International Phenomenological Society

Learning from Words

Author(s): Jennifer Lackey
Reviewed work(s):
Source: Philosophy and Phenomenological Research, Vol. 73, No. 1 (Jul., 2006), pp. 77-101
Published by: International Phenomenological Society
Stable URL: http://www.jstor.org/stable/40040993 .
Accessed: 16/03/2013 18:35

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
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.

International Phenomenological Society is collaborating with JSTOR to digitize, preserve and extend access to
Philosophy and Phenomenological Research.

http://www.jstor.org

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 Research
and Phenomenological
Philosophy
Vol. LXXIII,No. 1,July2006

from
Learning Words*
JENNIFER LACKEY
IllinoisUniversity
Northern

Thereis a widelyacceptedfamily of viewsin theepistemology of testimony

centering around theclaim that belief is thecentral item involved in a testi-
For in
monialexchange.1 instance, describing process learningvia tes-
the of
timony,ElizabethFrickerprovidesthefollowing:"one language-user has a
belief,which gives riseto an utterance by him; as a resultof observingthis
utterance anotheruserof the same language,his audience,comes to share
thatbelief'2 In a similarspirit,AlvinPlantingasaysthat"...a beliefon the
partofthetestifiee has warrant only if thatbeliefhas warrant forthetesti-
fier."3In both of these passages, we find strands of what I shall call the
BeliefView ofTestimony the
(hereafter, BVT).4
Thereareseveraldifferent yetrelatedcomponents oftheBVT. First,while
statements are necessaryfortheprocessof communication, theyare merely
vehiclesforexpressing -
beliefs they enable us to make publicwhatwould
otherwise remainprivate.Strictlyspeaking,then,we do not learnfromone
another'swords - we learnfromone another'sbeliefs.5This is one of the

Editor's Note: This paper won the Young Epistemologist Prize forthe RutgersEpiste-
mologyconference heldin 2005.
1 As MichaelWelbournesays,"...it is moreor less standard to describethe whole testimo-
nial processin thevocabulary of belief (1994, p. 305).
2
Fricker(1987, p. 68, emphasisadded). In a morepreciseformulation, Frickersays: "A
speaker,believingthatP, and wishingto communicate thisbelief,makes an utterance
whichconstitutes hisasserting thatP\ his audience,a hearer,observingand understand-
ingit...as a resultcomesalso to believethatP" (1987, pp. 68-9).
Plantinga(1993, p. 86, emphasisadded).
Proponents of different strandsof the BVT includeAustin(1979), Welbourne(1979,
1981, 1986,and 1994), Evans (1982), Hardwig(1985 and 1991), Fricker(1987, 1994,
and 1995), Ross (1986), Coady (1992 and 1994), Burge(1993 and 1997), Reid (1993),
Plantinga(1993), Dummett (1994), McDowell(1994), Williamson(1996 and 2000), Audi
(1997, 1998,2004, and 2006), Faulkner(2000), Owens (2000), Reynolds(2002), Adler
(2002), Pritchard (2004), and Schmitt (2006). It is of interest to notethatthe BVT cuts
acrosstheso-calledreductionist/non-reductionist debatein theepistemology of testimony.
Forinstance,Fricker(1987, 1994,and 1995) is a reductionist, Plantinga(1993) is a non-
reductionist, and Faulkner(2000) endorsesa hybridreductionist/non-reductionist view,
yetall threeespousestrandsof theBVT.
5 Forinstance,RobertAudisaysthat". . .ifone takesit(as ThomasReid mayhave) thatthe
recipientof testimony is (characteristically)
responding to theattester'sbelief- orat least

LEARNING FROM WORDS 77

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
reasons whynearlyeveryonetakes sincerityon thepartof the speakerto be a
necessarycondition for testimonial knowledge.6In orderto properly learn
froma speaker's belief,thereneeds to be a beliefpresentfromwhich to learn.
Thus, if a speaker is insincere and expresses what she herself does not
believe, thenthereis nothingforherto pass on to a hearer.
Second, the process of communicatingvia testimonyinvolves a speaker
transmitting herbelief to a hearer,along withthe epistemicproperties it pos-
sesses. For instance,following Fricker's talk of sharing beliefs and Plant-
inga's characterization of the transmissionof warrant,David Owens claims
that "...testimony transmitsknowledge of/? by transmittingthe probative
forceof reasons forbelief in p fromone partyto another."7Testimony,there-
fore,is said to be incapable of generatingnew epistemic features - it merely
has the capacity to transmitfrom one person to anotherbeliefs that have
already been renderedwarranted,justified, or instances of knowledge via
anothersource.8
Third,and closely related,statementsthemselves are not the bearers of
epistemicsignificance- beliefs are. Notice, for instance,that Plantinga does
not say thata belief on the partof the testifieehas warrantonly if the testi-
fier's statement has warrant;instead, he focuses entirelyon the epistemic
status of the speaker's belief. This is a naturalconsequenceof the previous
two points: if statementsare merely vehicles for transmittingbeliefs that

a presumption thereof- andnotjusttoa linguistic or othersymbolicact, one mayfindit

plausibleto arguethattestimony is a conditionally basic source of justification
and per-
haps of knowledgeas well" (2004, p. 21). Similarly, Duncan Pritchard claims thatthe
paradigmcase of testimony is "...the intentional transfer ofa belieffromone agent to
another,whetherin theusual way via a verbalassertionmade by theone agentto the
other,or bysomeothermeans,suchas through a note" (2004, p. 326, emphasisadded).
AndErnestSosa writes,"On thedefaultassumption... whichmustbe thatof sincerity, the
speaker's utterancedoes give to understand wliatthespeakerthinks.So, the utterance
carriesa deliveranceas to thespeaker's mind....If we are to knowa speaker's mind
through his utterances,thespeakermusthavea reliablecompetenceto statehis mind.He
mustbe able through his utterances to deliversafe deliverancesaboutwhathe thinkson
thetopicat hand"(2006, p. 121,emphasisadded).
For discussionsspecificallyabouttherole of sincerityin testimony, see, for instance,
Welbourne(1979, 1981, 1986, and 1994), Hardwig(1985 and 1991), Ross (1986),
Fricker(1987, 1994,and 1995), Plantinga(1993), McDowell (1994), Audi (1997 and
1998),Root(2001), Owens(2000), and Adler(2002). For indirectendorsements of sin-
cerityas a necessaryconditionfor testimonial knowledge(warrantedbelief,justified
belief)- e.g., via the strongerrequirement thatthespeakerhave the knowledge(war-
rantedbelief,justifiedbelief)in question-see Burge(1993 and 1997), Williamson(1996
and 2000),Audi(1997, 1998,and 2006), Dummett(1994), Reynolds(2002), and Schmitt
(2006).
7
Owens(2000, p. 169).
WhenI speakof warrantandjustification, I am interested in theseconceptsonlyinsofar
as theyhavesomeclose connection withknowledge.Thus,theremaybe some subjective
notionsof theseconceptsthatescape some of thearguments I give in thispaper. My
interesthere,however,is in theepistemology of testimony.

78 JENNIFERLACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 thenthestatements
alreadypossess epistemicproperties, themselves simply
the
dropoutof epistemic picture.
Whilethereis muchthatis intuitively plausibleabouttheBVT, in what
followsI arguethatthisentirepictureof theepistemology of testimony is
fundamentally In
incorrect. I
particular, show that,both causallyand epis-
temically,statements,not beliefs,are the crucialitems in a testimonial
exchange.9Hence,eachofthecomponents thatconstitute
theBVT is false.

1. The Transmissionof EpistemicProperties:Necessity

Let us beginwithwhatis perhapsthefundamental thesisof theBVT, which
I shall call the Transmissionof EpistemicProperties (hereafter, TEP).
Roughly,the thoughtexpressedby TEP is that a testimonialexchange
involvesa speaker'sbelief,alongwiththeepistemicproperties it possesses,
being transmitted to a hearer.There are two dimensions to TEP; one is a
necessitythesisand the otheris a thesis.
sufficiency More precisely,

TEP-N: For everyspeaker,A, and hearer,B, B's beliefthatp is

warranted known)on thebasis of A's testimony
(justified,
thatp only if A's beliefthatp is warranted (justified,
known).10

TEP-S: Foreveryspeaker,A, andhearer,B, if (1) A's beliefthatp

is warranted known),(2) B comes to believethat
(justified,
of A's testimony
p on thebasisofthecontent thatp9u and

9 thispaper,I shallfocuson therole of a speaker's statement in a testimonial

Throughout
exchange.However,since a beliefmaybe acquiredon thebasis of a speaker's testi-
monythatis notin theformofa statement - by a physicalgesture,forinstance - a more
precise characterization of thearguments in thispaper would be in terms of a speaker's
act of communication. (For more on the natureof acts of communication, see my
(2006a).) Nevertheless, because a discussionof therequisiteconditionsfor an act of
communication lies outsidethescopeofthispaper,I shall continueto frametherelevant
issuesin termsofstatements.
10 of thenecessitythesis(TEP-N) includeWelbourne(1979, 1981, 1986, and
Proponents
1994),Hardwig(1985 and 1991),Ross (1986), Burge(1993 and 1997), Plantinga(1993),
McDowell (1994), Williamson(1996 and 2000), Audi (1997, 1998,and 2006), Owens
(2000), Reynolds(2002), Adler(2002), and Schmitt (2006). Fora slightlyweaker version
of thisthesis,see Dummett (1994).
Two pointsshouldbe notedregardingthiscondition.First,since TEP-S expresses the
conditions requiredfortestimonial warrant(justification, knowledge),(2) specifiesthat
thehearermustformthe beliefin questionon thebasis of thecontentof thespeaker's
testimony. This is to precludecases wherea beliefis formedentirely on thebasis of fea-
turesaboutthespeaker'stestimony. Forinstance, ifyousay,in a sopranovoice, thatyou
have a sopranovoice and I come to believe thisentirelyon the basis of hearingyour
sopranovoice,thenthewarrant formyresulting
(justification) beliefis perceptualrather
thantestimonial. (Thisexampleis a slightvariation ofone foundin Audi (1997). For fur-
therdiscussionofthedistinction inspiredbythesesortsof cases, see my(1999).) Condi-
tion(2), therefore,is includedto prevent cases of thissortfromqualifying as instancesof

