solving the skeptical problem contents · 2017. 12. 22. · 2 those who think that hilary putnam...

55
CHAPTER 1: SOLVING THE SKEPTICAL PROBLEM CONTENTS 1. The Puzzle of Skeptical Hypotheses ......................................................................................... 1 2. Contextualist Solutions: The Basic Strategy .......................................................................... 4 3. Some Old Contextualist Solutions: Lewis’s ‘Rule of Accommodation’ ........................ 8 4. Some Old Contextualist Solutions: The ‘Relevant Alternatives’ Approach and the Rule of Relevance .................................................................................................................................. 14 5. The Subjunctive Conditionals Account (SCA) of the Plausibility of AI’s First Premise ...................................................................................................................................................... 18 6. SCA, Grandmothers, and Methods .......................................................................................... 21 7. SCA and Some Skeptical Hypotheses That Don’t Work ................................................. 23 8. SCA Confirmed ................................................................................................................................ 25 9. Nozick’s Own Solution and the Abominable Conjunction ............................................ 28 10. Strength of Epistemic Position and AI’s Second Premise ......................................... 31 11. Strength and Sensitivity ......................................................................................................... 35 12. The Rule of Sensitivity and the Beginnings of a New Contextualist Solution ... 37 13. The Rule of Sensitivity and SCA: A Comparison of Our New Solution with the Other Contextualist Solutions and with Nozick’s Solution ................................................... 40 

Upload: others

Post on 12-Dec-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

CHAPTER1:

SOLVINGTHESKEPTICALPROBLEM

CONTENTS1.  ThePuzzleofSkepticalHypotheses.........................................................................................1 

2.  ContextualistSolutions:TheBasicStrategy..........................................................................4 

3.  SomeOldContextualistSolutions:Lewis’s‘RuleofAccommodation’........................8 

4.  SomeOldContextualistSolutions:The‘RelevantAlternatives’ApproachandtheRuleofRelevance..................................................................................................................................14 

5.  TheSubjunctiveConditionalsAccount(SCA)ofthePlausibilityofAI’sFirstPremise......................................................................................................................................................18 

6.  SCA,Grandmothers,andMethods..........................................................................................21 

7.  SCAandSomeSkepticalHypothesesThatDon’tWork.................................................23 

8.  SCAConfirmed................................................................................................................................25 

9.  Nozick’sOwnSolutionandtheAbominableConjunction............................................28 

10.  StrengthofEpistemicPositionandAI’sSecondPremise.........................................31 

11.  StrengthandSensitivity.........................................................................................................35 

12.  TheRuleofSensitivityandtheBeginningsofaNewContextualistSolution...37 

13.  TheRuleofSensitivityandSCA:AComparisonofOurNewSolutionwiththeOtherContextualistSolutionsandwithNozick’sSolution...................................................40 

Page 2: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

14.  OurNewContextualistSolutionClarifiedandComparedwiththeStraightforwardSolutions..................................................................................................................41 

15.  BoldSkepticismandtheWarrantedAssertabilityManeuver.................................44 

16.  BoldSkepticismandSystematicFalsehood...................................................................48 

17.  BeggingtheQuestionAgainsttheSkeptic?.....................................................................51 

Page 3: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

CHAPTER1:

SOLVINGTHESKEPTICALPROBLEM

1. THEPUZZLEOFSKEPTICALHYPOTHESES

Manyofthemostcelebrated,intriguing,andpowerfulskepticalargumentsproceedbymeansofskepticalhypotheses.Brutallyparedtotheirbarestessentials,theyareroughlyofthefollowingform,where'O'isapropositionabouttheexternalworldonewouldordinarilythinkoneknows(e.g.,Ihavehands1)and‘H’isasuitablychosenskepticalhypothesis(e.g.,IamabodilessbraininavatwhohasbeenelectrochemicallystimulatedtohavepreciselythosesensoryexperiencesI'vehad,hencefortha'BIV'2):

1IchoosethisOpartlyforitshistoricalconnectionstoDescartes'sFirstMeditation,andalsobecauseIthinkitisanexemplarycaseofsomethingweordinarilythinkweknow.ButwhilewewouldordinarilythinkweknowthisO,we'dseldomhaveoccasiontosaythatweknowit,becausecasesinwhichsuchaclaimtoknowledgewouldbeconversationallyinorderarequiterare.(Exception:Ateacherbeginsanepistemologylecturebymatter‐of‐factlylistingvariousthingssheknows,andthatanyplausibletheoryofknowledgeshouldmakehercomeouttoknow.Inthecourseofthislisting,shesays,‘AndIknowthatIhavehands.’)Forthisandvariousrelatedreasons,somemightnotlikemychoiceofO.Suchreadersareinvitedtosupplytheirownfavoriteexemplarycasesofthingsweknowastheskeptic'starget.

2ThosewhothinkthatHilaryPutnammayhavealreadydisarmedBIV‐inspiredskepticismshouldunderstandtheBIVhypothesistobethehypothesisthatone'sbrainhasbeenrecentlyenvattedaftermanyyearsofnormalembodiment.ForevenifPutnamisrightinclaimingthatthecontentofthebeliefsoftheBIVsofhisscenarioissuchthattheseBIVsaren'tmassivelydeceived,itseemsthatrecentlyenvattedBIVsaresodeceived.

Page 4: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐2‐

The Argument from Ignorance (AI)3 1. I don’t know that not-H. 2. If I don’t know that not-H, then I don’t know that O. So, C. I don’t know that O.4

Settingasidethedistractingsideissuesthatimmediatelythreatenfromalldirections,andkeepingAIinthisstark,uncomplicatedform,Iwill,inwhatfollows,presentanddefend,atleastinbroadoutline,thecorrectsolutiontothepuzzleAIconfrontsuswith.AndAIdoespresentuswithapuzzle,because,forreasonswe'llinvestigateinlatersections,eachofitspremisesisinitiallyplausible,whenHiswellchosen.ForhoweverimprobableorevenbizarreitmayseemtosupposethatIamaBIV,italsoseemsthatIdon'tknowthatI'mnotone.HowcouldIknowsuchathing?

3AItakesitsnameprimarilyfromitsfirstpremise.ButsinceoneofAl'sbestformulations(towhichIherebyreferreadersseekingagoodversionofAIthathasnotbeensobrutallypared)isinchapter1ofPeterUnger'sbookIgnorance:ACaseforScepticism(1975)itisinmorethanonesensethatitisanargument‘fromignorance.’

4Iactuallyhaven'tparedAItoitsbarestessentials.Itcouldbefurtherparedtoaone‐premiseargument:Idon'tknowthatnot‐H;so,Idon'tknowthatO.Thesecond,‘bridge’premisehasbeenaddedtofacilitatemytreatmentoftheargument,nicelydividingthoseissuesthatimpactontheacceptabilityofthefirstpremisefromthosegermanetothesecond.

AIisthefirstandgreatargumentbyskepticalhypothesis.Andthesecond,likeuntoit,isTheArgumentfromPossibility(AP),which,likeAI,takesitsnamefromitsfirstpremise,andwhichhasthisform:

1.ItispossiblethatHind.2.IfitispossiblethatHind,thenitispossiblethatnot‐Oind.So,3.Itispossiblethatnot‐Oind.4.Ifitispossiblethatnot‐Oind,thenIdon’tknowthatO.So,C.Idon’tknowthatO.

(Thesubscript'ind'indicatesthatwhatoccursinthescopeof'Itispossiblethat'istobekeptintheindicativemood,sothatthepossibilityexpressedwillbeanepistemicone.The‘bridge’premises,2and4,canbeomitted.)InthispaperIaddressonlyAI,butletmequicklyindicatehowAPshouldbehandled.Premise4,whichinitiallystrikesmanyasAP'sweakestlink,isactuallycorrect(DeRose1991,sectionG).Thus,theAPskepticmustbestoppedbeforeshereachesstep3.Fortunately,thetreatmentofAIthatIpresentinthispapercanbegeneralizedtohandletheinitialphase(steps1‐3)ofAPaswell.ThistreatmentofAPislefthereasanexerciseforthereader,butisexplainedinchapter3,especiallysectionK,ofmy(199O).

Page 5: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐3‐

Anditalsoseemsthatif,forallIknow,IamaBIV,thenIdon'tknowthatIhavehands.HowcouldIknowthatIhavehandsif,forallIknow,I'mbodiless(andthereforehandless)?But,atthesametime,itinitiallyseemsthatIdoknowthatIhavehands.Sotwoplausiblepremisesyieldaconclusionwhosenegationwealsofindplausible.Sosomethingplausiblehastogo.Butwhat?Andequallyimportantly,how?

