spurposes.PW simply does not
say enough, in any sense of 'say' that is exacting enough to obstruct the
entailmentfrom 'If what PW says were true, it would be that A' to 'PW
says that A'. It can't do so, since there are potentially infinitelymany A
for which the fictionalist will want to substitute 'Accordingto PW,A',
while PW can explicitly (or near enough explicitly) state only a finite
selection of them. To overcomethis, the fictionalistwould be obliged to
adopt a less exiguous readingof 'say', on which 'PW says that A' holds
true not only when PW explicitly states that A, but equally when it
follows from things which PW explicitly states that A. But this takes us
straightback to the originaldifficulty.In short, any readingof the fictionalist'sprefixin termsof 'says that' will be too restrictive,unless he allows
that the truth of a strong conditional 'If PW were true, it would be true
that A' or an entailment'PW entails that A' sufficesfor the truth of 'PW
says that A' - in which case he is lost unless he endorses strategy (b)
anyway.
There is another,quicker,way to see that the fictionalistmust follow
strategy(b)- a way whichsimplybypassesanywranglingover what is/isn't
an apt paraphrase for 'According to PW, ...'. For the upshot of the imme-
diately precedingobservationsis that it is a second constraint(on top of
non-factivity)upon any acceptablereadingof this prefixthat whenever,if
PW were true, it would be true that A (or wheneverPW entails that A), it
must also be true that accordingto PW,A. If not, then the fictionalist's
savingmove will be unavailablein cases whereit is needed.But so long as
the strongconditionalis standard(or the entailmentclassical),this means
that the necessaryfalsehoodof PW will guaranteethe truthof everystatement of the form 'Accordingto PW,A'. In sum, strategy (a) is hopeless
unless backed by strategy (b); and redundant in the presence of (b), so a
waste of time at best.
A DESPERATE FIX
77
Strategy(b) In pursuitof strategy(b), Rosen writes:
... the feature of the standard semantics for counterfactualswhich
Hale'sobjectionexploits is plausiblyregardedas a defectin that analysis. As HartryField has observedin anothercontext, we do seem to
be able to make discriminatinguse of counterfactualswhose antecedents we suppose to express necessaryfalsehoods:if arithmeticwere
inconsistent,set theorywould be inconsistent;if the God of the philosophers (i.e. a perfect, necessarybeing) existed, the righteouswould
have nothingto fear;if the Queenwere your mother,Diana would be
your sister-in-law.1
The appositionhereof the referenceto Fieldto the claimthat the standard
semanticsis defective certainlysuggests that Rosen thinks Field's point
providesa good reasonfor rejectingthe standardsemantics- though to be
fair, he doesn't expressly say as much. If so, he presumablyhas in mind
some such argumentas this: We do in fact make pointful, discriminating
uses of counterfactualswith necessarilyfalse antecedents.Butwe could not
do so if (all) such conditionals were vacuously true, as on the standard
semanticalaccount. Hence the standardaccount is defective.The weakness of this argumentlies, of course,in the second premiss,i.e. in the claim
that vacuoustruth entails incapacityfor pointfulor discriminatinguse.
One kind of case wherewe can makea discriminatinguse of a vacuously
true conditionalis - ironically- noticed by Rosen himself:the case where
someone believeswhat is in fact a necessaryfalsehood,and we attemptto
wean her off it by getting her to see that if what she believeswere true,
something else which she recognises as quite certainly false, or even
absurd,would also be true. The strategyof reductiodemandsthe truthof
our conditional,but not its non-vacuoustruth.It does requirethat we can
get her to appreciatethe truth of the conditionalother than by appealing
to the (necessary)falsehood of its antecedent.But that is quite another
matter,and one which bearson other cases - such as those mentionedby
Rosen - in which we may make discriminatinguses of conditionalswith
impossibleantecedents,includingcases- such as Rosen'sfirst,and perhaps
his second- where an argumentby reductiois not in the offing.The truth
of a conditional (vacuous or not) never by itself guaranteesthat it can
pointfully be asserted. In general, and somewhat roughly, what makes
discriminatinguse of one of a pair of vacuouslytrue conditionalspossible
is that thereis a way of defendingthe one we favourwithout appealingto
the necessaryfalsehoodof its antecedent.The standardsemanticsby itself,
unsupplementedby such broadly pragmatic considerations, does not
1 [4],pp.69-70. Thereferenceto Fieldis to [1],pp.237-8. Thelastexampleobviously
assumesthatyou arenot PrinceCharles.
78
BOB HALE
explain how we can make discriminatinguses, but it does not rule them
out. Only if it did would we have an argumentagainstit.