LEARNINGFROMWORDS 79

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 testimonialwarrant (justification,knowledge).Notice,however,thatI do notrequirethat
thehearer'sbeliefbe formedentirelyon thebasis of thecontentof thespeaker's testi-
mony.This is to allowcases wherea givenbeliefis formedon thebasis of boththecon-
tentof a speaker'stestimony and,say,perception, butwhere,bothcausallyand epistemi-
cally,thetestimony is sufficient
forthesubjectto holdthewarranted(justified, known)
beliefin question.In suchcases,thoughperception playsbotha causal and an epistemic
role,it maybe arguedthatthesufficiency of testimony rendersthe belief in question
genuinely testimonial.
Second,thereis a weaker versionof (2) (hereafter,(2*)) thatrequiresonlythatB
cometo believethatp on thebasisofthecontent of A's testimony, notthatsuch a belief
be basedon thecontent ofA's testimony thatp. There is also a correspondingly weaker
versionofTEP-N:
TEP-N*: Foreveryspeaker,A, and hearer,B, B's beliefthatpis warranted (justi-
fied,known)on thebasisof A's testimony onlyif (i) A's beliefthatp is
warranted(justified, known),and (ii) B's beliefthatp is appropriately
connectedwiththecontent of A's testimony.
(Condition(ii) is includedto prevent beliefsformed on thebasis of featuresdisconnected
fromthecontent ofthetestimony in question,suchas theone above aboutyourhavinga
sopranovoice, fromqualifying as testimonial. Of course, if TEP-N is substituted with
TEP-N*,morewillneedto be said aboutwhattheappropriate connectionin (ii) amounts
to.) Now, thereare at leasttwodifferent kindsof cases thatmightsupportincludingboth
(2*) ratherthan(2) and TEP-N* ratherthanTEP-N. For instance,supposethatVirginia
asks me whether itis cold outsideand I respondby saying,"Your hat and glovesare in
thecloset." Here, one mightsay that,given the context,Virginiaacquires warranted
belief(justified belief,knowledge)thatitis cold outsideon thebasis of mytestimony that
herhat and glovesare in thecloset.Thus,one mightclaim thatboth(2) of TEP-S and
TEP-N are,in fact,too strongsinceone can acquiretestimonially warranted belief(justi-
fiedbelief,knowledge)thatpon thebasisofthecontentof a speaker's testimony thatq.
The secondtypeof case thatmaypose a problemforbothof thesethesescan be seen by
consideringthefollowing:supposethatEdna asks me whetherI see any bald eagles
through mybinocularsand I respondwitha nod. Here, it mightbe argued thatEdna
acquireswarranted belief(justifiedbelief,knowledge)thatI see bald eagles through my
binoculars on thebasisof thenon-verbal testimony providedbymynod. In thisway,one
mightclaimthatwhilewarranted belief(justifiedbelief,knowledge)thatp is acquiredon
thebasisofthecontent ofmytestimony, itis notbasedon thecontent ofmytestimony that
P-
Although I thinkbothtypesof examplesprovideconvincingevidence forincluding
the weaker versionsfoundin (2*) and TEP-N*, thereare also reasons to preferthe
stronger (2) and TEP-N. Withrespectto thefirsttypeof case, forinstance,one might
arguethatVirginia'swarranted belief(justifiedbelief,knowledge)thatit is cold outside
reliestoo heavilyon memoryand inferenceto qualifyas testimonial. In particular, one
mightclaimthatVirginiamustrelyon backgroundinformation storedin memoryabout
therelationship betweencold weatherand the wearingof hatsand glovesand explicitly
inferthatit is cold outsidefrommytestimony thather hat and glovesare in thecloset.
Andeventhoughmemory and inference arguablyplaya rolein theacquisitionof all tes-
timonialknowledge, one mightclaimthatheretherole of memoryand inferenceis sim-
plytoo significant fortheresulting warranted belief(justifiedbelief,knowledge)to prop-
erlybe regardedas testimonial. Regarding thesecondtypeof case, it maybe arguedthat
thoughI do notexplicitly statethatI see bald eagles through mybinoculars,thereis a
clearsensein whichthenon-verbal testimony providedby mynod has thisas itscontent.
Forjustas myresponding "Yes" in thiscontextamountsto "I see bald eagles through my
binoculars,"so toodoes mynoddingin thiscontext amounttotestifying tothisproposition.
I do notwishto here choose betweeneither(2) and (2*) or TEP-N and TEP-N*.
ThoughI am leavingthestronger (2) andTEP-N in mycharacterizations of therelevant

80 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 (3) B has no undefeated defeatersforbelieving thatp, then
B's beliefthatp is warranted(justified,known).12

There is much that is intuitive about both of these theses. For, in many
respects,a testimonialchain seems to be much like a bucket brigade:in order
to give you a fullbucketof water,I musthave a full bucket of waterto pass
to you. Moreover, if I give you a full bucket of water, then- spills
aside- thebucketof wateryou now possess as a resultof our exchange will
also be full. Similarly,in orderto transmitto you a warrantedbelief, I must
have a warrantedbeliefto pass to you.13Moreover, if I transmitto you a war-
rantedbelief, then- defeatersaside- the belief that you now possess as a
resultof our exchange will also be warranted.Despite theirintuitiveplausi-
bility,however,I shall argue thatbothof these theses are false and, therefore,

theses, they can easily be substitutedwith the weaker (2*) and TEP-N* for those con-
vinced by the cases discussed above.
12
Proponentsof differentversions of the sufficiencythesis (TEP-S) include Austin (1979),
Evans (1982), Fricker(1987), Coady (1992), and Owens (2000). Burge (1993), William-
son (1996 and 2000), and Audi (1997) endorse qualified versions of this thesis. For
instance, Burge claims that "[i]f one has acquired one's belief from others in a normal
way, and if the othersknow the proposition,one acquires knowledge" (1992, p. 477, fn.
16, emphasis added). Timothy Williamson writes that "[i]n normal circumstances, a
speaker who asserts thatP therebyputs a hearerin a position to know that P if (and only
if) the speaker knows that P" (1996, p. 520, emphasis added). Similarly, Audi writes,
"Concerning knowledge, we mightsay thatat least normally,a belief thatp based on tes-
timonythereby constitutesknowledge... provided that the attester knows thatp and the
believer has no reason to doubt eitherp or the attester's credibilityconcerning it" (1997,
p. 412, emphasis added). It should be noted, however, thatthe arguments offered in Sec-
tion 2 against TEP-S subsume even these three qualified versions of the sufficiency the-
sis.
It is of furtherinterestto note that there is a version of the sufficiency thesis that is
even strongerthan TEP-S. With respect to warrant and justification,it can be expressed
as follows:
TEP-S*: For every speaker, A, and hearer, B, if (1) A's belief that p has x amount
of warrant(justification),(2) B comes to believe thatp on the basis of the
contentof A's testimonythatp, and (3) B has no undefeated defeaters for
believing that/?,then B's belief thatp has x amount of warrant (justifica-
tion).
A similar principle for knowledge can be formulatedby substituting"has x amount of
warrant"with "is known with degree x of certainty"both in condition (1) and in the con-
sequent of the conditional. In Section 2, I shall frame my arguments against the weaker
TEP-S since theywill apply just as well to the strongerTEP-S*. For instance,if A's belief
thatp is not warrantedand B acquires warranted belief thatp on the basis of A's testi-
mony thatp, thenobviously A and B do not share the epistemic propertyin question to the
same degree. Nevertheless, I should mentionthat, given the picture of testimonypainted
by proponentsof the BVT - in which a speaker's belief along with its epistemic proper-
ties is transmittedto a hearer- it is not entirelyclear what would justifyendorsing only
the weaker TEP-S.
13
For ease of exposition, I shall sometimes focus on only one epistemic property,such as
warrant.The argumentsin this paper, however, are completely general, and apply just as
well to knowledge and to many conceptions of justification(and even, at times, to ration-
ality).

LEARNING FROM WORDS 81

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
that each of the componentsof the BVT is false as well. In this section, I
focus on the necessitythesis(TEP-N), leaving the sufficiencythesis (TEP-S)
forSection 2.
To begin,considerthe following:

CONSISTENT LIAR: When Berthawas a teenager,she suffereda

head injurywhile ice skating and, shortlyafterwards, became quite
prone to telling lies, especially about her perceptual experiences
involving wild animals. Afterobservingthis behavior, her parents
became increasinglydistressedand, afterconsulting various psy-
chologistsand therapists,finallytook her to see a neurosurgeon,Dr.
Jones. Upon examining her, Dr. Jones noticeda lesion in Bertha's
brainwhichappearedto be the cause of her behavior, and so it was
decidedthat surgerywould be the best option to pursue. Unfortu-
nately, Dr. Jones discoveredduring the surgerythat he couldn't
repairthe lesion- instead, he decidedto modifyher currentlesion
and create another one so that her pattern of lying would be
extremelyconsistentand would combine in a veryprecise way with
a patternof consistentperceptualunreliability.Not only did Dr.
Joneskeep theprocedurethat he performed on Berthacompletelyto
himself,he also did this with the best of intentions,wanting his
patientto functionas a healthy,happy,and well respectedcitizen.
As a resultof thisprocedure,Bertha is now- as a young adult
- a radicallyunreliable,yet highly consistent,believer with respect
to her perceptual experiences about wild animals. For instance,
nearlyeverytime she sees a deer, she believes that it is a horse;
nearly every time she sees a giraffe,she believes that it is an
elephant;nearlyeverytimeshe sees an owl, she believes that it is a
hawk, and so on. At the same time, however,Berthais also a radi-
cally insincere,yet highly consistent,testifierof this information.
For instance,nearlyeverytimeshe sees a deer and believes that it is
a horse,she insincerelyreportsto othersthat she saw a deer; nearly
everytimeshe sees a giraffeand believes that it is an elephant,she
insincerelyreportsto others that she saw a giraffe,and so on.14
Moreover, because of her consistencyas both a believer and a liar,
those aroundher do not have any reason fordoubtingBertha's reli-
abilityas a source of information.Indeed,in her home community,
she is regardedas one of themost trustworthy people to consulton a

14
Noticethatbecause Bertha'sice skatingaccidenttookplace whenshe was a teenager,
thecontentof her wild-animal beliefswas alreadyfixedin a normalway, and hence
thereis nothing
abouttheexamplethatis incompatiblewiththetruthofa causal theoryof
content.