Tobesure,thepremisesareonlyplausible,notcompelling.Thus,wewillalwayshaverecoursetotheMooreanreactiontothisargument:Declarethatitismorecertainthatoneknowsthatonehashandsthanitisthateitherofthepremisesoftheargumentistrue(muchlessthattheirconjunctionistrue),andthereforerejectoneofthosepremises,ratherthanaccepttheconclusion.Butalsoavailableistheskepticalreaction,whichistoaccepttheconclusion.

Butweshouldhopeforabettertreatmentoftheargumentthansimplychoosingwhichofthethreeindividuallyplausiblepropositions—thetwopremisesandthenegationoftheconclusion—seemsleastcertainandrejectingitonthegroundsthattheothertwoaretrue.Inseekingasolutiontothispuzzle,weshouldseekanexplanationofhowwefellintothisskepticaltrapinthefirstplace,andnotsettleformakingasimplechoiceamongthreedis‐tastefulwaysoutofthetrap.Wemustexplainhowtwopremisesthattogetheryieldaconclusionwefindsoincrediblecanthemselvesseemsoplausibletous.Onlywithsuchanexplanationinplacecanweproceedwithconfidenceandwithunderstandingtofreeourselvesfromthetrap.

ManyofthoseworkingonAIinrecentyearsseemtohaveunderstoodthis.5AndIhavegoodnewstoreport:Substantialprogresstowardsfinallysolvingthisskepticalpuzzlehasbeenmadealongtwoquitedifferentfronts.Thebadnewsisthat,asIshallargue,neitherapproachhassolvedthepuzzle.Buttheculminatinggoodnewsisthat,asIwillalsoargue,thenewsolutionIpresenthere,whichincorporatesimportantaspectsofeachofthetwoapproaches,canfinallysolvethis

5ThisisespeciallytrueofStewartCohen,towhomI'mindebtedforhisgeneralsetupofthepuzzleasaconflictofintuitions,asatisfactorysolutionofwhichrequiresanexplanationofwhythepuzzlearises.See(Cohen1988:93‐94).

Page 6: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐4‐

perenniallythornyphilosophicalproblem.WhilemoredetailsandprecisionwillbecalledforintheresultingsolutionthanIwillprovide,therewillbeenoughmeatonthebonestomakeitplausiblethatthefullyarticulatedsolutionliesinthedirectionIpointtohere.

Insections2‐4ofthispaper,Iexplorethecontextualistapproachtotheproblemofskepticism,andshowwhyithasthusfarfallenshortofsolvingthepuzzle.Insections5‐9,IturntoRobertNozick'sattempttosolveourpuzzle.SincetheshortcomingsofNozick'streatmentofknowledgeandskepticismhavebeen,atleasttomysatisfaction,dulydemonstratedbyothers,itwillnotbemypurposeheretorehearsethoseshortcomings,butrathertoexploreandexpanduponthesubstantialinsightthatremainsintactinNozick'saccount.Insections1O‐17,Ipresentanddefendmyowncontextualistsolution,whichIargueisthebestsolutiontoourpuzzle.Since,asIargueinsections15‐17,theskeptic'sownsolution,accordingtowhichweacceptAl'sconclusion,isamongthesolutionsinferiortotheoneIpresent,AIdoesnotsuccessfullysupportthatconclusion.

2. CONTEXTUALISTSOLUTIONS:THEBASICSTRATEGY

SupposeaspeakerA(for‘attributor’)says,‘SknowsthatP,’ofasubjectS'struebeliefthatP.Accordingtocontextualisttheoriesofknowledgeattributions,howstronganepistemicpositionSmustbeinwithrespecttoPforA'sassertiontobetruecanvaryaccordingtofeaturesofA'sconversationalcontext.6

6Forabitmoreonthenatureofcontextualisttheories,seemy(1992a).Thenotionof(comparative)strengthofepistemicposition,centraltomycharacterizationofcontextualism,willbeexplicatedbelowinsections1Oand11.

Forexemplarycontextualisttreatmentsoftheproblemofskepticism,inadditiontothepaperscitedbelowinsections3and4,seeespecially(Unger1986)and(Cohen1988).

Page 7: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐5‐

Contextualisttheoriesofknowledgeattributionshavealmostinvariablybeendevelopedwithaneyetowardprovidingsomekindofanswertophilosophicalskepticism.ForskepticalargumentslikeAlthreatentoshow,notonlythatwefailtomeetveryhighrequirementsforknowledgeofinterestonlytomisguidedphilosophersseekingabsolutecertainty,butthatwedon'tmeeteventhetruthconditionsofordinary,out‐on‐the‐streetknowledgeattributions.Theythusthreatentoestablishthestartlingresultthatwenever,oralmostnever,truthfullyascribeknowledgetoourselvesortoothermeremortals.

But,accordingtocontextualists,theskeptic,inpresentingherargument,manipulatesthesemanticstandardsforknowledge,therebycreatingacontextinwhichshecantruthfullysaythatweknownothingorverylittle.7Oncethestandardshavebeensoraised,wecorrectlysensethatweonlycouldfalselyclaimtoknowsuchthingsasthatwehavehands.Whythenarewepuzzled?Whydon'twesimplyaccepttheskeptic'sconclusionandhenceforthrefrainfromascribingsuchknowledgetoourselvesorothers?Because,thecontextualistcontinues,wealsorealizethis:Assoonaswefindourselvesinmoreordinaryconversationalcontexts,itwillnotonlybetrueforustoclaimtoknowtheverythingsthattheskepticnowdeniesweknow,butitwillalsobewrongforustodenythatweknowthesethings.Butthen,isn'ttheskeptic'spresentdenialequallyfalse?Andwouldn'titbeequallytrueforusnow,intheskeptic'spresence,toclaimtoknow?

Whatwefailtorealize,accordingtothecontextualistsolution,isthattheskeptic'spresentdenialsthatweknowvariousthingsareperfectlycompatiblewithourordinaryclaimstoknowthoseverypropositions.Oncewerealizethis,wecanseehowboththeskeptic'sdenialsofknowledgeandourordinaryattributionsofknowledgecanbecorrect.

Thus,itishoped,ourordinaryclaimstoknowcanbesafeguardedfromtheapparentlypowerfulattackoftheskeptic,while,atthesametime,thepersuasivenessoftheskepticalargumentisexplained.Forthefactthattheskepticcaninvokeveryhighstandardsthatwedon'tliveuptohasnotendencytoshow

7Thisisatleastsoaccordingtoskeptic‐friendlyversionsofcontextualistsolutions,aswillbeexplainedlaterinthissection.

Page 8: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐6‐

thatwedon'tsatisfythemorerelaxedstandardsthatareinplaceinmoreordinaryconversationsanddebates.

ThreeimportantpointsaboutcontextualiststrategiesasdescribedaboveshouldbemadebeforeImoveon.First,thistypeofstrategywillleaveuntouchedthetimidskepticwhopurportsbyAImerelytobeestablishingtheweakclaimthatinsome(perhaps‘high’or‘philosophical’)sense(perhapsinducedbythepresentationofAl)wedon'tknowtherelevantO,whilenotevenpurportingtoestablishtheboldthesisthatourordinaryclaimstoknowthatsamepropositionarefalse.Whethersuchatimidskepticalstanceisofanyinterestisatopicforanotherpaper.ThecontextualiststrategyisimportantbecauseAIinitiallyseemstothreatenthetruthofourordinaryclaims—itthreatenstoboldlyshowthatwe'vebeenwrongallalonginthinkingandsayingthatweknowthisandthat.Foritdoesn'tseemasifit'sjustinsome‘high’or‘philosophical’sensethatAl'spremisesaretrue:Theyseemtrueintheordinarysenseof'know'.Infact,oneisinitiallytemptedtosaythatthere'snogoodsenseinwhichIknowthatI'mnotaBIVorinwhichIcanknowIhavehandsifIdon'tknowthatI'mnotaBIV.How(andwhether)toavoidtheboldskepticalresultispuzzleenough.

Second,inpresentingthecontextualiststrategy,Ihaveaboveassumedaskeptic‐friendlyversionofcontextualism—oneaccordingtowhichthephilosophicalskepticcan(fairlyeasily),anddoes,succeedinraisingthestandardsforknowledgeinsuchawayastomakeherdenialsofknowledgetrue.Somecontextualistsmaythinkthatit'snotsoeasytosoraisethestandardsforknowledge,andthatadeterminedopponentoftheskepticcan,bynotlettingtheskepticgetawaywithraisingthem,keepthestandardslow.Buttheimportantpointistoidentifythemechanismbywhichtheskepticatleastthreatenstoraisethestandardsforknowledge.Whethertheskepticactuallysucceedsagainstadeterminedopponentinsoraisingthestandardsisoflittleimportance.Tosafeguardordinaryclaimstoknowwhileatthesametimeexplainingthepersuasivenessoftheskepticalarguments(whichisthegoalofhisstrategy),thecontextualistcanprovisionallyassumeaskeptic‐friendlyversionofcontextualism,leavingitasanopenquestionwhetherandunderwhichconditionstheskepticactuallysucceedsatraisingthestandards.Thecontextualist'sultimatepointwillthenbethis:Totheextentthattheskepticdoessucceed,shedoessoonlybyraisingthestandardsforknowledge,and

Page 9: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐7‐

sothesuccessofherargumenthasnotendencytoshowthatourordinaryclaimstoknowareinanywaydefective.