Referringback to the examplescited above, Rosen writes:
The significantfeatureof these examplesis that the alleged impossibilities supposed in the antecedentsare not logical impossibilities.
They are substantiveimpossibilities,metaphysicalor mathematical;
andwhile theremay be insuperableobstaclesto makingsenseof counter-logicalconditionals,conditionalswhose antecedentsare impossibilitiesof these substantivesorts seem much betterbehaved...
He seems to concede here - albeit somewhattentatively- that there may
be no sensibleuse of conditionalswith logicallyfalse antecedents.2If that
is his view, his fictionalistmust hold that it is at least logically possible
that PW be true, even if this is not a metaphysicalpossibility.And in that
case, the atheist fictionalist should hold that the falsehood of PW is a
logically contingent matter. Since my dilemma is intended to be understood in terms of (broadly) logical necessity and contingency, the
fictionalistwho takes this tack is effectivelyrefusingthe first horn of my
originaldilemma,and opting for the second. That leaves him needingan
account of what he means by 'PossiblyPW is true' - as on the original
dilemma;so far as I havebeen ableto see, Rosen does nothingto bluntthis
second horn.
If that is right, then Rosen had better rescindhis apparentconcession
about logical impossibilitiesand repudiate the standard semantics for
counterfactualconditionalsacross the board. Indeed, since the dilemma
can just as well be constructedusing the other glosses - in terms of strict
conditionals or entailment statements- Rosen originally suggested, he
cannot stop short at tinkeringwith the logic or semanticsof counterfactuals: a thoroughgoing implementationof strategy (b) requires him to
repudiatethe standardclassical account of logical consequence.It is, of
course,not unknownfor peopleto thinkthat a good thing. I am not about
to arguethat they are wrong (thoughI think they are). My presentpoint is
justthat it is an unexpected,and not entirelyendearing,featureof Rosen's
proposalthat a commitmentto some form of relevantistrevisionismlurks
in its shadows.3
The considerationsjustrehearsedin connectionwith strategy(b) do not,
of course, add up to a conclusivecase againstit. Nor do they, or anything
2 I have to confessto some uncertaintyhere over what Rosen meansby logicalas
I am assumingthatthe former
opposedto substantiveor metaphysical
impossibility.
is intendedto includenot onlynarrowlylogicalimpossibilities
(roughly,instancesof
unsatisfiable
schematainvolvingonly logicalvocabularyessentially)but any statementswhosetruthcan be ruledout on purelyconceptualgrounds.
A DESPERATEFIX 79
else I have said, directlyaddressthe furthermanoeuvresin which Rosen
engages,premissedupon a somewhatobscuredistinctionbetweenpossible
worldsand 'islanduniverses'.Butthey do, it seemsto me, indicaterespects
in which strategy (b) is unattractiveand lacking in independentmotivation. And if the final point in my discussionof strategy(a) is well-taken,it
shouldbe clearthat it would be sheerconfusionto think that those further
manoeuvrescould help the fictionalist'scause.
Appendix:Have Menziesand Pettit bluntedthe first horn?
In the final section of [3], PeterMenziesand PhilipPettitclaim to provide
a satisfactoryway for the fictionalist to construe potentially awkward
claimsaboutthe modal statusof PW.They agreethat 'Possibly,PW is false'
is 'a very specialsort of modal claim'- a 'modal dangler'which demands
a specialanalysis,differentfrom that offeredby the fictionalistfor modal
claims in general.They would, presumably,acceptthat the same goes for
'Possibly,PW is true'. They also agree that it must be analysed'in a way
that marksthe peculiarityof the claim but that remainscontinuouswith
the fictionalistanalysisof modal talk generally'.They proposethe following translationschemesfor 'dangling'modal claims:
PossiblyA iff A is truein the actualworld or at one of the otherworlds
posited in PW
NecessarilyA iff A is true at the actual world and at all of the other
possibleworlds posited in PW
Let us supposethe fictionalisttakes PW to be false, but maintainsthat it is
only contingentlyso. That is, she holds that:
PW is false (at the actualworld) but possibly,PW is true.
ApplyingMenzies and Pettit'sproposal,this comes to:
PWis false at the actualworld & (PWis true at the actualworld or at
one of the other possibleworlds posited in PW)
3 It is perhapsworthbrieflynotinga furtherapparentdrawbackof strategy(b). It is
difficultto see how an analysisof counterfactuals
in termsof possibleworldstheory
could avoid the featurewhich, in Rosen'sview, rendersstandardaccounts(i.e.