82 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 wide range of topics. Yesterday,Bertha ran into her next door
neighbor,Henry,and insincerelythough correctlyreportedto him
thatshe saw a deer on a nearbyhikingtrail.Since, in additionto his
trustin Bertha,it is not at all unlikely forthereto be deer on the
hikingtrailin question,Henryreadilyaccepted hertestimony.16

The firstpoint to notice about CONSISTENT LIAR is that even though

Bertha is a radicallyunreliable believer withrespectto her animal sightings,
she is nonetheless an extremelyreliable testifier of this information -
indeed,even more reliable than many average testifierswho frequently exag-
gerate, or
distort, are simply wrong in their reportsabout what is true. For,
as a resultof theprocedureperformedby Dr. Jones,nearly every time Bertha
sees a deer, she believes thatit is a horse yet reportsto othersthat she saw a
deer; nearly everytimeshe sees a giraffe,she believes that it is an elephant
yetreportsto othersthat she saw a giraffe,and so on. Moreover, this point
can be furtherillustratedby noticing that Bertha's statementssatisfy the
counterfactualsthatare, perhaps,most likelyto capturereliability,despite the
factthatherbeliefsdo not.For instance,while Bertha's beliefs are not sensi-
tive in Nozick's sense, her statementsare, i.e., though Bertha would still
believe that/?ifp were false, she would not statethat/?if p were false.17The
same can be said withrespectto Sosa's safetyrequirement,i.e., though Ber-
tha would believe thatp withoutit being so that/?,she would not statethatp
withoutit being so thatp. Her statements,then, are safe in Sosa's sense,
even thoughher beliefs are not.18Thus, it is clear that Bertha's statements,

15
In order to avoid the worrythat Bertha's false beliefs would eventually be detected, we
can also stipulate that,now that she is a young adult, there are very few occasions in
which she either interacts with or discusses wild animals. For instance, we can assume
that she lives in an urban environment,is allergic to most non-human animal hair and
hence rarely has contact with such creatures,and so on.
16
A similar, though substantially less developed, example can be found in my (1999).
There, however, my purpose was simply to argue against a version of TEP framed in
termsof knowledge, ratherthan the general BVT that is my concern here. Moreover, it
could plausibly be argued that the example found in my (1999) is a Gettier-type case,
whereas I shall later argue that there is simply no compelling sense in which
CONSISTENT LIAR is such a case. One furtherpoint- compare CONSISTENT LIAR
withthe following: I know that Margot only reports that it is snowing outside when it is
not. Thus, when she reports to me that it is snowing outside, I supplement her testimony
with the background informationI possess about her testimonialhabits and come to know
thatit is not snowing outside. My resultingknowledge, though in part based on testimony,
also relies quite heavily on perception,memory,and inductive inference. Thus, one may
plausibly argue thatthis sort of case does not pose a problem for TEP-N since this thesis
applies only to cases of pure testimonialbelief. But notice that this kind of move cannot
be made with respect to CONSISTENT LIAR - the beliefs thatHenry forms on the basis
of Bertha's testimonyare purely testimonialin every relevant sense.
17
See Nozick (1981).
18
See Sosa (1996, 1999, 2000, and 2002). See also Williamson (2000).

LEARNING FROM WORDS 83

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
unlike her beliefs, are an excellent epistemic source of informationabout
wild animals.
The second point that should be emphasizedabout CONSISTENT LIAR
is that though the procedureDr. Jones performedon Bertha rendersher a
highly reliable testifierwith respect to her wild animal sightings, she is
nonethelessstill an agent with intentionalstates that affectthe statements
that she offersto others. In this sense, Berthais importantlydifferent from
non-agentialmechanisms, such as thermometersand odometers, that may
reliablyconvey informationbut not throughanythingreasonablyregardedas
testimony.For instance, Bertha can, if she chooses, refrainfrom telling
Henryanythingat all; she can talk about the weatherinstead; she could tell
an unrelatedlie (e.g., "I saw an escaped elephantin the woods")- All of these
possibilitiesare available to Berthain a way that they are not forthermome-
ters,odometers,and othersuch non-agentialsources of information.For this
reason,Henry's learningfromBertha's wordsfalls underthe generalcategory
picked out by paradigmaticinstancesof testimonyratherthan that picked out
by such non-testimonialmechanisms.19
The thirdsalientfeatureof CONSISTENT LIAR is thatDr. Jones delib-
eratelyperformedprocedureson Bertha's brain, not only so that her percep-
tual errorsand lying practiceswould be highly consistent,but also so that
theywould combine in such a way so as to renderher an extraordinarily reli-
able testifier.In large partbecause of therole thatDr. Jones plays in Bertha's
condition,thereis no relevantaccidentality in the belief that Henry forms
on the basis of her testimonyand, hence, CONSISTENT LIAR cannot plau-
sibly be regardedas a Gettier-typecase.20 With respect to the truthof the
belief in question, it is neitheran accidentthat Berthareportedthat therewas
a deer on thehikingtrailnorthatHenrycame to hold this belief on the basis
of her testimony.For, because of her two brain lesions, Berthais such that
nearly every time she sees a deer,she believes it to be a horse, yet reports

19 It maybe further objectedthateven if Bertha'sstatements are an excellentsource of

information, theydo not qualifyas testimony since our ordinaryusage of this term
requiresthatthespeakerbelieve theproposition in question.By way of responseto this
objection,noticethatit makesperfectsense to speak of a witnesstestifying in a court-
roomto something thatshe herselfdoes notbelieve- it is called perjury.Moreover,I
have elsewhereargued(2006a) thatthecentralfeatureof testimony is thatthespeakerin
questionlias theintention to expresscommunicable contentwhenmakingthestatement in
question.Bertha,unlikenon-testimonial sources of information, such as thermometers
and odometers, clearlysatisfiesthiscriterionsince she intendsto expresscommunicable
content,evenifshedoes notintendto expresstheparticular contentthatshe does. (I am
gratefulto Liz Camp,MarkHeller,and IshaniMaitraforpressingan objectionthatled to
theinclusionofthispoint.)
See Gettier(1963) fortheinitialformulation of theproblemand Shope (1983) forsome
of themanypermutations ofGettier cases. Forthisresponseto the sortof testifier found
inCONSISTENT LIAR (i.e., thatit is a Gettier-type case), see, forinstance,Plantinga
(1993, pp. 82-3) and Chakrabarti (1994, p. 110).

84 JENNIFERLACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 that it is a deer. Thus, Bertha's deer-reports nearly always covary with her
deer-sightings, despite taking a slight detourthroughhorse-believings.Fur-
thermore,given all of Henry's excellent inductiveevidenceon behalfof Ber-
tha's testimonialpractices,thereis also no relevantaccidentally in his com-
ing to formthe belief in question on the basis of her report.With respectto
the warrantorjustificationof Henry's belief, similar remarkscan be adduced:
Dr. Jones deliberatelymodifiedBertha's firstbrain lesion and addeda second
one preciselyso thatBertha would turnout to be a reliable testifier.21 More-
over, there are no close possible worlds in which Dr. Jones did not perform
thesurgeryon Bertha,and hence thereare no close possible worlds in which
Berthadoes nothave theconditionshe has. Because of this, thereis no sense
in which it is an accidentthatBertha is a reliable testifierand hence a source
of warrantedorjustifiedbelief.
The fourthfeatureof CONSISTENT LIAR that is here relevantis that
Henry not only fails to possess reasons to doubt Bertha's competence and
sincerity,he is also in possession of excellentpositive reasons to accept her
testimony.For, Henry has no reason to doubt that a deer was in facton the
hiking trail in question, nor does he have counterevidenceof any kind for
Bertha's testimony.Moreover,because beliefsthatare formedon the basis of
Bertha's statementsnearly always turn out to be true, Henry has acquired
excellentinductiveevidence forbelievingherto be one of the most trustwor-
thypeople to consult on a wide rangeof topics. So, when Bertha's extreme
reliabilityas a testifieris combined withhis veryhigh degreeof justification
foracceptingher report,thereis simply no reason to deny that Henry's true
belief that therewas a deeron a nearby hiking trail qualifies as warranted,
justified,and known.Thus, in CONSISTENT LIAR, we finda clear case in
whicha hearernotonlyacquires warrantedand justifiedbelief on the basis of
testimonythatis insincere,but also comes to have testimonialknowledgeof
the propositionin question. Furthermore, notice thatBertha fails to possess a
belief withthe epistemicpropertiesin question, yet Henry formsa belief on
the basis of her testimonythatdoes possess these epistemic properties.Here,
then,we finda case of testimonialknowledge {warrant,justification) with-
out testimonialtransmission?2