Third,AIcanbepuzzlingevenwhenoneisnotinthepresenceofaskepticwhoispresentingit.Theargumenthasaboutthesamedegreeofintuitiveappealwhenoneisjustconsideringitbyoneself,withoutanybody'ssayinganything.Butthecontextualistexplanation,asdescribedabove,involvesthestandardsforknowledgebeingchangedbywhat'sbeingsaidinaconversation.8Forthemostpart,Iwillframethecontextualistexplanationintermsofsuchconversationalrules,largelybecausethat'swhatbeendonebymycontextualistpredecessors,withwhomIwanttomakecontact.Butwemustrealizethattheresultingsolutionwillhavetobegeneralizedtoexplainwhytheargumentcanbesoappealingevenwhenoneisconsideringitinsolitude,withnothingbeingsaid.Thebasicideaofthegeneralizationwilltakeeitherorbothofthefollowingtwoforms.First,itcanbemaintainedthatthereisaruleforthechangingofthestandardsforknowledgethatgovernsthetruthconditionsofourthoughtsregardingwhatisandisnotknownthatmirrorstheruleforthetruthconditionsofwhatissaidregardingknowledge.Inthatcase,ananalogueofthecontextualistsolutioncanbegivenforthought,accordingtowhichthepremisesandconclusionofAIaretrulythought,butmytruethoughtthat,say,Idon'tknowthatIhavehands,hadwheninthegripofAI,willbecompatiblewithmythought,madeinanothercontext,thatIdoknowthatverything.Second,ourjudgmentregardingwhethersomethingcanorcannotbetrulyasserted(underappropriateconditions)mightbeheldtoaffectourjudgmentregardingwhetherit'strueorfalse,evenwhenwemakethisjudgmentinsolitude,withnothingbeingsaidatall.ThatthepremisesofAIcouldbetrulyasserted,then,makesthem(atleast)seemtrueevenwhenthey'rejustbeingthought.

Myownsolutionwillemploythebasiccontextualiststrategyexplainedinthissection.But,asshouldbeapparentalready,wehaven'texplainedthepersuasivenessofAl,andthushaven'tsolvedourpuzzle,ifwehaven'tlocatedandexplainedtheconversationalruleormechanismbywhichtheskepticraises(orthreatenstoraise)thestandardsforknowledge.Andherecontextualistshavehad

8ThankstoRichardGrandyandtoPeterUngerforpressingthispoint.

Page 10: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐8‐

littletooffer.Thetwomainproposalsthathavebeenputforwardarediscussedinthefollowingtwosections.

3. SOMEOLDCONTEXTUALISTSOLUTIONS:LEWIS’S‘RULEOFACCOMMODATION’

ThoughsubstantialpapershavebeenlargelydevotedtocontextualismanditsabilitytoexplaintheworkingsofskepticalargumentslikeAl,oneofthebestattemptstoexplainhow(bywhatruleorconversationalmechanism)skepticsraisethestandardsforknowledgeistobefoundinDavidLewis's‘ScorekeepinginaLanguageGame’(1979a),apaperthat,whilenotprimarilyaboutknowledgeattributions,doestreattheminpassing.9

AccordingtoLewis,‘rulesofaccommodation’operateinmanyspheresofdiscoursethatcontaincontext‐sensitiveterms.10Suchrulesspecifythatwhenastatementismadecontainingsuchaterm,then—ceterisparibusandwithincertainlimits—the‘conversationalscore’tendstochange,ifneedbe,soastomakethatstatementtrue.Forexample,'flat',accordingtoLewis,isacontext‐sensitiveterm:howflatasurfacemustbeinorderforasentencedescribingitas‘flat’tobetrueisavariablematterthatisdeterminedbyconversationalcontext.Andonewaytochangetheconversationalscorewithrespecttothestandardsinplaceforflatnessistosaysomethingthatwouldrequireforitstruthsuchachangeinstandards.Suppose,forexample,thatinacertainconversationthestandardsforflatnessarerelaxedenoughthatmydesktopcountsasbeingflat.IfIwerethentosay,‘My

9Iamheredistinguishingamongcontextualistsolutionsaccordingtothemechanismorrulethattheyallegeraisesthestandardsforknowledge.AlthoughtherearesuggestionsoftheRelevantAlternatives(RA)approachin‘Scorekeeping,’Lewis'sRuleofAccommodationisquitedifferentfromthemechanismmostRAtheoristsposit—thustheseparatetreatmentofLewis.TotheextentthatLewisisarelevantalternativist,theRAaspectsofhistreatmentareaddressedbelowinsection4.

10Seeespecially346‐47.

Page 11: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐9‐

desktopisnotflat,’whatIsaywouldbefalseifitwereevaluatedaccordingtothestandardsforflatnessinplaceimmediatelybeforethisissaid.ButtheRuleofAccommodationspecifiesthatinsuchasituation—atleastundertherightcircumstances,wheretheceterisparibusclauseismet—thestandardsforflatnessareraisedsoastomakemystatementtrue.

Lewissuggeststhatskepticsmanipulateasimilarruletochangethestandardsforwhatistocountasknowledge.AccordingtoLewis'sexplanationoftheplausibilityofskepticism,then,theskeptic'sstatementschangetheconversationalscore—here,raisethestandardsforknowledge11—soastomaketheskeptic'sstatementstrue.Oncethestandardsforknowledgehavebeensoraised,then

the commonsensical epistemologist must concede defeat. And yet he was in no way wrong when he laid claim to infallible knowledge. What he said was true with respect to the score as it then was.12 (355)

HereLewisdisplaysthebasiccontextualiststrategy:Heprotectsthetruthofwhatweordinarilysay,orsaybeforetheskepticgetsaholdofus,fromtheskeptic'sattackbyexplainingthesuccessofthatattackintermsoftheskeptic'schangingwhatcountsasknowledge,or,here,‘infallibleknowledge.’Thus,thepersuasivenessoftheskeptic'sattackisexplainedinsuchawayastomakeitunthreateningtoourordinaryclaimsofknowledge.

Andthisexplanationinitiallyappearstobetailor‐madeforAI,forAl'sfirstpremiseisadenialofknowledge—preciselythetypeofassertionthatariseinthestandardsforknowledgecanhelptomaketrue.Suchadenial,then,isjustthesortofthingthatcanraisethestandardsforknowledgeviaaRuleofAccommodation.Perhapswhentheskepticassertsthisfirstpremise,thestandardsforknowledgeare

11ForLewis,asforRelevantAlternativestheorists(seesection4,below),thisraisingofepistemicstandardsconsistsinexpandingtherangeofrelevantalternativestowhatonebelieves,thatis,therangeofalternativesthatonemustbeinapositiontoeliminateinordertocountasknowing.

12Whycan'tthecommonsensicalepistemologistsimplydeclareagainthatheknows,andrelyonaRuleofAccommodationtolowerthestandardsbackdownsoastomakehisclaimtrue?TothisLewisrespondsthat,forsomeadmittedlyunknownreason,thestandardsaremoreeasilyraisedthanlowered(355).

Page 12: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐10‐

raised,viatheRuleofAccommodation,toalevelatwhichwecountasknowingneitherthatwe'renotBIVs,northatwehavehands.13

ButaRuleofAccommodationcannotreallyexplainthepersuasivenessofAI,or,moregenerally,ofanyargumentbyskepticalhypothesis.Tovividlyillustratewhythisisso,letusimagineandcomparetwoskepticswhoaretryingtoconvinceyouthatyoudon'tknowthatyouhavehands.The‘AIskeptic,’truetohername,reliesonAI,which,asInotedinsection1,isprettypowerful.The‘simpleskeptic,’ontheotherhand,simplyinsiststhatyoudon'tknowthatyouhavehands,offeringnoreasoningatallforthisskepticalassertion.

InseekingasolutiontothepuzzlegeneratedbyAI,weshouldhopeforasolutionthat,attheveryleast,explainswhytheAIskepticismoreconvincingthanthesimpleskeptic.Ifourexplanationdoesnotdothismuch,thenwehaven'texplainedhowtheskepticalargumentworksonusinanywaysufficienttodifferentiateitfromabald(anddogmatic!)skepticalassertion.

