Documente Academic
Documente Profesional
Documente Cultură
.
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 Analysis Committee are collaborating with JSTOR to digitize, preserve and
extend access to Analysis.
http://www.jstor.org
The GettierProblem
SCOTTSTURGEON
Thirty years ago this journal published the most influentialpaper of
modernanalyticepistemology- EdmundGettier's'Is JustifiedTrueBelief
23, 1963, pp. 121-23). In it Gettier refuted a clasKnowledge?' (ANALYSIS
sic theory of propositionalknowledge by constructingthought experiments to test the theory.A cottage industrywas born. Each responseto
Gettierwas quicklymet by a new Gettier-stylecase. In turntherewould be
a responseto the case, a furtherGettierscenario,and a reiterationof the
process. The industry'soutput was staggering.Its literaturebecame so
complicated,its thoughtexperimentsso baroque,that commonsensewas
stretchedbeyondlimit.The deepsignificanceof Gettier'swork drownedin
the resultingcacophony.That significancecan be seen by reflectingon two
points:first,why the problemarises;and second, how it is to be solved.
1. WhatgeneratesThe GettierProblem?
Infallibilismis the view that whatevergives rise to epistemicjustification
guaranteestruth.Accordingto the infallibilist,it is impossibleto justifiedly
believesomethingfalse. This has the tidy result that justificationis sufficient to convert belief into knowledge. Thus the InfallibilistView of
knowledge:
(IV) S knows P iff (a) S believesP
(b) S'sbeliefin P is justified.
But since guaranteesof truth are hardto come by, Infallibilismleads to a
dilemma:either we are justifiedin believingfar less than commonsense
admits,or justificationis fallible.
Fallibilistsopt for the latter horn of the dilemma. This has the messy
result that justification is not sufficient to convert belief into knowledge. For you cannot know something false. Since you can, according
to the fallibilist, justifiedly believe something false, it follows that
knowledge is more than justifiedbelief. The question then is, what has
to be added?
The view attacked by Gettieranswers this question very simply:add
truth.This is the so-called'traditionalview of knowledge':
(TV) S knows P iff (a) S believesP
(b) S's beliefin P is justified
(c) P is true.
ANALYSIS
53.3, July 1993, pp. 156-164. @ Scott Sturgeon
158
SCOTT STURGEON
159
justification.Let me explain.
debate springsfrom this quesEpistemology'sinternalism/externalism
tion: what determinesepistemicjustification?Internalistssay justification
is determinedby an agent'sperspective:reasonerswho are perspectivally
equivalentare justificationallyequivalent;while externalistssay justification is determinedby featuresindependentof an agent'sperspective.But
what does this differencereallycome to?
Notice first that two scenarios appear the same to a person iff her
perspectiveon each is the same.So if we can reckonthe featuresgivingrise
to appearance,we will have found those definingan agent'sperspective.In
this regardit is customaryto say that two scenariosappearthe same to a
person when her relevantappearance-statesare phenomenallythe same.
More carefully,scenariosP1 and P2 appearthe same to S just in case her
mental-representationsof them are, or would be, tokens of the same
phenomenaltype. Epistemicinternalismthen amountsto this idea:
justificationis determinedby the phenomenalfeaturesof an agent's
mind:reasonerswhose mental states are phenomenallyidenticalare
justificationallyidentical;
and epistemicexternalismamountsto this idea:
justificationis determinedby featuresindependentfrom the phenomenal featuresof an agent'smind.
Now, we can easily see that full justificationis determinedby features
independentfrom an agent'sperspective.Thus externalismis true of the
crucial ingredientneeded to convert belief into knowledge:
Lesson One - externalismis true of full justification.
To see this, notice that full justificationmustsatisfyall externalconstraints
on knowledge.For instance:
(El) one is fullyjustifiedin believingP only if one is reliableaboutP,
(E2) one is fully justifiedin believinga truepropositiononly if one
does so becauseone is reliable,
beliefonly if one's
(E3) one is fully justifiedin a perceptually-based
states
are
caused.
perceptual
non-deviantly
And so forth.The point is: everyobjectivereal-worldconnectionnecessary
for knowledgewill be necessaryfor full justificationas well. The failureof
any such connectionwill functionas a defeaterfor condition (b) of (M).
Hence the justificationof that condition will convert to full justification
exactlywhen the externalconstraintson knowledgeare satisfied.Gettier's
work thus provides a direct route to the conclusion that epistemic externalism is true of the crucial ingredient needed for knowledge.
I60
SCOTTSTURGEON
4. Infallibilismrevisited.
(M) asserts that knowledge is fully justified true belief. The second lesson
of Gettier is this:
161
you know all lawsof natureandall mattersof factup to the momentin question. In other words, suppose you know everythingin T. Finally,let this
knowledgerenderit indeterminate,
thoughlikely,that the photonwill go to
the right.On the basisof this you come to believethat the photonwill go to
the right.And supposeit will. Do you know the photonwill go to the right?