21
For a compellingdiscussionof thedistinctionbetweenaccidentaltruthand accidental
see Reed (2000).
justification,
For thosewho findsome of thedetailsof CONSISTENT LIAR a bit too fanciful,the
as a counterexample
followingcase can be substituted to TEP-N:
DOUBTING THOMAS: Thomasis a scientist conducting researchon the migratory
patternsof graywhales and, thoughhe is brilliant,he is crippledby self-doubt.
Afteruncovering evidencethatthe whales' patternof migration is actuallyimpor-
fromwhatscientific
tantlydifferent consensushad thought, his doubtspreventhim
fromcomingto believe in thetruth of thisevidence.Nevertheless, his intellectual
leads himto feel an obligationto presenthis trueand reliablyacquired
integrity
researchto his colleagues,mostof whomreadilyaccept his testimony about the

LEARNING FROM WORDS 85

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 The upshotof theseconsiderations is thatthestatements of speakersare
notonlythebasis,bothcausallyand epistemically, of the beliefsthathear-
ers acquire via testimony, they are also the bearers of epistemicsignifi-
cance. For as CONSISTENT LIAR reveals,a speakerdoesn'tevenneedto
believethestatement she is reportingin orderfora hearerto acquirea belief
thatis warranted,justified, or known on the basis of hertestimony. This
shows thatin orderto have a unifiedepistemology of testimony, the epis-
temicconditions fortestimonial beliefsneedtobe imposedon thestatements
ofspeakers,noton theirbeliefs.23 Hence,TEP-N is false.24

whales' migratory patterns. (Thanks to David Bullerfor suggestingthistypeof

case.)
Now, even thoughThomas's self-doubtpreventshim frombeing a reliable believer
regarding themigratory patterns of graywhales,he is nonethelessa highlyreliabletesti-
fierof thisinformation and so his colleaguescan acquire knowledge(warrantedbelief,
justifiedbelief)via histestimony. Hence,as withCONSISTENT LIAR, we have a case
in whicha heareracquiresknowledge(warranted belief,justifiedbelief) on the basis of
thetestimony of a speakerwhodoes notbelieveand hence does notknow(believe with
warrant, justifiedly believe) the proposition to whichhe is testifying, therebyshowing
TEP-N to be false.
For additionalcounterexamplesto TEP-N like DOUBTING THOMAS, though
employedfordifferent purposes,see my(forthcoming). See also my(1999) forcounter-
examples to TEP-N thatinvolvespeakers whose own beliefsare defeatedbut who
nonethelessreliablyconvey information to hearers withouttherebyconveyingtheir
defeaters.
It shouldbe emphasizedthatitis notopento theproponent oftheBVT to argueon behalf
of a so-calleddisjunctivist viewof testimony in whichbeliefsare thebearersof epistemic
propertiesin cases wheretestimonial knowledgeis successfullyacquired, while state-
mentsare thebearersof epistemicproperties in thosecases where it is not.For notice:
testimonial knowledgeis acquiredin CONSISTENT LIAR. Hence,a proponent of a dis-
junctiveview of testimony would have to endorse a completelyarbitrarydivision,
accordingto whichbeliefsare thebearersof therelevantepistemicproperties in all of
thosecases thatdo notpose a problemforthe BVT whilestatements are the bearersof
thesepropertiesin cases thatare counterexamples to the BVT. Butthisview beginsto
looksuspiciously ad hoc. Itis muchlikean epistemologist responding to theGettierprob-
lembyinsisting thatknowledge is justifiedtruebeliefin non-Gettier cases, whileadmit-
tingthatitis morethanthisinGettier cases.
24
It is of interestto notethatevenin thoserarecases whereithas beenarguedthatbeliefis
notthecentraliteminvolvedin a testimonial exchange,itis simplyreplaced withanother
mentalor cognitive itemrather thanwitha linguistic or communicative item,as I am here
suggesting. Forinstance,in his (1994), Welbourneargues againstcertainstrandsof the
BVT on thegroundsthatwhilebeliefsarenotappropriate itemsfortransmission, statesof
knowledgeare.Thus,he claimsthat"...our conceptof knowledge, unlikeour conceptof
belief,is a conceptof something whichis essentiallytransmissible" (1994, p. 309). Of
course,sincethearguments offered in thissectionsubsumea knowledgeversionof TEP-
N, movingfrombeliefto knowledgeleadsus in thewrongdirection. Butevenif we were
to bracketthispointand accept Welbourne' s proposal,we wouldstillbe leftwiththe
problemof characterizing howbeliefsthatfallshortof knowledgebutnevertheless pos-
sess warrant andjustification are acquiredvia testimony.

86 JENNIFERLACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 2. The Transmissionof EpistemicProperties:Sufficiency
In this section,I take up the remainingthesis of TEP- the sufficiency
dimension. I shallarguethatspeaker-warrant (justification,knowledge) is not
sufficient forhearer-warrant (justification, knowledge), and hence that the
picturedominating theepistemology of testimony, in whichepistemicprop-
ertiesaretransmitted froma speaker'sbelieftoa hearer's,is thoroughly mis-
guided.
Beforeproceeding, however,one feature of TEP-S thatis important to
noticeis the no-defeater clause (condition(3)). For, on this view, even if
thereare some cases in whichspeaker-warrant (justification,knowledge)is
sufficientforhearer- warrant(justification, knowledge), thebelief thata hearer
acquires from a speaker does not alwayspossess therelevant epistemicprop-
ertyto thesame degreeas thespeaker's - a hearermayhave a defeater for
believingtheproposition in questionthatthespeakersimplydoes not pos-
sess. Therearetwo different kindsof defeaters thatare hererelevant.First,
thereare what we mightcall psychologicaldefeaters. A psychological
defeateris an experience, doubt,or beliefthatis hadby S, yetindicatesthat
S's beliefthatp is eitherfalseor unreliably formed or sustained.25Defeaters
in thissensefunction by virtue of being had by S, regardless of theirtruth
value orjustificatory status.26 Second,thereare whatwe mightcall norma-
tivedefeaters. A normative defeater is a doubtor beliefthatS oughtto have,
yet indicates thatS's beliefthat p eitherfalseor unreliably
is formed or sus-
Defeatersinthissensefunction
tained.27 by virtue of beingdoubtsor beliefs

25
To be even more precise, thereare two different kindsof psychologicaldefeaters:
rebuttingdefeatersare thosethatindicatethatthetargetbeliefis false whileundercutting
defeatersare thosethatindicatethatthetargetbeliefis unreliably formedor sustained.
See Pollock(1986) for furtherdevelopmentof the distinction betweenrebutting and
undercutting defeaters.
Forvariousdiscussionsof whatI call psychologicaldefeaterssee, forexample,BonJour
(1980 and 1985), Nozick (1981), Pollock(1986), Goldman(1986), Plantinga(1993),
Lackey(1999, 2003, 2005a, 20056, and 20066), Bergmann(1997 and 2004), and Reed
(forthcoming).
Followingthe distinction in note 25, thereare rebutting and undercutting normative
defeaters.The centraldifferenceis thatwhile psychologicaldefeatersare doubtsor
beliefshad by thesubject,theirnormative counterparts are doubtsor beliefs thatthe
subjectshouldhave. Moreover,psychologicaland normative defeatersmaythemselves
be eitherdefeatedor undefeated. For instance,supposethatDaphne acquires a psycho-
logicaldefeaterforher beliefthatthereis bamboogrowingin her backyardwhenshe
acceptsClifford'stestimony thatbamboodoesn'tgrowin California.Butsupposefurther
thatDaphne laterlearnsthatClifford compulsivelylies aboutCaliforaianplantlife. In
such a case, Daphneacquiresa psychological defeaterforthe beliefthatshe formedon
thebasisofClifford'stestimony, therebyproviding her witha defeater-defeater forher
originalbeliefthatbamboois growingin her backyard.And, as mightbe suspected,
defeater-defeaterscan be defeatedby further experiences,doubts,beliefs,and reasons,
which,in turn,can be defeatedby further experiences,doubts,beliefs,and reasons,and
so on. Now,whenone has a defeaterD forone's beliefthatp thatis notitselfdefeated,
one has whatis calledan undefeated defeaterforone's beliefthatp. As specifiedin con-

LEARNINGFROMWORDS 87

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
thatS should have (whetheror not S does have them) given the presenceof
certainavailable evidence.28
However, even though the presence of a defeatermay rendera hearer's
beliefless warrantedthanthespeaker's fromwhom it was acquired,this does
not threatenthe spiritof TEP-S. For what happens in such a case is that the
speaker's belief thatp, along with its warrant,is transmittedto the hearer,
butthewarrantforthehearer'sbeliefthatp is thendefeated.What would con-
tradictthe spiritof TEP-S would be a case in which a hearerbelieves thatp
purelyon the basis of a speaker's belief that p, possesses no defeatersfor
believing thatp, and yet still has a significantlydifferent epistemic relation
to p thanthe speaker.29This is preciselythe sortof case I shall provide.
To begin,considerthe following:

COMPULSIVELY TRUSTING: Bill is a compulsively trusting

person with respect the testimonyof his neighbor,Jill, in whom
to
he has an obsessive romanticinterest.Not only does he always trust
Jillwhen he has verygood reason to believe her, he is incapable of
distrustingherwhen he has verygood reason to not believe her. For

dition(3), itis the presenceof undefeateddefeaters,notmerelydefeaters,thatis rele-

vanttoTEP-S.
For discussionsinvolving whatI call normative defeaters,approachedin a numberof
different ways,see BonJour (1980 and 1985),Goldman(1986), Fricker(1987 and 1994),
Chisholm(1989),Burge(1993 and 1997), McDowell (1994), Audi (1997 and 1998),
Lackey (1999, 2003, 2005a, 20056, and 20066), Williams(1999), BonJourand Sosa
(2003), Hawthorne (2004), and Reed (forthcoming). Whatall of thesediscussionshave
in commonis simplytheidea thatevidencecan defeatknowledge(warrant,justification)
even whenthesubjectdoes notformany corresponding doubtsor beliefsfromtheevi-
dencein question.
In additionto psychologicaland normative defeaters,thereare whatwe mightcall
factualdefeaters.A factualdefeater is a trueproposition,D, suchthatifD wereadded to
S's beliefsystem, thenS wouldno longerbejustifiedin believingthatp. Defeatersin this
sensefunction byvirtueof beingtrue.For instance,you maycorrectlybelieve thatyou
saw a coyotein JoshuaTree NationalParkand yetthefactthatthereare dogs thatyou
wouldnotbe able to distinguish fromcoyotesthatfrequently visitthisparkmaynonethe-
less preventsucha belieffrombeingan instanceof knowledge.In particular, thatthere
are dogsindistinguishable toyoufromcoyotestliatfrequent theparkin whichyou saw the
real coyoteis a trueproposition which,if added to yourbeliefsystem,wouldresultin
yourbeliefbeingunjustified. (See, forinstance,(Lehrer(1965 and 1974), Lehrer and
Paxson(1969), Klein (1971, 1976, 1979,and 1980), Sosa (1974 and 1979), and Swain
(1981).) In whatfollows,however,I shallfocusonlyon psychologicaland normative
defeaterssince theseare thekindsof defeatersgenerallytakenby proponents of the
BVT to be incompatible withtestimonial knowledge.In contrast, theattempt to rule out
Gettier-type cases via thisthirdclass of defeaters is notendorsedby all such proponents
of theBVT.
29 In his (2000), PeterGrahamproposestwo different to a versionof
counterexamples
TEP-S, bothof whichcan be interpreted as involvingthepresenceof factualdefeaters.
In contrast, whatI shallprovidearecounterexamples to TEP-S thatdo notdependin any
way on defeaters,therebyshowingthatthe entirepictureof testimony as involving
transmission is fundamentally incorrect.

88 JENNIFERLACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 instance,even when he has available to him overwhelmingevidence
forbelieving that she is deliberatelylying or being deceitful,Bill
cannot come to believe thisabout Jill.Indeed, Bill is such that there
is no amountof evidence thatwould convince him to not trustJill.
Yesterday,while taking his afternoonwalk, Bill ran into Jill, and
she told him that she had seen an orca whale while boating earlier
that day. Bill, of course, readilyaccepted Jill's testimony. It turns
out that Jill did in factsee an orca whale on the boat trip in ques-
tion,thatshe is veryreliable withrespectto her epistemic practices,
both in generaland in this particularinstance,and that Bill has no
reason to doubt the proffered testimony.Given his compulsively
trustingnature withrespectto Jill,however,even if he had had mas-
sive amounts of evidenceavailable to him indicating,for instance,
thatJilldid not see an orca whale, thatshe is an unreliable epistemic
agent,thatshe is an unreliabletestifier,thatorca whales do not live
in thispartof the country,and so on, Bill would have just as readily
accepted Jill's testimony.30

It is, of course, clear that Jill's belief about the orca whale possesses all of
the epistemic propertiesin question- she is a reliable epistemic agent, both
in generaland in theparticularcase at issue, and she did, in fact,see an orca
whale while boating yesterday.31 The crucial question for our purposes is
whetherBill knows,or is warranted(justified)in believing, that therewas an
orca whale in therelevantbody of wateron the basis of Jill's testimony.And
here,theanswershould clearlybe no.
To see this, notice that because of his compulsively good naturewith
respectto Jill's testimony,Bill is simply incapable of being sensitive to the
presenceof defeatersregardingherreports.In thisrespect,he is no betterepis-
temicallythana subject who has been brainwashedor programmedto accept
any reportthat Jill makes. For were Bill to be inundatedwith massive
amountsof counterevidence, he would have acceptedJill's testimonyjust as
readilyas he did in thecompleteabsence of such counterevidence.Indeed,Bill
is such that he would have accepted Jill's testimonyunder any circum-
stances?1 Because of this, Bill's belief that therewas an orca whale in the

30 An examplethatbearssomesimilarities to COMPULSIVELY TRUSTING can be found

in my(2003). In thatpaper,however,I have a different - thereI argue specifi-
target
callyagainststandardcharacterizations of non-reductionism
in theepistemology of testi-
mony,nottheBVT.
Moreover,since thereis nothing unusualaboutJill'sexperienceof theorca whale,we
can assumethatshe has excellentreasonsfor formingthe corresponding belief. This
avoidstheobjectionthatJillpossessestherelevantknowledgeonlyif an externalist view
ofepistemic justificationis assumed.
In thissense,Bill's epistemic toJillis importantly
relationship different
fromtheepistemic
a childbearsto her parentsor teachers.For whilechildrenmaybe trusting
relationship

LEARNING FROM WORDS 89

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
relevantbodyof wateris evidentiallyinsensitivein a way thatis clearly
incompatible withwarrant, justification,and knowledge.33 Therefore, while
Jill's beliefpossessesall of theepistemicproperties in question,the belief
thatBill formson thebasisof hertestimony possessesnoneof them.Hence,
TEP-S is false.
WhatCOMPULSIVELY TRUSTING revealsis thatwhilea speakermay
be botha perfectly reliablebelieverandtestifier, a hearermaybe so consti-
tutedas topreventtheepistemic properties ofa speaker's belieffrombeing
transmitted to her.This has significant consequences fortheBVT. For pro-
ponents of the BVT portray the transmission of beliefs as being muchlike
thegivingof gifts:ifI giveyoua wrappedbox,all youhave to do to receive
thegiftis to receivethewrappedbox- thegift,so to speak,comesalong for
theride.Similarly, itis thought thatifI transmit to youmybeliefthat/?,all
you have to do to receive the epistemicproperties it possessesis to receive
-
thebelief theepistemicproperties itpossesses,so to speak,come along for
theride.But as we saw in COMPULSIVELY TRUSTING, beliefsare not
connectedwiththeirepistemicproperties the way thatgiftsare connected
withthewrappedboxes thatcontainthem.A hearermayacquirea beliefon
thebasisof a speaker'stestimony, andyetnotacquiretheepistemicproper-
tiespossessedbythespeaker'sbelief.Moreover,thiscan happenwhereit is
notexplainableby appealto thedefeatof epistemicproperties thatwerein
facttransmitted. For Bill's compulsively trusting naturedoes not makehim
suchthatepistemicproperties aretransmitted to himvia Jill's testimony,yet
immediately defeated;he is such that epistemicproperties cannot even be
transmitted to him in thefirstplace?* Hence, the pictureof testimony
paintedby proponents of theBVT, in whichbeliefsare passedalong with
theirepistemicproperties froma speakerto a hearer,is fundamentally incor-
rect.

with respect to theircaretakers,they certainlyare not compulsively trusting.For instance,

when I said to my young daughter that houses withoutchimneys do not pose a problem
for Santa Claus because childrenperhaps do not live in such houses, even at three years
old she skeptically responded, "Surely thereare some houses without chimneys that have
childrenin them!" In this sense, children, unlike Bill, certainly have the capacity to be
sensitive to defeaters with respect to the testimonyof their caretakers, even if such a
capacity is not frequentlyexercised. Moreover, if children were like Bill with respect to
the testimonyof theirparentsand teachers, then I would have no problem denying testi-
monial knowledge to them as well. (I am grateful to comments from Fred Dretske and
Ishani Maitra thatpromptedthe inclusion of this point.)
For a more detailed discussion of the natureof this incompatibility,see my (2005a).
3
Furthermore,as COMPULSIVELY TRUSTING is described, Bill, as a matter of fact,
does not have any relevant defeaters for the testimonyin question: Jillis very reliable
with respect to her epistemic practices, both in general and in this particularinstance, and
Bill has no reason to doubt the proffered testimony.The epistemic problem with Bill is
that,even if therewere defeaters, he would accept Jill's testimonyjust as readily as he
did in theircomplete absence.

90 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 Furthersupportforthisconclusion can be providedby examininga second
typeof case thatunderminesTEP-S. Consider thefollowing:

ALMOST A LIAR: Phil is compulsively-trusting-B ill's twin

brother,thoughhe is not himselfcompulsivelytrustingwith respect
to Jill's testimony.In fact, he is, froman epistemic point of view,
quite healthy:he truststhose whom he has good reason to trust - or
-
at least those whom he has no clear reason to distrust and distrusts
those whom he has good reason to distrust.Yesterday,while taking
his afternoonwalk, Phil ran intoJilljust afterBill did, and she also
told him that she had seen an orca whale while boating earlierthat
day. Phil, having acquiredverygood reasons for trustingJill over
the five years he has known her, readilyacceptedher testimony. It
turnsout thatJill did in factsee an orca whale on the boat trip in
question,thatshe is veryreliable withrespect to her epistemic prac-
tices,both in general and in thisparticularinstance,that she is gen-
erally a veryreliable testifier,and thatPhil has no reason to doubt
the profferedtestimony. However, in order to promote a whale
watchingbusiness she is in the process of starting,she would have
-
reportedto Phil- in preciselythe same manner thatshe had seen
an orca whale even if she hadn't. (Of course, she wouldn't have
believed that she had seen an orca whale if she hadn't.) Moreover,
given the patternof the whales' travelcombined with the particular
time of year it is, it is in factquite surprisingthat Jill saw an orca
whale whenand whereshe did.