ButtheRuleofAccommodation,asitstands,appearstobeequallyaccommodatingtobothofourimaginedskeptics.WhenthesimpleskepticclaimsthatIdon'tknowthatIhavehands,thesupposedRuleofAccommodationshouldraisethestandardsforknowledgetosuchapointastomakeherclaimtrue.Ofcourse,theceterisparibusclausemayblockthisresult,dependingonhowitisfleshedout.ButthereisnothingtothisRule,atleastasithassofarbeenarticulated,thatwouldfavortheAIskepticoverthesimpleskeptic.Thus,theexplanationbasedonthisRuledoesnotdifferentiatebetweenthesetwoskeptics.Butifitdoesn'tdothat,itdoesn'tsolveourpuzzle.

Toavoidpossiblemisunderstanding,letmeclearlystatethatmyobjectionisnottotheproposedsolution'slackofprecision—thatwe'renotgivenaveryclearideaofwhentheRuleofAccommodationtakeseffect,thattheRulesaysmerelythat

13Tobefair,Lewis,asI'vepointedout,treatsknowledgeonlyinpassing.Althoughtheskepticheimaginesdoesutilizeaskepticalhypothesis(thatoneisthevictimofadeceivingdemon(355)),suggestingthatthetreatmentLewisoffersshouldbehelpfulinsolvingthepuzzleofskepticalhypotheses,heneverexplicitlyattemptsasolutiontoourpuzzle.Still,sincethesolutionatleastsuggestedbyLewisisoneofthebestonoffer,it'sworthestablishingthatitcan'treallysolvethepuzzle.

Page 13: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐11‐

thestandardstendtochangeinacertainwayprovidedthatthe(highlyunarticulated)ceterisparibusclauseismet.Myownsolutionwillbelikewiseimprecise.No,theproblemisn'tthattheRuleisn'tcompletelyfilledin,butratherthat,forthereasonsgivenabove,sincetheexplanatoryworkneededtosolvethepuzzleisn'tdonebytheaspectsoftheRulethathavebeenprovided,itwillhavetobedonebyjustthoseaspectsoftheRulethathaven'tbeenprovided.And,aswe'velittleideawhattheseaspectsare,we'velittleideaofwhatitisthatmaysolvethepuzzle.14

Perhaps,whenit'smorefullyarticulated,theoperativeRuleofAccommodationcanbeseentocontainafeaturethatfavorstheAIskepticoverthesimpleskeptic.Inthatcase,thesolutiontoourpuzzle,whichhassofareludedus,may(atleastinpart)befoundinafullerarticulationofthatRule.

ButIdoubtthatthesolutionevenliesinthatdirection.One(secondary)reasonformydoubtisthatpositiveclaimstoknowthatskepticalhypothesesdon'tobtainseemtoraisethestandardsforknowledgeaswellasdodenialsofsuchknowledge.

ToillustratethisI'lluseFredDretske'sfamiliarexampleofmulescleverlypaintedtolooklikezebras(Dretske197O,1O15‐16).IfIsawwhatlookedtobezebrasinthezebracageatazoo,Iwouldordinarilyclaimtoknowthattheanimalsinthecagearezebras.(Suppose,forinstance,thatmysonaskedme,‘Doyouknowwhatthoseanimalsare?’Iwouldrespondpositively.)AskepticmightchallengethissupposedknowledgewithaninstanceofAlwhereOisThoseanimalsarezebrasand

14NoneofthisistodenythatthereissomeRuleofAccommodationaccordingtowhichthestandardsforknowledgetendtoberaisedto‘accommodate’denialsofknowledge.NorisiteventodenythatsuchRulesofAccommodationhelptheAlskeptic.Infact,Ifinditplausibletosupposethatmanydenialsofknowledge,includingthoseofAlskeptics,oftendoexertanupwardpressureonthestandardsforknowledgeviasomesuchrule.Likewise,certainsettings(inadditiontocourtsoflaw,certainphilosophyclassesaregoodexamples),itseemstome,tendtomilitateinfavorofhighepistemicstandards.Alskepticsmaytakeadvantageofthesefactors,theinfluenceofwhichmayexplainsomeofthepersuasivenessoftheirskepticalperformances.Buttosolveourpuzzle,wewantprimarilytoexplainwhatthenatureoftheskepticalargumentitselfaddstotheeffectivenessoftheskeptic'sperformancethatgoesbeyondwhatiscontributedbytheskeptic'ssettingandthefactthatsheassertsherconclusion.

Page 14: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐12‐

HisThoseanimalsaremulescleverlypaintedtolooklikezebras.Theresultingpremisesareindividuallyplausible,sinceIcouldn'ttellacleverlypaintedmulefromazebra.AcontextualisttreatmentofthisinstanceofAlwillclaimthatinassertingthefirstpremise,theskepticraisesthestandardsforknowledgetoalevelatwhichIcountasknowingneitherthattheanimalsarenotcleverlypaintedmulesnorthatthey'rezebras.

Anditindeeddoesseemthatoncethisskepticalhypothesisisbroughtintoplay,IcannothappilyclaimtoknowwhatIsohappilyclaimedtoknowbefore.Tobeinagoodenoughpositiontoclaimtoknowthattheanimalsarezebrasaccordingtothestandardsbroughtintoplaybytheskeptic,onemustbeinagoodenoughpositionthatonecanruleout15thehypothesisthattheyarecleverlypaintedmules.SinceI'mnotinthatkindofepistemicposition,Idon'tcountasknowing,althoughperhapssomeonemorefamiliarwithmulesandzebraswouldstillcountasknowing,evenatthesehigherstandards—someone,forinstance,whowasinapositiontosay,‘No,theycan'tbemules:nomule'sheadisshapedlikethat.’

Butthesesamehigherstandardsseemtobeinducedwhentheskepticalhypothesisisbroughtintoplaybyapositiveclaimtoknowthatitdoesn'tobtain.Suppose,tovaryDretske'sexample,thatIamconfronted,notbyaskeptic,butbyaboastfulzoologist.Hebrags,‘Duetomyvastknowledgeofzebraandmuleanatomy,Iknowthatthoseanimalsarenotmulescleverlypaintedtolooklikezebras;soIknowthatthey'rereallyzebras.’Thiszoologist,asmuchastheskeptic,seemstoinvokehigherstandardsforknowledgeatwhichhe,butnotI,willcountasknowingthattheanimalsarezebras.HecertainlyseemstobeclaimingmorethanthemundaneknowledgethatevenIpossess—andclaimtopossess—inanordinaryzoosetting,wherethere'snosuchzoologisttellingmewhat'swhat.

ButaRuleofAccommodationcannotaccountforthisriseinstandards,forthezoologistdoesn'tdenyanysupposedknowledge.Tothecontrary,whathedoesismakepositiveclaimstoknow,andariseinstandardsforknowledgecanneverhelp

15Forsomecommentsonthisnotionof‘rulingout’seesections4and5,below.

Page 15: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐13‐

tomaketrueapositiveclaimtoknow.So,asIsaid,aRuleofAccommodationcan'tdoanythingtoexplainthisnotableriseinepistemicstandards.16

MyprimaryreasonfordoubtingthatoursolutionistobefoundinafullerarticulationoftheRuleofAccommodationisthis:ToexplainthepersuasivenessofAl(and,inparticular,ofitsfirstpremise)insuchawayastodifferentiatetheAlskepticfromthesimpleskeptic,wemustidentifythefeatureofskepticalhypothesesthatmakesitparticularlyhardtoclaimortothinkthatoneknowsthattheyarefalse.FarfrombeingfoundinaRuleofAccommodation,then,asolutiontoourpuzzle,ifit'stobefoundatall,istobefoundinanexplanationofwhatitisaboutskepticalhypothesesthatmakesthesepropositions,asopposedtoeversomanyotherpropositions,sucheffectiveskepticalweapons.So,tosolvethepuzzle,wemustlocateorarticulatethispeculiarlypotentfeatureofjustthesepropositions(theskepticalhypotheses).And,onceweseewhatthisfeatureisandhowitworks,theRuleofAccommodationisdestinedtoplayonlyarathersubsidiaryrole(seenote14)inexplainingtheeffectivenessoftheskeptic'sattack.

MysecondaryreasonfordoubtingthattheRuleofAccommodationmightsolveourpuzzlewasworthbringingupbothbecauseitseemstometohavesomeforce,andbecauseitvividlyillustratesthisimportantfact:Theupwardpressureonthestandardsforknowledgethatbringingskepticalhypothesesintoplayexertsisexertedwhetherthehypothesesareraisedindenialsofknowledgeorinpositiveclaimstoknow.

16It'sbeenproposedtome,onbehalfoftheRuleofAccommodationandthesolutiontoAlthatcanbebasedonit,thattheboastfulzoologist,whilehedoesnotsaythatIdon'tknow,doesstronglysuggestorimplythatIdon't,andtheRuleofAccommodationoperateshereonhissuggestion:thestandardsgoupsoastomakethesuggestiontrue.Iamskepticalofthisattempttosalvagethesolutionfortworeasons.First,Isuspectthattherulebecomesfartoopowerfulifit'sallowedtoworkonwhatwesuggestaswellasonwhatwesay.Second,thestandardsforknowledgeseemlikewiseraisedeveniftheboastfulzoologistthinksIamalsoanexpert,andthinksheisinformingmethathetooknowswhat'swhat.Herehe'snotevensuggestingthatIdon'tknow.