I submit that you do not know where the photon will go. After all,
there'sanotherpossible world exactly like yours save the photon goes to
the left. In it you believethe photon will go to the right with exactly the
sameevidence- namelyall the evidence- but you are mistaken.Hencethe
view that you actuallyhave knowledgeimpliesthat bruteluck makesthe
differencebetweenknowledgeand ignorance.This is unacceptable.Luck
does not contributeto our possessionof knowledge.
Eitherwe have located a counter-exampleto the standardpost-Gettier
model of knowledge,or there lurks a defeatersomewherein the photon
case. The latteroption seemsappropriate:indeterminacyitselffunctionsas
a defeater.When all the facts save the issue at hand do not fix the issue,
then this is reasonto withholdjudgement.Case-(iii)scenariosof this sort
are incompatiblewith full justification.
On the other hand, nomic indeterminaciesare not unknowable.If, for
example,the photon goes to the rightand then strikesa screen,we might
verywell come to know the path taken.The lesson is: nomic indeterminacies are knowablewhen they fit into mattersof fact which forge an epistemic route to them. In particular,nomic indeterminaciesare known
through their effects. They are known when the moment of knowledge
itself post-datesthe moment known. But T will logically imply the fact
known in these cases (as suggestedfive paragraphsback). We will have
shiftedfromcategory(iii) to category(i).
This suggeststhe following lemma:
LemmaA If microscopicindeterminaciesclimb into the macroscopic
domain, then we have much less knowledge than commonsense
would admit.
Indeed:if the wobblinessof the tiny reachesup into the world of commonsense, then I submitwe know truthsalmost exclusivelypost facto. Naturally, this does not mean that our commonsensebeliefs are in any sense
unreasonableor less useful.Theyjust do not amountto knowledge.As far
as I can tell, this differencedoes not make any difference.
(b) Supposewe expressthe fact that all Fs are H thus:
UF
(Vx)(Fx -+ Hx).
162.
SCOTTSTURGEON
L claimsthatallmovingobjectstravelat or belowthespeedof
Perhaps
how we can
light.Does (M) providea usefulmodelfor understanding
knowfactsof thissort?
Yes.ForT willcontaina nomictruthabouteveryspeedgreaterthanthat
Thesewilljointly
at thatspeedis impossible.
of light,to wit,thattravelling
on
a
with
entailL . Thus,(M)rendersnomicknowledge par
anyothersort
to
of knowledge.Wehaveit whenour evidencestandsup the facts,and
our evidenceso standswhenthe relevantfactsimplythe nomicfact in
question.
On the otherhand,(M) explainswhy nomicknowledgeis problematic froma Humeanperspective.Thereare two Humeanperspectives
worth distinguishinghere:weak and strong.Accordingto the weak
perspective, nomic knowledge is grounded solely in non-nomic fact.
According to the strong perspective, nomic knowledge is grounded
solely in empiricalfact.
163
164
ROBERTJ. FOGELIN
placedontheontologyof ourepistemic
theory.Butonceenforced,thelimit
as well.Weareleftthenwith
of ourknowledgeseemsnaturallyrestricted
definitionof knowledge:
a streamlined
post-Gettier
(PG) S knowsP iff (a)S believesP
(b)S'sbeliefin P is fullyjustified.
Therecan be no doubtbutthatGettierput his fingeron somethingvery
indeed.2
important
King'sCollegeLondon
Strand,London WC2R2LS
2 I thank the discussiongroup to which I belong for encouragement:Tim Crane,Keith
Hossack,Mike Martin,LucyO'Brien,David Papineau,GabrielSegal,BarrySmithand
BernhardWeiss;and I thank Tom Senor,Peter Smith and an anonymousrefereefor
helpfulcomments.
Hookway on KnowledgeInferences
ROBERT
J. FOGELIN
Inhisbook,Scepticism
1990),Christopher
(London:
Hookway
Routledge,
callsattentionto a classof knowledgesentenceshe labelsQ-claims:
Some[knowledge
sentences]likethefollowingtakean indirectquestioncompliment:
themurder
X knowswhocommitted
X knowswhywaterexpandson freezing
X knowswhenthetrainleavesfor London
X knowswhethertheatomicweightof sodiumis 29
X knowshowtheprisonerescaped.
Theirformcanbe expressed:
X knowsQ. (pp.196-97)
Hookwaycontrastssuch Q-claimswith sentenceshe call P-claims.Pclaimshavetheform:
X knowsthatP.
Hookwaypoints out, quite correctly,that the analysisof knowledge
on P-claimsto the relative
sentenceshas to a largeextentconcentrated
ANALYSIS
53.3, July 1993, pp. 164-168. @ Robert J. Fogelin