There are two centralfeaturesof ALMOST A LIAR that are importantto

notice. First,because she wishes to promote her futurewhale watchingbusi-
ness, Jill would have reportedto Phil that there was an orca whale in the
relevantbody of watereven if therehadn't been one. And second, given the
schedule of the whales' travelcoupled with the time of year it is, it was in
fact much more likely forJill not to have seen an orca whale when she did
thanforherto have seen one.
The combinationof these two featureshas the following result: thereare
nearbypossible worlds in which an orca whale was not in the relevantbody
of water,Jillnonethelessreportsthattherewas, and Phil, being in the same
evidentialsituationas he is in the actual world,readilyformsthe correspond-
ing belief on the basis of Jill's testimony.More generally, Phil's belief
about therehavingbeen an orca whale in the relevantbody of wateris coun-
ter/actuallyinsensitiveto the truth.For instance, his belief is neithersensi-
tive (a la Nozick) norsafe (a la Sosa): ifp were false,Phil would still believe
thatp, and Phil would believe thatp withoutit being so thatp. Otherwise
put,in nearbyworlds whereit is false that an orca whale was in the relevant

LEARNING FROM WORDS 91

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
body of water,Phil believes that therewas one, and thereare nearbyworlds
wherePhil believes that an orca whale was in the relevantbody of water
withoutthis being so. Given this,the truebelief thatPhil formson the basis
of Jill's testimonynot only fails to qualifyas knowledge,it also fails to be
warrantedor justified.35Thus, while Jill's belief possesses all of the epis-
temic propertiesin question, the belief thatPhil formson the basis of her
sinceretestimonypossesses none of them. Hence, once again, we see that
TEP-S is false.36
To fullyappreciatethe consequences of ALMOST A LIAR, it will be
helpfulto compare it withCONSISTENT LIAR fromSection 1. Notice first
that while CONSISTENT LIAR shows that an unreliable believer may
nonethelessbe a reliable testifier,ALMOST A LIAR reveals that a reliable
believer may nonetheless be an unreliable testifier. This is why Henry
acquires knowledge(warrantedbelief,justifiedbelief) on the basis of Bertha's
testimony,despite the factthat she fails to have the knowledge (warranted
belief,justifiedbelief) in question, and why Phil fails to acquire knowledge
(warrantedbelief,justifiedbelief)on thebasis of Jill's testimony,despite the
factthat she possesses the knowledge (warrantedbelief, justified belief) in
question. For, recall that while Bertha's belief is neithersensitive nor safe,
the statementthatshe offersto Henry is both sensitive and safe. In contrast,

35
I should again emphasize, as I did in note 8, that I am concerned with the epistemic prop-
erties in question only insofaras theybear an intimateconnection with knowledge. Thus,
theremay be some subjective notions of these propertiesthatPhil does plausibly possess.
3 TRUSTING or
For those who have reservations about either COMPULSIVELY
ALMOST A LIAR in isolation, we can simply combine the features from both so as to
have COMPULSIVELY TRUSTING ALMOST A LIAR. This would simply involve
imagining Bill in Phil's scenario, so thatwe have a hearer who is both evidentially insen-
sitive and counterfactuallyinsensitive to the truth.Given this combination, both reduc-
tionistsand non-reductionistsin particular, and externalists and internalistsin general,
should agree thatthe epistemic propertiesin question fail to be acquired by the hearer.
Moreover, the following represents a thirdkind of counterexample to TEP-S that
should be generally persuasive:
A LUCKY CHOICE: Upon arrivingin Chicago forthe firsttime,Alvin asks the first
passerby thathe sees, Zoe, for directionsto the Sears Tower and she reports that it
is six blocks east. While Zoe knows thatthis is the case, and Alvin has no reason to
doubt eitherher credibilityas a speaker or the truthof the proposition to which she
is testifying,she is the only reliable speaker in this part of Chicago, completely sur-
rounded by incompetentsand liars. Because of this,that Alvin chooses a reliable
testifierwho correctly points him in the direction of the Sears Tower is entirelya
matterof good luck.
Now, even thoughZoe knows thatthe Sears Tower is six blocks east, and Alvin does not
possess any relevant defeatersfor the reportin question, A LUCKY CHOICE represents
a testimonialGettier-typecase for the recipientof testimony.In particular, Alvin's luck-
ily choosing the only reliable testifieramong the surroundingincompetentsand liars is
analogous to a perceiver luckily seeing the only real barn among surrounding barn
facades. Hence, Alvin does not come to know thatthe Sears Tower is six blocks east on
the basis of Zoe' s testimony,despite the fact that conditions (l)-(3) are satisfied. Once
again, we see thatTEP-S is false.

92 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
Jill's orca-whale-belief
is both sensitiveand safe,but the orca-whale-state-
mentthatshe offers to Phil is neither.
In particular,
thoughJill would not
believethatp ifp werefalse,she wouldstatethatp if p werefalse,and
thoughJillwouldnot believethatp withoutit beingso thatp, she would
statethatp withoutit beingso thatp. Thus,just as Berthacan give epis-
temicpropertiesto othersthatshe failsto possessherself,
Jillcan failto give
epistemicpropertiestoothersthatshepossessesherself (evenwhentestifying
sincerely).Bothcases, therefore,supportthesame generalconclusion:state-
mentsof speakers,nottheirbeliefs,arewhatmatterforan epistemology of
testimony.

3. The SVT
We haveseen thatboththesesof TEP arefalse and, accordingly, thateach
component of the BVT is falseas well. We are now in a position to also see
thatthereis a counterpart versionof each componentthat is true - one
involvingthestatements of speakers.Let us call this alternative familyof
thesestheStatement ViewofTestimony (hereafter,theSVT).
According to theSVT, theprocessof communicating via testimony does
notinvolvea speakertransmitting herbeliefto a hearer, along with the epis-
temicproperties itpossesses.Instead,a speakeroffers a statement to a hearer,
alongwiththeepistemic properties itpossesses,anda hearerformsthecorre-
sponding beliefon the basis of understandingandacceptingthestatement in
question. Statements arenot,therefore, merelyvehicles for expressing beliefs
but,rather,theyarethecentralbearersofepistemic significance themselves.
Thereareat leastthreesignificant consequences of rejectingthe BVT in
favorof theSVT. First,becauseproponents of theBVT portray thetestimo-
nial exchangeas involvingmerelytransmission, testimonyis said to be
incapableofgenerating newepistemicfeatures forbeliefs - it merelyhas the
capacitytotransmit fromone personto another beliefsthathavealreadybeen
rendered warranted, or
justified, an instanceof knowledge via anothersource.
As AlvinPlantingasays,"...testimonialwarrant, like water,risesno higher
thanitssource...ifyou tell me something andI believeit on yoursay-so,I
havewarrant foritonlyifyoudo."37Butas we saw in CONSISTENT LIAR,
this is not so: a hearercan acquirea beliefthatis warranted, justified,and
knownfroma speakerwhoseownbeliefpossessesnoneof therelevantepis-
temicproperties. Thus,notonlycan testimony function as a generativeepis-
temic source, but testimonialwarrant(justification, knowledge) - unlike
water - can risehigherthanitssource.38

37
Plantinga ( 1993, p. 84 and p. 87).
For arguments leading to a similar conclusion with respect to the epistemic status of
memory,see my (2005/?).

LEARNING FROM WORDS 93

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 Second, and closely related, proponents of the BVT maintain
that- defeatersaside- testimonialwarrant(justification,knowledge)cannot
fail to rise as high as its source. We have seen, however,that this is false as
well: a hearer'sbelief may fail to be as warranted(justified,known) as its
source, either for reasons having to do with the hearer - as
COMPULSIVELY TRUSTING revealed- or forreasons having to do with
thespeaker- as ALMOST A LIAR revealed.
Third,it is nearlyreceived wisdom in the epistemologyof testimonythat
thereare two centralcriteriathat need to be fulfilledin orderfora speaker's
testimonyto be epistemicallyacceptable: "...the sinceritydimension, con-
cerningtheattester'shonesty,and the competencedimension,concerningthe
attester'shavingexperienceor knowledgesufficientto make it at least likely
thatif the attesterformsa belief thatp, thenp is true."39Otherwiseput, in
orderfora hearerto acquireknowledge(warrantedbelief,justified belief) on
the basis of a speaker's testimony,the speaker must be both a competent
believer and a sincere testifier. As we saw in CONSISTENT LIAR, how-
ever, neithera speaker's competentbelieving nor her sincere testifyingis
necessary for the acquisition of testimonial knowledge (warrantedbelief,
justified belief). And, as we saw in COMPULSIVELY TRUSTING and
ALMOST A LIAR, a speaker's competentbelieving combined with her sin-
cere testifyingis not sufficientforthe acquisition of testimonialknowledge
(warrantedbelief,justifiedbelief).
What really mattersforthe epistemic status of testimonyis whetherthe
speaker is a competent testifier, where this is understoodin terms of the
reliabilityof the statementin question. More precisely, according to the
SVT, TEP-N should be replacedwith the following Reliability of the State-
(RS-N) thesis:
ment-Necessity

RS-N: For everyspeaker, A, and hearer,B, B's belief that p is

warranted(justified,known) on the basis of A's testimony
that/?only if A's statementthatp is reliable or otherwise
truth-conducive.40

39 in note6.
Audi (1997, p. 409). See also thereferences
For thosewho findthecases in note 11 compelling, RS-N maybe substituted withthe
weaker:
RS-N*: For everyspeaker,A, and hearer,B, B's beliefthatp is warranted(justi-
fied,known)on thebasisof A's testimony onlyif (i) A's statementis reli-
able or otherwise and (ii) B's beliefthatp is appropriately
truth-conducive,
connectedwiththecontentof A's statement.
Of course,ifRS-N is replacedwithRS-N*,morewillneedto said aboutthe"appropriate
connection" foundin condition (ii).