Page 16: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐14‐

4. SOMEOLDCONTEXTUALISTSOLUTIONS:THE‘RELEVANTALTERNATIVES’APPROACHANDTHERULEOFRELEVANCE17

Perhapsthemostpopularsolutiontoourpuzzlehasbeenputforwardbyadvocatesofthe‘RelevantAlternatives’theoryofknowledge(RA).AgainsupposeaspeakerAsays,‘SknowsthatP.’AccordingtoRA,suchanassertionismadewithinandmustbeevaluatedagainstacertainframeworkofrelevantalternativestoP.ToknowthatPistohaveatruebeliefthatPandtobeabletoruleouttheserelevantalternatives.ButnoteverycontraryoforalternativetoPisarelevantalternative.18Inanordinarycaseofclaimingtoknowthatsomeanimalsinthezooarezebras,toagainuseDretske'sexample,thealternativethatthey'recleverlypaintedmulesisnotrelevant.Thus,Icantruthfullyclaimtoknowthey'rezebrasdespitemyinabilitytoruleoutthisfancifulalternative.

Butinvariousextraordinarycases,thepaintedmuleshypothesisisarelevantalternative.ItmightbemaderelevantbysomeextraordinaryfeatureofS(theputativesubjectofknowledge)orhersurroundings.19ButmostRAtheoristsarecontextualists,andallowthatfeaturesoftheconversationalcontextinwhichA(theascriberofknowledge)findshimself,inadditiontofeaturesofSandhersurroundings,caninfluencewhichalternativesarerelevant.20AlvinGoldman,for

17FredDretske(seehis197O,1971,198la,198lb),althoughhedoesadvocateaRelevantAlternativestheoryofknowledge,proposesatreatmentofAIquitedifferentfromthatdescribedbelow.I'mnotcertainwhetherDretske'sisevenacontextualistversionofRA.(AsInoteinpart2ofmy1992a,onecanbeanRAtheoristwithoutbeingacontextualist.)OnethingisclearaboutDretske'streatmentofAI:Hedeniespremise(2).Giventhis,histreatmentrunsintothesamedifficultiesasdoesNozick's;seeespeciallysection9below.

18See,forexample,Dretske197O,1O22;Goldman1976,772;andStine1976,249.

19Thus,ifSisatazoothatfairlyconsistentlyusespaintedmulesinanattempttofoolthezoo‐goingpublic,thenthepaintedmulehypothesisisrelevant.So,eventhoughSisluckyenoughtobeatthiszooononeoftheraredayswhenactualzebrasarebeingused,Scannottruthfullybesaidtoknowthatthey'rezebrasunlesssheisabletoruleoutthepaintedmulehypothesis,whichshecan'tdounlesssheknowsmorethanIdoaboutzebrasandmules.

20AsIexplaininpart2ofmy1992a,anRAtheoristcanbeaninvariantistifheallowsonlyfactorsabouttheputativesubjectofknowledgeandhersurroundings,andnotconversationalfactorspertainingtothespeaker

Page 17: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐15‐

instance,suggeststhat‘ifthespeakerisinaclassinwhichDescartes'sevildemonhasjustbeendiscussed,’thencertainalternativesmayberelevantthatordinarilyarenot(1976:776).

Itisthiscontextualistaspectof(mostversionsof)RAthatfacilitatesthemostcommonlyproposedsolutiontoourpuzzle,theRelevantAlternativesSolution(henceforth,'RAS').Withsomeslightvariationsindetailindifferentpresentationsofit,thebasicideaofRASisthis:TheAlskeptic'smentioningoftheBIVhypothesisinpresentingthefirstpremiseofAlmakesthathypothesisrelevant.Oncetheskepticalhypothesishasbeenmaderelevant,wecorrectlysensethatwecannottruthfullyclaimtoknowanythingcontrarytoitunlesswecanruleitout.Sinceweareunabletoruleitout,andsinceitisanalternativetobothIamnotaBIVandtoIhavehands,wecorrectlysensethatwecouldonlyfalselyclaimtoknowthesethings.Sotheskeptictruthfullyassertsthatwedon'tknowthatthehypothesisdoesn'tobtain,andthentruthfullyconcludesthatwedon'tknowthatwehavehands.21

Whythenarewepuzzled?BecauseweatthesametimerealizethattheBIVhypothesisisnotordinarilyrelevant.Werealizethatinmostoftheconversationalcircumstancesinwhichwefindourselves,ourinabilitytoruleouttheskeptic'sfar‐fetchedhypothesisisnobartoourtruthfullyclaimingtoknowsuchthingsasthatwehavehands.Thus,evenaswefindtheskeptic'sdenialsofknowledgepersuasive,werealizethatwhenweagainfindourselvesinmoreordinarycontexts,itwillnotonlybecorrectforustoclaimtoknowsuchthings,itwouldbewrongtodenythatweknowthemmerelybecausewecan'truleouttheBIVhypothesis.Whatwefailtorealize,accordingtoRAS,isthatourordinaryclaimstoknowsuchthingsasthatwe

(theascriberofknowledge),toaffectwhichalternativesarerelevant.Mattersgettrickywithfirst‐personknowledgeclaims,whereSandAareidentical.Here,inadditiontoallowingfeaturesthataffecthowgoodanepistemicpositionoursubjectactuallyisin,andthattherebyattachtoherquaputativesubjectofknowledge,contextualistRAtheoristswillalsoallowfeaturesofherconversationalcontext,whichaffecthowgoodapositionshemustbeintocountasknowing,andwhichtherebyattachtoherquaattributorofknowledge,toinfluencewhattherangeofrelevantalternativesis.

21Again,hereI'monlygivingtheskeptic‐friendlyversionofthiscontextualistsolution.AnRAtheoristmightbelessfriendlytotheskepticbyholding,forexample,thatmentioninganalternativemakesthatalternativerelevantonlyifone'sconversationalpartnerletsonegetawaywithmakingitrelevant.

Page 18: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐16‐

havehandsarecompatiblewiththeskeptic'spresentdenialthatweknowthoseverythings.

RAS,then,isaninstanceofthegeneralcontextualiststrategy—oneaccordingtowhichtheraisingofthestandardsconsistsinenlargingtherangeofalternativesthatarerelevantandthatonemustthereforebeinapositiontoruleoutinordertocountasknowing.TheconversationalruleormechanismthatRASpositsforenlargingthatrange(raisingthestandardsforknowledge),then,isthatmentioningapropositionQ—ceterisparibusandwithincertainlimits,nodoubt—tendstomakeQacontextuallyrelevantalternativetoanyPthatiscontrarytoQ.CallthistheRuleofRelevance.22

NotethatthisRuleofRelevance,asopposedtotheRuleofAccommodation,canhandlecaseslikethatoftheboastfulzoologist,inwhichapositiveclaimtoknowthataskepticalhypothesisdoesn'tobtainseemstohavethesameeffectonthemeaningofsentencescontaining'know'aswouldadenialofsuchknowledge.ThisistobeexpectedonthepresentRuleofRelevance,onwhichboththedenialandtheclaimtoknowwill,byincludingamentionoftheskepticalhypothesis,expandtherangeofrelevantalternativessothatitwillincludethathard‐to‐rule‐outhypothesis.

ButtoexplainthepersuasivenessofAI(particularlyofitsfirstpremise),andtotherebysolveourpuzzle,atreatmentofAImusttelluswhatitisaboutskepticalhypothesesthatmakesitdifficulttoclaimtoknowthattheydon'tobtain.ThekeyfeatureofskepticalhypothesesthatRASseizesonisclearlythis:wecan'trulethemout.

22Ofcourse,itshouldn'tbeheldthatjustanymentionofapropositionmakesthatpropositionarelevantalternative.Inordertobemaderelevant,thepropositionmust,nodoubt,beinsertedintoaconversationintherightway.ButtheadvocateofRAScanplausiblyclaimtohaveexplainedthepersuasivenessofAIevenifhehasn'tgivenanexactspecificationoftheconditionsunderwhichamentioningofapropositionmakesthatpropositionarelevantalternative.PlausiblyholdingthatinpresentingAItheskepticdoesinsertherskepticalhypothesisintotheconversationintherightway,theadvocateofRAScanleaveitasafutureprojecttospecifymoreexactlyjustwhichwaysaretherightways.Althoughthisbyitselfwillbeneithernecessarynorsufficientforthementioningofapropositiontobeoftherightkindtoenlargetherangeofrelevantalternativessoastoincludeit,itnonethelessmayberelevantthatintheskeptic'spresentationofAl'sfirstpremise,thementioningofthehypothesisoccurswithinthescopeofanepistemicoperator—’Sdoesnotknowthat....’