Lewis'sandStalnaker's)
objectionable.
Certainlyrejectingthestandardaccountscuts
awayone of the centralplanksin Lewis'smainargumentfor realismaboutpossible
worlds- that is, its (alleged)capacityto supporta good analysisof counterfactuals.
To be sure,the fictionalistdoes not wish to regardthis argumentfromexplanatory
virtueas supportingmodalrealism:but he does wantto commendhis own view as
securingthe explanatoryadvantagesof realism(withoutthe ontologicalcosts)- so
he is, indirectly,shootinghimselfin the foot; not a directhit perhaps,but a rather
ricochet!
regrettable
80
BOB HALE
How is this to be understood?Obviouslyit is no good construingit as:
PW is false at the actual world & (PW is true at the actual world or
thereis a possibleworld w suchthat w ? the actualworld & PWposits
w & PW is true at w)
- no good becauseboth componentsof the disjunctiveright conjunctare
inconsistentwith the left conjunct.It must ratherbe understoodalong the
lines of:
PW is false at the actual world & (PW is true at the actual world or
accordingto PW,there is a possibleworld w, distinctfrom the actual
world, at which PW is true)
More generally,the proposedanalysansfor (dangling)'PossiblyA' must be
construedas:
EitherA is true at the actual world or accordingto PW, there is a
possibleworld w, distinctfrom the actualworld, at which A is true
A not unfairgloss on this is:
Either there is a world - the actual world - at which A is true or there
would be one, if PW were true
I think it is a fair question whether this concoction captures a genuine
notion of possibility.Note that the proposed analysans is, as it stands,
consistentwith the necessaryfalsehoodof PW.But if PW were necessarily
false,the rightdisjunctwould be vacuouslytrue.So if the left disjunctwere
false, the truth of the analysanswould scarcelyvindicate the claim that
possiblyA. To get somethingmoreplausiblytakenas sufficientfor truthof
'PossiblyA', we need to incorporatean additionalclause,4stipulatingthe
possibletruth of PW,i.e. we need somethinglike:
Either there is a world - the actual world - at which A is true or (there
would be one, if PW were true, and PW might be true)
But then the claim that PW might be true is being analysedas:
Either there is a world - the actual world - at which PW is true or
(therewould be one, if PW were true, and PW might be true)
And this is clearlyeitherviciouslycircularor regressive.
The University,St. Andrews
Fife, KY16 9AL, Scotland
[email protected]
4 Rememberthatwhat'sbeingofferedis supposedto be an analysis- so it won'tdo
assumption.
justto takethe possibilityof PW'struthas a background
A DESPERATEFIX 81
References:
[1] Hartry Field, Realism, Mathematicsand Modality (Oxford: Blackwell, 1989).
[2] Bob Hale, 'Modal fictionalism:a simple dilemma', Analysis this issue 63-67.
[3] Peter Menzies and Philip Pettit, 'in defence of fictionalism about possible worlds'
Analysis 54 (1994) 27-36.
[4] Gideon Rosen, 'Modal fictionalism',Mind 99 (1990) 327-54.
[5] Gideon Rosen, 'Modal fictionalism fixed', Analysis this issue 68-73.
Modal fictionalism cannot deliver possible
worlds semantics
JOHN DIVERS
According to deflationismyou can have all the benefits of talking
about possible worlds without the ontological costs. (Rosen [9]
p. 330, my emphasis)
1. Rosenconceiveshis modal fictionalismas a versionof deflationism.The
fictionalistwho is preparedto assertthe propositionexpressedexplicitly
by the ordinarymodal sentencetype:
(1) Theremight have been blue swans.
wishes to gain the benefits associated with expressinghis belief, as the
genuine modal realist would, in terms of possible worlds by using the
sentencetype:
(2) Thereis a possibleworld in which there are blue swans.
but he wishes to do so while avoidingan ontologicalcommitmentto nonactualworlds. Rosen'sfictionalistproposalis that one may gain the desirable form of expression while avoiding the unwanted ontological
commitmentby using (2) as an ellipticalmeansof expressingthe proposition which is expresseddefinitivelyand explicitlyby the sentencetype:
(3) Accordingto PW,thereis a possibleworld in which thereare blue
swans.
In general the fictionalist interpretationof possible world discourse is
derivedfrom a genuine modal realist theory (story) PW. PW associates
with each ordinarymodal sentenceP a genuinemodal realistanalysis,P*
and so entailsevery instanceof the schema:
ANALYSIs
55.2, April1995, pp. 81-88. ? JohnDivers