94 JENNIFERLACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
The reliabilityof the statementin question can, in turn,be fleshed out in any
number of ways. 41 For instance, it may be necessary that the speaker's
statementbe sensitive,safe, properlyor virtuouslyformed,and so on.42What
is of importforour purposes, however,is that being a competentbeliever
and a sinceretestifierare epistemicallyrelevantonly insofaras they bear on
the speaker's capacityto be a competenttestifier.43

41 It maybe objectedthatrequiring thereliability

of thestatement in questionis too stronga
condition fortestimonial justification. Forinstance,wouldn'tI bejustifiedin believingthat
thePresident in Iraqon thebasisof readingtheNew YorkTimeseven if,for
is currently
somereason,the reporter'sstatement failsto be reliableor otherwisetruthconducive?
WhatI wouldsay is first, as mentioned in notes8 and 35, 1am concernedwithtestimonial
justification onlyinsofaras ithas an intimate connectionwithtestimonial knowledge.So,
theremay be some subjectivenotionsof testimonial justificationthatdo notrequirethe
reliability ofthestatement in questions.Second,theconceptionof justification intimately
connectedwithknowledgethatI favor has two essentialaspects: (i) a reliability, or
objective,componentand (ii) a rationality, or subjective,component.(I argue for this
view of testimonial justification in my(2006b).) Given this,mybelief about the Presi-
dent'swhereabouts clearlysatisfiesthesubjective,rationality constraint,despiteitsfail-
ureof theobjective,reliability constraint. Moreprecisely, eventhoughsuch a belieffails
(i) and is therefore notjustified, mysatisfaction of (ii) showsthatI am nonethelesssub-
jectivelyrationalin myacceptance of thereporter'stestimony. (Indeed, such a belief
possessesmanyotherpositiveepistemicproperties, suchas beingheldin an epistemically
responsible way,beingepistemically virtuous,and so on.) This enables us to explainthe
intuition thatthe belief in questionbothpossesses and lacks something epistemically
important: it possessessubjectiverationality, butlacks objectivereliableformation and,
hence,justification. (Alternatively, itcouldbe arguedthatmybeliefaboutthe President's
whereabouts is subjectively
justified, butnotobjectively justified.For moreon thedistinc-
tionbetweensubjectiveand objectivejustification, see BonJour and Sosa (2003, pp. 153-
5). Itis of further interestto notethatit is notat all uncommonin the literature to find
justification beingdiscussed,eitherentirely or partially, terms.To namejust
in externalist
a few,see Alston(1989), Goldman(1992), and BonJour and Sosa (2003).)
42 See Nozick(1981), Sosa (1996, 1999,2000,and 2002) and Williamson(2000), Plantinga
(1993), and Sosa (1991), respectively.
Of course,oftentimes,it is preciselybecause a speakeris insincereor an incompetent
believerthatshe is an incompetent or unreliabletestifier.For instance,if I frequentlylie
or forminaccuratebeliefs,moreoftenthannot thiswill preventyou fromacquiring
knowledge(warranted belief,justifiedbelief)on the basis of mytestimony. Buttherea-
son whyyouareso prevented is thatmyinsincerity or incompetence has made mytesti-
monyunreliable.Moreover,a hearer's beliefsabouta speaker's sincerity and compe-
tence can have epistemicsignificance.For instance,if I believe that you are a
compulsiveliaror an unreliable epistemicsource,theneven if you are neitherof these,
themerefactthatI believe thatyou are can provideme witha defeaterforaccepting
yourtestimony. Hence,mybeliefsaboutyoursincerity and competencecan preventme
fromacquiringknowledge(warrantedbelief,justifiedbelief)on thebasis of yourtesti-
mony.
Butwhatif statements themselvescannotbe understood apart fromthe beliefsof
speakers?Will myview stillbe importantly differentthanthe BVT? (I am gratefulto
Kent Bach, Michael Glanzberg,and GurpreetRattanforpressingthispoint.)Yes. For
evenifstatements cannotbe understood apartfromsome of thebeliefsof therelevant
speakers,all I need formypurposesis thatsuch speakersdo not need to believe the
particularpropositions In particular,
to whichtheyaretestifying. whatI need formyRS-
N to be importantly fromthe BVT is the falsityof TEP-N. And thereis no
different

LEARNING FROM WORDS 95

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 Now, as withTEP-N, RS-N expresses only a necessaryconditionfor tes-
timonial warrant(justification,knowledge).A complete view of the episte-
mology of testimonywill, then,requirefurtherconditions,such as (2) and (3)
fromTEP-S. Thus, at a minimum,the SVT will include the following:

SVT: For everyspeaker, A, and hearer,B, B's belief that p is

warranted(justified,known44)on the basis of A's testi-
mony thatp only if (1) A's statementthatp is reliable or
otherwisetruth-conducive, (2) B comes to believe thatp on
the basis of the contentof A's statementthatp, and (3) B
has no undefeateddefeatersforbelievingthat/?.45

It is not my purpose here to providea fullycomplete account of testimonial

warrant(justification,knowledge),and so I shall not defendSVT as such.46
Rather,SVT providestheframeworkfor a complete epistemologyof testi-
mony,one thatplaces the statementsof speakersat the centerof the acquisi-
tionof testimonialknowledge.47

reason to thinkthatplausible accounts of statementswill need to depend on the truthof

TEP-N.
Of course, for testimonialknowledge, a condition will need to be added to SVT requiring
the truthof B's belief thatp.
Again, following notes 11 and 40, SVT may be substitutedwith the weaker:
SVT*: For every speaker, A, and hearer,B, B's belief thatp is warranted (justified,
known) on the basis of A's testimonyonly if (1) A's statementis reliable or
otherwise truth-conducive,(2) B comes to believe thatp on the basis of an
appropriateconnection with the content of A's statement,and (3) B has no
undefeated defeaters for believing thatp.
As mentionedin note 40, more will need to be said about the "appropriate connection" in
(2) if SVT is replaced with SVT*.
In fact, in other work, I have argued for conditions on testimonialwarrant beyond those
found in (l)-(3). See, for instance, my (2003 and 2006fc).
What if the proponent of the BVT were to argue that while the SVT may be true of a
broader category of testimonialknowledge, the BVT nonetheless accurately captures a
narrowercategory of testimonialknowledge, such as knowledge thatwe acquire through
transmissionof belief or via trustin testimony?(For this type of response, see Fricker
(forthcoming). I am also grateful to Mark Heller, Brian McLaughlin, and, especially,
David Sosa, for pressing this point.) By way of response to this objection, notice that in
the absence of any reason to think that this narrower category of testimonialknowledge
is of epistemic interest,it is unclear why such a distinctionshould be drawn in the episte-
mology of testimony.For my view of testimonyprovides an account thatsubsumes both of
these categories of knowledge. Moreover, my view has the resources to explain trans-
mission (if or when such a process actually occurs): the epistemic properties of a
speaker's belief is transmittedvia her statement(or other act of communication) to the
recipientof her testimony.Thus, transmissionis epistemically importantonly insofar as it
bears on the satisfactionof RS-N. (This is evidenced by the fact that testimonialknowl-
edge can be acquired even after the counterfactual removal of a speaker's belief, so
long as the statementin question remains reliable.) What would a theory of testimonyin
the narrowersense add to this? Proponentsof this approach do not say. And why would
anythingneed to be added? We already have a theory- the SVT - that explains the
epistemic propertiesof both of these categories of testimonialknowledge.

96 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 - we
speaking,we do not learnfromone another'sbeliefs
Hence,strictly
learnfromone another'swords**A failureto appreciate in
thishas resulted
the widespreadacceptanceof a pictureof testimonythat is fundamentally
Thus, in orderto makegenuineprogressin the epistemology
incorrect. of
testimony, we need to stop lookingat what speakersbelieve and focus,
instead,on whatspeakerssay.49

References
Adler,Jonathan E. (2002) Beliefs OwnEthics.Cambridge, MA: The MIT
Press.
Alston,WilliamP. (1989) EpistemicJustification: Essays in the Theoryof
Knowledge. Ithaca:Cornell University Press.
Audi,Robert.(1997) "The Place of Testimonyin theFabricof Knowledge
andJustification."AmericanPhilosophicalQuarterly 34: 405-22.
. (1998) Epistemology: A Contemporary Introduction to the Theory
ofKnowledge.London:Routledge.
. (2004) "The A prioriAuthority of Testimony," in ErnestSosa and
Enrique Villanueva (eds.), Philosophical Issues,14: 18-34.
. (2006) "Testimony, Credulity, andVeracity," in Jennifer Lackeyand
ErnestSosa (eds.), The Epistemology of Testimony.Oxford:Oxford
University Press:25-49.
Austin,J.L. (1979) "OtherMinds,"in his PhilosophicalPapers, 3rdedn.
Oxford:OxfordUniversity Press.
Bergmann, Michael. (1997) "Internalism, Externalism and the No-Defeater
Condition."Synthese110: 399-417.
. (2004) "EpistemicCircularity: MalignantandBenign."Philosophy
and Phenomenological Research 69: 709-27.
BonJour, Laurence.(1980) "Externalist Theoriesof EpistemicJustification."
MidwestStudiesin Philosophy5: 53-73.
. (1985) The Structureof EmpiricalKnowledge.Cambridge,MA:
HarvardUniversity Press.

48
As mentionedin note 9, it would be more precise to say thatwe learn from one another's
acts of communication (so as to allow for learning from others in non-linguisticways,
such as througha nod or pointing).
49
For veryhelpfulcommentson previous draftsof this paper, I am grateful to Kent Bach,
Mike Bishop, David Buller, Liz Camp, Fred Dretske, Jeremy Fantl, Richard Fumerton,
Michael Glanzberg, Sandy Goldberg, Peter Graham, John Greco, Liz Harman, John
Hawthorne, Mark Heller, Tom Kapitan, Jason Kawall, Matt McGrath, Brian McLaugh-
lin, Ishani Maitra, Gurpreet Rattan, David Sosa, Ernie Sosa, and audience members at
Northern Illinois University,the Universityof Iowa, the 2005 Central Division of the
American Philosophical Association, the 2005 Rutgers Epistemology Conference, and the
2005 Bellingham Summer Philosophy Conference. Most of all, I am indebted to Baron
Reed for, among countless other things,being the most stimulatingand invaluable phi-
losophical testifierI have ever met.