Page 19: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐17‐

Andisn'ttheresomethingtothisexplanation?Foritseemsthatweindeedcan'truleout(effective)skepticalhypotheses,anditfurtherseemsthatitispreciselythisfactthatmakesthemsucheffectiveskepticalweapons.

Butthoughitisplausibletosupposewecan'truleoutskepticalhypotheses,andalsoplausibletosaythatwedon'tknowthattheydon'tobtain,itisfutiletotrytoexplaintheplausibilityofthelatterbythatoftheformer.

Indeed,thereareplentyofotherphrasesthatcanbeusedplausiblytodescribeourapparentlylimitedepistemicpositionwithregardtoeffectiveskepticalhypotheses.Allofthefollowingdescriptionsaboutmypositionvis‐a‐vistheBIVhypothesishavesomeinitialplausibility:Icannotruleitout,Idon'tknowthatitdoesn'tobtain(anddon'tknowwhetheritobtains),Ican'ttellthatitdoesn'tobtain(andcan'ttellwhetheritobtains),Ican'tdiscernthatitdoesn'tobtain(andcan'tdiscernwhetheritobtains),andIcan'tdistinguishitsobtainingfromitsnotobtaining,andsoon,andsoforth.Butcitingoneofthesetoexplaintheplausibilityofanotherdoesn'toccasioneventheslightestadvanceinourunderstanding.

WhataccountsfortheplausibilityofsayingthatIdon'tknowthatI'mnotaBIV?ThefactthatIcan'tdiscernthatI'mnotone?Thisisnoexplanation.Itseemsjustasgood(infact,tome,better)toreversethingsandclaimthatthefactthatIdon'tknowthatI'mnotaBIVaccountsfortheplausibilityofsayingthatIcan'tdiscernthatI'mnotone.

Likewiseforrulingout.Itisindeedplausibletosupposethatwecan'truleoutskepticalhypotheses.Andit'splausiblethatwedon'tknowthattheydon'tobtain.Butitdoesn'tseemtoadvanceourunderstandingmuchtoexplaintheplausibilityofeitherbythatoftheother.

(Anexerciseforthereader:Randomlypicktwooftheabovenegativeassessmentsofourepistemicpositionvis‐a‐viseffectiveskepticalhypotheses.Thenconsiderwhethertheplausibilityofthefirstcanbeexplainedbyreferencetothesecond.Thenreversethingsandconsiderwhethertheplausibilityofthesecondcanbeexplainedbyreferencetothefirst.Trythesameprocedureonanotherpairofdescriptions.(Ifyou'rerunninglowonsuchnegativeassessments,you'llfindit'seasy,followingmylead,tocomeupwithmanymoreonyourown.)Thenevaluate

Page 20: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐18‐

thesuccessofexplainingtheplausibilityofAl'sfirstpremisebyreferencetothefactthatwecan'truleouteffectiveskepticalhypotheses.)

Toexplainwhywefeelsomepulltowardsdescribingourepistemicpositionwithregardtoskepticalhypothesesinanyoftheabovelessthanflatteringways—aswellasverymanyotherwaysthatIdidn'tbothertomention—weneedanexplanationthatreachesoutsidethiscircleofall‐too‐closelyrelatedtermsofepistemicappraisal.23Indeed,aswillemergeinthefollowingsections(especiallysection8),thebestexplanationfortheplausibilityofAl'sfirstpremisealsoseemstoprovideagoodaccountofwhyitseemsthatwecan'truleoutskepticalhypotheses,aswellasanexplanationoftheplausibilityofthevariousotherpessimisticevaluations.Oncethisexplanationisinplace,itbecomesevenclearerthatnoneofthethingsit'susedtoexplaincanbeproperlyusedtoexplaineachother.

5. THESUBJUNCTIVECONDITIONALSACCOUNT(SCA)OFTHEPLAUSIBILITYOFAI’SFIRSTPREMISE

Themainstumblingblockofthecontextualistsolutionswe'vediscussedhasbeenafailuretoexplainwhatitisaboutskepticalhypothesesthatmakesitsoplausibletosupposethatwedon'tknowthatthey'refalse.ThispointofweaknessinthecontextualistsolutionsistheparticularpointofstrengthofNozick'streatmentofAIinhisPhilosophicalExplanations(1981).InthisandthefollowingthreesectionsI'llpresentanddefendtheSubjunctiveConditionalsAccount(SCA)oftheplausibilityofAl'sfirstpremise,whichI'veabstractedfromNozick'saccountofknowledgeandskepticism.

23Goldman(1976)cashesout‘discriminating’whatonebelievesfromarelevantalternativetoitintermsofwhatonewouldbelieveifthealternativeobtained.This,combinedwiththeRuleofRelevance,couldyieldanapproachtoskepticismclosetotheoneI'llheredefend.Goldmanhimselfdoesnotproposeasolutiontotheskepticalproblem;hestrivestoremainneutralontheissue.ButI'llbeworkinginthegeneraldirectionIthinkGoldmanpointsto.

Page 21: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐19‐

AccordingtoSCA,theproblemwithmybeliefthatI'mnotaBIV—andIdohavesuchabelief,asdomostofus—isthatIwouldhavethisbelief(thatI'mnotaBIV)evenifitwerefalse(evenifIwereone).ItisthisthatmakesithardtoclaimtoknowthatI'mnotaBIV.For,accordingtoSCA,wehaveaverystronggeneral,thoughnotexceptionless,inclinationtothinkthatwedon'tknowthatPwhenwethinkthatourbeliefthatPisabeliefwewouldholdevenifPwerefalse.Let'ssaythatS'sbeliefthatPisinsensitiveifSwouldbelievethatPifPwerefalse.SCA'sgeneralizationcanthenberestatedasfollows:WetendtojudgethatSdoesn'tknowthatPwhenwethinkS'sbeliefthatPisinsensitive.

Asiswellworthnoting,thisgeneralinclinationexplainstheoperationofnonphilosophicalskepticalhypothesesthatarefarlessradicalthantheBIVhypothesisoreventhepaintedmulehypothesis.Justso,itservestoexplainwhy,eventhoughIfeelinclinedtosaythatIknowtheBullswontheirgamelastnightbecauseIreadtheresultinasinglenewspaper,Istillfeelstronglypulledtowardadmittingthe(mildly)skepticalclaimthatIdon'tknowthatthepaperisn'tmistakenaboutwhichteamwon:Irealizethatmybeliefthatthepaperisn'tmistakenisabeliefIwouldholdevenifitwerefalse(evenifthepaperweremistaken).

Indeed,afterencounteringacoupleofinstancesofAIwithdifferentskepticalhypothesespluggedintothe'H'slot(forexample,theBIV,thepaintedmules,andthemistakenpaperhypotheses),onedevelopsasenseofwhatmakesforaneffectiveskepticalhypothesisand,thus,anabilitytoconstructconvincinginstancesofAIoneself.TomakeAl'ssecondpremiseconvincing,itisusuallysufficient(thoughnotnecessary)thatHbeincompatiblewithO.Butwhataboutthefirstpremise?Tomakeitconvincing,weinstinctivelylookforahypothesisthatelicitsinthelistenerboththebeliefthatthehypothesisdoesn'tobtainandanacknowledgementthatthisbeliefisoneshewouldholdevenifthehypothesisdidobtain.

Uponhearingthehypothesis,typicallyonecan'thelpbutprojectingoneselfintoit.Howwouldthingsseemtomeifthatsituationobtained?Well,prettymuch(orsometimesexactly)astheyactuallyseemtome.And,so,whatwouldIbelieveifsucha‘strange’situationobtained?Prettymuch(orexactly)whatIactuallybelieve.

Page 22: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐20‐

Forexample,andinparticular,ifIwereaBIV,IwouldbelieveeverybitasfirmlyasIactuallydothatIwasn'tone.ButifthisbeliefisoneIwouldholdevenifitwerefalse,howcanIbeinapositiontotellthat,ordiscernthat,orknowthat,it'strue?

AsI'vejusthinted,asimilarexplanation,intermsofsubjunctiveconditionals,canexplaintheplausibilityoftheotherwayswefeelinclinedtodescribeourseeminglylimitedepistemicpositionvis‐a‐viseffectiveskepticalhypotheses.Considerespeciallythedescriptioninvolving'rulingout'.Inanormalzoosetting,mostofuswouldtakeourselvestoknowthattheanimalsinthezebracagearezebras.Fromthis,itseems,weshouldbeabletoinferthatthey'renotcleverlypaintedmules,sincezebrasaren'tmules.Sowhyarewereluctanttocountourseeingthezebrasandperformingthisinferenceasacaseofrulingoutthepaintedmulehypothesis?Because,theexplanationgoes,evenafterperformingtheinference,itstillseemswewouldbelievetheobservedanimalsweren'tpaintedmulesiftheywerepreciselythat.Whydoesitseemwecan'ttellthatthey'renotpaintedmules?Becausewewouldbelievetheyweren'teveniftheywere.Dittoforwhyweseeminglycan'tdiscernthatthey'renotandwhyitseemswecan'tdistinguishtheirbeingcleverlypaintedmulesfromtheirnotbeingsuch,etc.