LEARNING FROM WORDS 97

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
BonJour, LaurenceandErnestSosa. (2003) EpistemicJustification: Internal-
ismvs. Externalism, Foundationsvs. Virtues.Oxford:BlackwellPub-
lishing.
Burge,Tyler. (1993) "ContentPreservation." The PhilosophicalReview
102: 457-88.
. (1997) "Interlocution, Perception,and Memory."Philosophical
Studies86: 21-47.
Chakrabarti, Arindam.(1994) "Tellingas LettingKnow,"in Bimal Krishna
Matilal and ArindamChakrabarti(eds.), Knowingfrom Words.
Dordrecht: KluwerAcademicPublishers: 99-124.
Chisholm,RoderickM. (1989) Theoryof Knowledge,3rdedn. Englewood
Cliffs,N.J.:Prentice-Hall.
Coady,C.A.J.(1992) Testimony: A PhilosophicalStudy.Oxford:Clarendon
Press.
. (1994) "Testimony, Observation and 'AutonomousKnowledge',"in
Bimal KrishnaMatilal and ArindamChakrabarti (eds.), Knowingfrom
Words.Dordrecht: KluwerAcademicPublishers: 225-50.
Dummett,Michael. (1994) "Testimonyand Memory,"in Bimal Krishna
Matilal and ArindamChakrabarti(eds.), Knowingfrom Words.
Dordrecht: KluwerAcademicPublishers: 251-72.
Evans, Gareth. (1982) The VarietiesofReference. Oxford:ClarendonPress.
Faulkner,Paul. (2000) "The Social Character of TestimonialKnowledge."
TheJournalofPhilosophy 97: 581-601.
Fricker,Elizabeth.(1987) "The Epistemologyof Testimony."Proceedings
of theAristotelian Society,supp.vol. 61: 57-83.
. (1994) "AgainstGullibility," in BimalKrishnaMatilalandArindam
Chakrabarti (eds.), KnowingfromWords.Dordrecht: KluwerAcademic
Publishers:125-61.
. (1995) "TellingandTrusting:Reductionism andAnti-Reductionism
in theEpistemology of Testimony." Mind104: 393-411.
. (forthcoming) "KnowledgefromTrustin Testimonyis Second-Hand
Knowledge."Philosophy and Phenomenological Research.
Gettier, Edmund.(1963) "Is Justified TrueBeliefKnowledge?'Analysis23:
121-3.
Goldman,Alvin I. (1986) Epistemology and Cognition.Cambridge,MA:
HarvardUniversity Press.
. (1992) Liaisons: PhilosophyMeets theCognitiveand Social Sci-
ences.Cambridge, MA: The MIT Press.
Graham, PeterJ. (2000) "ConveyingInformation." Synthese123: 365-92.
The
Hardwig,John.(1985) "EpistemicDependence." Journalof Philosophy
82: 335-49.

98 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
 . (1991) "The Role of Trustin Knowledge."The Journalof Philoso-
phy88: 693-708.
Hawthorne, John.(2004) Knowledgeand Lotteries.Oxford:OxfordUniver-
sity Press.
Klein,Peter.(1971) "A ProposedDefinitionof PropositionalKnowledge."
JournalofPhilosophy 68: 471-82.
. (1976) "Knowledge, Causality,and Defeasibihty.Journalof Phi-
losophy 73: 792-812.
. (1979) "Misleading'MisleadingDefeaters'." Journalof Philosophy
76: 382-6.
. (1980) "MisleadingEvidenceandthe Restoration of Justification."
Philosophical Studies 37: 81-9.
Lackey,Jennifer. (1999) "TestimonialKnowledgeand Transmission."The
PhilosophicalQuarterly 49: 471-90.
. (2003) "A MinimalExpressionof Non-Reductionism in the Epis-
of
temology Testimony." Nous 37: 706-23.
. (2005a) "Testimony and theInfant/Child Objection."Philosophical
Studies126: 163-90.
. (20056) "Memoryas a Generative EpistemicSource', Philosophy
and Phenomenological Research70: 636-58.
. (2006a) "The Natureof Testimony."Pacific PhilosophicalQuar-
terly87: 177-97.
. (20066) "It Takes Two to Tango: BeyondReductionism and Non-
Reductionism in the of
Epistemology Testimony," in JenniferLackey
andErnestSosa (eds.), The Epistemology of Testimony.Oxford:Oxford
University Press: 160-89.
. (forthcoming) "Normsof Assertion." Nous.
Lehrer,Keith.(1965) "Knowledge, Truth, Evidence."Analysis25: 168-
and
75.
. (1974) Knowledge.Oxford:OxfordUniversity Press.
Lehrer, Keith and Thomas Paxson. (1969) "Knowledge:Undefeated Justified
TrueBelief."JournalofPhilosophy 66: 225-37.
McDowell,John.(1994) "KnowledgebyHearsay,"in Bimal KrishnaMatilal
and ArindamChakrabarti (eds.), Knowingfrom Words. Dordrecht:
KluwerAcademicPublishers: 195-224.
Nozick,Robert.(1981) PhilosophicalExplanations.Cambridge,MA: The
BelknapPress.
Owens,David. (2000) Reason without Freedom:The Problemof Epistemic
Normativity. London: Routledge.
Plantinga,Alvin. (1993) Warrantand Proper Function.Oxford:Oxford
University Press.

LEARNING FROM WORDS 99

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
Pollock,John.(1986) Contemporary TheoriesofKnowledge.Totowa,N.J.:
RowmanandLittlefield.
Pritchard, Duncan.(2004) "The Epistemology of Testimony." Philosophical
Issues 14: 326-48.
Reed,Baron.(2000) "Accidental TruthandAccidental The Phi-
Justification."
losophicalQuarterly 50: 51-61 .
. (forthcoming) "EpistemicCircularity Squared?Skepticismabout
CommonSense." Philosophy and Phenomenological Research.
Reid, Thomas. (1993) The Works of Thomas Reid. Sir William Hamilton
VA: Lincoln-Rembrandt
(ed.). Charlottesville, Publishing.
Reynolds,StevenL. (2002) "Testimony, Knowledge,andEpistemicGoals."
PhilosophicalStudies110: 139-61.
Root,Michael.2001. "Hume on theVirtuesof Testimony." AmericanPhi-
losophicalQuarterly38: 19-35.
Ross, Angus.(1986) "WhyDo We BelieveWhatWe AreTold?"Ratio 28:
69-88.
Schmitt,Frederick F. (2006) "TestimonialJustification and Transindividual
Reasons,"in Jennifer LackeyandErnestSosa (eds.), The Epistemology
ofTestimony. Oxford:OxfordUniversity Press:193-224.
Shope, Robert. (1983) The AnalysisofKnowing.Princeton: Princeton Uni-
versity Press.
Sosa, Ernest.(1974) "How Do You Know?"AmericanPhilosophicalQuar-
terly11: 113-22.
. (1979) "EpistemicPresupposition," in GeorgeS. Pappas (ed.), Jus-
tificationand Knowledge:New Studies in Epistemology.Dordrecht:
Reidel:79-92.
. (1991) Knowledgein Perspective:Selected Essays in Epistemol-
ogy.Cambridge: Cambridge University Press.
. (1996) "Postscript to 'ProperFunctionalism andVirtueEpistemol-
ogy\"in JohnL. Kvanvig(ed.), Warrantin Contemporary Epistemol-
ogy.Lanham, MD: Rowman & Littlefield:271-81.
. (1999) "How Must KnowledgeBe Modally Relatedto What Is
Known?"PhilosophicalTopics26: 373-84.
. (2000) "Contextualism andSkepticism," in J. Tomberlin(ed.), Phi-
losophical Issues, supp. to Nous 34: 94-107.
. (2002) "Tracking,Competence,and Knowledge,"in Paul Moser
(ed.), TheOxfordHandbookofEpistemology . Oxford:OxfordUniversity
Press:264-87.
. (2006) "Knowledge:Instrumental and Testimonial,"in Jennifer
LackeyandErnestSosa (eds.), The Epistemology of Testimony. Oxford:
OxfordUniversity Press:116-23.

100 JENNIFER LACKEY

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions
Swain,Marshall.(1981) Reasonsand Knowledge.Ithaca,NY: CornellUni-
versity Press.
Welbourne,Michael. (1979) "The Transmission of Knowledge."The Phi-
losophical Quarterly 29: 1-9.
. (1981) "The Community of Knowledge."The PhilosophicalQuar-
terly31: 302-14.
. (1986) The Community of Knowledge.Aberdeen: Aberdeen Univer-
sity Press.
. (1994) "Testimony, KnowledgeandBelief,"in Bimal KrishnaMati-
lal andArindam
Chakrabarti (eds.), Knowing fromWords.Dordrecht: KluwerAcademicPub-
lishers:297-313.
Williams,Michael.(1999) GroundlessBelief:AnEssay on thePossibilityof
Epistemology, 2ndedn.Princeton: Princeton Press.
University
Williamson,Timothy.(1996) "KnowingandAsserting." The Philosophical
Review105: 489-523.
. (2000) Knowledgeand itsLimits.Oxford:OxfordUniversity Press.

LEARNING FROM WORDS 101

This content downloaded on Sat, 16 Mar 2013 18:35:28 PM

All use subject to JSTOR Terms and Conditions