AlsoworthnotingistheusefulnessofSCAinexplainingourreluctancetoascribeknowledgeincertainlotterysituations.Evenwheretheoddsofyourbeingaloserareastronomicallyhigh(thereare2Omilliontickets,onlyoneofwhichisawinner,andyouhavebutoneticket),itcanseemthatyoudon'tknowthatyou'realoserofafairlotteryifthewinnerhasn'tyetbeenannounced.SCAaccountsforthisseeming:Yourbeliefthatyou'realoserisoneyouwouldholdevenifyouwerethewinner.

SCAisapowerfulexplanation.Butthereareproblems.AsIsuggestedabove,thereareexceptionstothegeneralinclinationtowhichSCAappeals:TherearecasesinwhichitseemstousthatsomeSdoesknowthatPeventhoughwejudgethatSwouldbelievethatPevenifPwerefalse.Someoftheseexceptionswillbequicklydiscussedinsections6and7below.Thefirstandmainpointtomakeregardingsuchexceptions,ofcourse,isthatthisverygeneralinclinationneedn'tbeexceptionlesstoperformtheexplanatoryroleSCAassignsit.Insection8wewillseestronggroundsforendorsingSCAasbeingatleastontherighttrackdespitethe

Page 23: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐21‐

exceptionstothegeneralizationtowhichitappeals.Buttheseexceptionsarestillworthexamining,fortheywillindicatecertainimportantdirectionsinwhichSCAcanbeimproved,eventhoughwewon'tbeinapositiontomakeSCAideallyprecisehere.

6. SCA,GRANDMOTHERS,ANDMETHODS

First,then,consideracasediscussedbyNozick:

A grandmother sees her grandson is well when he comes to visit; but if he were sick or dead, others would tell her he was well to spare her upset. Yet this does not mean she doesn't know he is well (or at least ambulatory) when she sees him. (1981, 179)

Here,itseems,thegrandmotherknowshergrandsoniswell,thoughitcanseemthatshedoesn'tsatisfythethirdconditionofapreliminaryformofNozick'sanalysisofSknowsthatP,whichis:

(3) If p weren't true, S wouldn't believe that p.

Nozick'sresponseistorelativizethisthirdconditiontothemethodbywhichShascometobelievethatp,yielding:

(3) If p weren't true and S were to use M to arrive at a belief whether (or not) p, then S wouldn't believe, via M, that p (179),

where'M'isthemethodbywhichShascometobelievethatp.24

UnlikeNozick,I'mnotpresentingananalysisofpropositionalknowledge.ButhisgrandmothercasealsoseemstobeanexceptiontothegeneralinclinationSCAappealsto:Herewe'renotatallinclinedtothinkthegrandmotherdoesn'tknowher

24Precisely,whatNozickdoesisthis:Heanalyzesthetechnicallocution'Sknows,viamethodM,thatp',andtheninturnanalyzestherelationofS'sknowingthatpintermsofthistechnicallocution.TherevisedthirdconditionI'vedisplayedispartofNozick'sattempttoanalyzethetechnicallocution.

Page 24: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐22‐

grandsoniswell,eventhoughitcanseemthatifheweren'twell,shewouldstillbelievehewas.ThegeneralizationSCAutilizessaysthatwetendtojudgethatSdoesn'tknowwhereSdoesnotsatisfyNozick'sthirdconditionforknowledge.OnepossibilityhereistofollowNozickverycloselybymodifyingthatgeneralizationsothatitreferstoNozick'smodified,ratherthanhisoriginal,thirdcondition,andthus,likeNozick,explicitlyrelativizingouraccounttothemethodbywhichSbelievesthatP.

Often,though,contexttakescareofthisforus.Eventooneawareofthelikelihoodthatthegrandmother'sfamilywouldhavekeptherinthedarkabouthergrandson'sconditionwerehenotwell,itcanseemthatevenNozick'sinitialformulationofthethirdconditionforknowledgeismetbythegrandmother.Ononewayofevaluatingthatsimpleconditional,itseemsthatifthegrandsonwerenotwell,thegrandmotherwouldnotbelievehewaswell.Afterall,she'slookingrightathim!Thestandardpossible‐worldssemanticsforcounterfactualconditionalscanilluminatewhat'sgoingonhere.Whenonesearchesforthepossibleworldsmostsimilartotheactualworldinwhichthegrandsonisnotwell,therespectsinwhichthepossibleworldsaretoresembletheactualworldisahighlycontext‐sensitivematter.Especiallywherethecontextfocusesone'sattentiononthegrandmotherandhercognitiveandrecognitionalabilities,onecanplaceheavyweightuponsimilaritywithrespecttothemethodsheisusingtoarriveatherbelief,andthenitcanseemthatintheclosestworldinwhichthegrandsonisnotwell,she'slookingrightathimandseeingthathe'snotwell,andsodoesnotbelieveheiswell.Onthiswayofevaluatingtheconditional,thegrandmotherdoessatisfyeventheinitialformulationofNozick'sthirdcondition,andshe'snocounter‐exampletothegeneralizationutilizedbySCA.But,inevaluatingthatsimpleconditional,onecanalsostressothersimilarities,particularlyonesinvolvingthepropensitiesandplansofthevariousfamilymembers(orwhateverfactsgroundthejudgmentthatifhergrandsonweren'twell,thegrandmotherwouldbeeffectivelyliedto),toreachtheverdictthatifhewerenotwell,shewouldbelievethathewaswell.

WecansharpenSCAbyspecifyingthatwetendtojudgethatSdoesn'tknowwhenshefailstosatisfyNozick'sinitialformulationof(3),where(3)isevaluatedinsuchawaythatheavyemphasisisputuponsimilaritywithrespecttothemethodofbeliefformationutilizedbyS,or,followingNozick,wecaninsertaspecificationof

Page 25: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐23‐

themethodintotheantecedentof(3).25Butinneithercaseisthistomakeaveryprecisemodification;rather,itmerelyindicatesthedirectioninwhichamorepreciseaccountmightlie,foranysuchuseofthenotionofmethodsofbeliefformationinouraccountinvitesahostofquestions(manyofwhichNozickwrestleswith)involvinghowsuchmethodsaretobespecifiedandindividuated.

7. SCAANDSOMESKEPTICALHYPOTHESESTHATDON’TWORK

CertaininstancesofAIaren'tverypersuasive.ThefirstpremiseoftheargumentcanbequiteunconvincingdespitethefactthatSCApredictsthatwe'dfinditplausible.Suppose,forinstance,thatinanattempttoshowbyAIthatIdon'tknowIhavehands,askepticutilizes,insteadoftheBIVhypothesis,thefollowingsimpleH:IfalselybelievethatIhavehands.TheresultinginstanceofAlseemstopacklittleornomorepunchthanasimpleskeptic'sunsupportedclaimthatIdon'tknowIhavehands.It'satthefirstpremisethatthisill‐fatedinstanceofAlfizzles.ButmybeliefthatIdon'tfalselybelievethatIhavehandsisinsensitive:Ifthisbeliefwerefalse(ifIdidfalselybelievethatIhavehands)Iwouldstillbelieveitwastrue(I'dstillbelievethatIdon'tfalselybelievethatIhavehands).Likewiseinsensitiveismybeliefthatthefollowinghypothesisisfalse:I'manintelligentdogwho'salwaysincorrectlythinkingthatIhavehands.Ifthisbeliefofminewerefalse(ifIweresuchadeludedintelligentdog)I'dstillbelieveitwastrue(I'dstillbelievethatIwasn'tsuchacreature).SoSCA,asithassofarbeenformulated,predictsthatitwillseemtousthattheabovebeliefsdon'tamounttoknowledgeandthatwe'llfindplausiblethefirstpremiseofAlthatresultswhentheabovehypothesesareused.Butinfact

25Thesearenotidenticalmodifications.Onthefirstoption,similaritywithrespecttomethodisweightedheavily,butcanbeoutweighedbyotherfactors.Thus,evensoevaluated,themostsimilarworld(s)inwhichtheantecedentoftheoriginal(3)aretruemaybeworldsthatdivergefromtheactualworldwithrespecttothemethodbywhichScametobelievethatP.Bycontrast,onthesecondoption,sincethemethodbywhichSbelievesthatPbecomespartoftheantecedentoftheconditionalwe'reevaluating(themodified(3)),theclosestpossibleworld(s)inwhichthatantecedentistruecannotbeworldsthatdivergefromtheactualworldwithrespecttomethod.

Page 26: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐24‐

theseinstancesofAl'sfirstpremisearefarfromconvincing.AsopposedtotheBIVhypothesis,itseemsthatonedoesknowthatthedeludeddoghypothesisandthesimplefalsebeliefhypothesisarefalse.

Again,themainpointtomakehereisthatSCA'sgeneralizationneedn'tbeexceptionlesstobeexplanatory.WhileamorepreciselyChisholmedrefinementofSCAmightnothavethenegationsoftheseineffectiveH'sasinstancesofthosepropositionsitsayswetendtojudgewedon'tknow,I'llherejustmakeapreliminaryobservationastowhatmightbegoingwrong.Partoftheproblemwiththese‘hypotheses’isthattheydon'tgiveusmuchofanideaofhowIcometohavethefalsebelieftheyassigntome.Hypothesesaresupposedtoexplain;skepticalhypothesesshouldexplainhowwemightcometobelievesomethingdespiteitsbeingfalse.ThefirstofthesehypothesessimplystipulatesthatI'mwrongaboutmyhavinghands,withoutindicatinghowIcametobesosadlymistaken.ThesecondaddstothefirstthatI'madog,whichaddslittletoourunderstandingofhowmymistakeabouthavinghandscameabout.Bycontrast,whenweencountereffectiveskepticalhypotheses,wehavesomeunderstandingofhow(ifHistrue)wehavecometofalselybelievethatO.IfeitherofourineffectivehypothesesisfilledinsoastomakeitcleartoushowIcametofalselybelieveIhavehands,itbecomeseffective.

SCA'sgeneralizationwasthis:WetendtojudgethatSdoesn'tknowthatPwhenwethinkthatS'sbeliefthatPisinsensitive(whenwethinkthatSwouldbelievePevenifPwerefalse).ThelimitationofSCA'sgeneralizationthat'ssuggestedbythesecasesisthis:Wedon'tsojudgeourselvesignorantofPwherenot‐Pimpliessomethingwetakeourselvestoknowtobefalse,withoutprovidinganexplanationofhowwecametofalselybelievethisthingwethinkweknow.Thus,IfalselybelievethatIhavehandsimpliesthatIdon'thavehands.SinceIdotakemyselftoknowthatIhavehands(thisbeliefisn'tinsensitive),andsincetheaboveitalicizedpropositiondoesn'texplainhowIwentwrongwithrespecttomyhavinghands,I'lljudgethatIdoknowthatpropositiontobefalse.Butthisagainisjustapreliminarystatement,andthere'sroomforalotmorerefinementhere.Whatweneednowissomeassurancethatwe'reheadedintherightdirection.

Page 27: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐25‐

8. SCACONFIRMED

Suchassuranceistobefoundbyconsideringwhatitwouldtaketomakeitseemtousthatwedoknowskepticalhypothesestobefalse.

Butlet'sfirstreconsiderthelotterycase.Asnotedaboveinsection5,wearepuzzlinglyreluctanttoclaimknowledgeincertainlotterysituations.TheexplanationprovidedbySCAforthisphenomenonisintuitivelyappealing:Itdoesseemthatthefactthatwewouldbelievethatwewerelosersevenifwewerewinnersislargelywhat'sbehindourjudgmentthatwedon'tknowwe'relosers.SCAreceivesfurtherpowerfulsupportwhenweconsiderthegroundsthatdoseemtoussufficientforknowledgeofone'sbeingaloser.Inthelotterysituation,evenaveryminutechanceofbeingwrongseemstodepriveoneofknowledge.Butifwe'regoingtoworryaboutevensuchminutechancesoferror,thenwhydoesitseemthatyoudoknowyou'realoserafterthewinningnumberhasbeenannouncedontheradioandyou'vecomparedthenumbersonyourticketwiththesadlydifferentnumbersannounced?Afterall,radioannouncementscanbeinerror;whatyou'rehearingmaynotbearealradioannouncementbutthevoiceofafriendwho'sriggedupapracticaljoke;youmightbesufferingfromsomeweirdmomentaryvisualillusionandmisreadingthenumbersonyourticket;andsoforth.Allveryremotepossibilities,tobesure.But,sincewe'realreadycountenancingeventhemostminutechancesoferror,whydon'tthesepossibilitiesrobusofknowledgeevenaftertheannouncementhasbeenmadeandheard?

SCA'sexplanationofwhywedon'tthinkweknowbeforetheannouncementismadeisthatweatthattimejudgethatifweweren'tlosers,we'dstillbelievethatwewere.Notethatonceyou'veheardtheannouncementofthewinningnumbersandcomparedthemwiththenumbersonyourticket,itnolongerseemsthatifyouhadbeenthewinner,you'dbelieveyouwerealoser.Rather,wejudgethatinthatcaseyou'dnowbelieveyouwerethewinnerorwouldatleastbesuspendingjudgmentasyoufranticallydouble‐checkedthematch.It'sveryimpressivethattheveryoccurrencethatwouldsufficetomakeitseemtousthatyoudoknowyou'realoser(theradioannouncement)alsoreversesourjudgmentregardingthetruthoftheconditionalappealedtoinSCAtoexplainwhyitseemstousthatyoudon'tknow

Page 28: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐26‐

beforetheannouncementismade.TheoccurrencewhichgetsustojudgethatweknowherealsoremoveswhatSCApositsastheblocktoourjudgingthatweknow.ThisisanindicationthatSCAhascorrectlyidentifiedtheblock.

SCAsimilarlyprovidesaveryintuitivelyappealingexplanationforwhyitseemstousthatwedon'tknowthatskepticalhypothesesarefalse,aswasalsonotedinsection5.Itagainreceivespowerfulfurtherconfirmationaswelooktocasesinwhichoneseeminglydoesknowthataskepticalhypothesisdoesn'tobtain(casesinwhichskepticalhypothesesthatareordinarilyeffectivefailtobeeffective).TheboastfulzoologistIintroducedtowardtheendofsection3,itseems,knowsthattheanimalsinthezebracagearenotcleverlypaintedmules,whileI,itseems,donot.Buttheveryanatomicalknowledgethatseeminglyenableshimtoknowthey'renotpaintedmulesalsohastheconsequencethatiftheanimalswerecleverlypaintedmules,thezoologist,unlikeme,wouldnotbelievethattheyweren't.AndalthoughIdon'tseemtoknowthey'renotpaintedmulessimplybylookingatthem,Icould,itseems,gettoknowthisifIundertooksomespecialinvestigation—perhaps,ashasbeensuggestedintheliterature(Stine1976:252),oneinvolvingpaintremover.Whichspecialinvestigationswoulddothetrick(andunderwhichcircumstanceswouldthey)?Asurveyofvariousscenariosyieldsanimpressivecorrelation:Theinvestigationsthatwouldseeminglyallowmetoknowthattheanimalsaren'tpaintedmuleswouldalsoaffectourjudgmentastothetruthvalueofthesubjunctiveconditionalsocriticaltoSCA.OnceIhavecompletedtheinvestigation,itseemsthatI,likethezoologist,wouldnotbelievethattheanimalsweren'tpaintedmulesifinfacttheywere.Likewise,bycheckingappropriatelyindependentsources,IcouldgetmyselfintoapositioninwhichIseeminglywouldknowthatthenewspaperisn'tmistakenaboutwhethertheBullswonlastnight.Butthechecksthatwouldseeminglyallowthisknowledgewouldalsomakeitseemthatifthepaperweremistaken,Iwouldnotbelieveitwasn't.Againandagain,SCApositsacertainblocktoourjudgingthatweknow,andthechangesthatwouldclearthewayforourjudgingthatweknowalsoremovethisblock.ThismakesitdifficultnottobelievethatSCAisatleastroughlycorrect.

InthecaseoftheBIVhypothesis,it'shardtotestSCAinthisway,forit'sdifficulttoimagineasituationinwhichitseemsasubjectdoesknowthatshe'snotaBIV.ButthisonlyconfirmsSCA:Whileit'sdifficulttoimagineasituationinwhich

Page 29: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐27‐

oneseemstoknowthatone'snotaBIV,it'slikewisedifficulttoimaginecircumstancesinwhichtheblockSCApositsisremoved.It'sdifficult,thatis,toimagineasituationinwhichsomeonebelievesthey'renotaBIVbutinwhichtheconditionalIfSwereaBIV,thenSwouldbelieveshewasn'taBIVisn'ttrue.For,astheBIVhypothesisisformulated,one'sbrainiselectrochemicallystimulatedsothatonehaspreciselythosesensoryexperiencesoneactuallyhashad.Butwouldn'tonethenhaveformedpreciselythosebeliefsthatoneactuallyhasformed,includingthebeliefthatone'snotaBIV?

UnlikethatinvolvedintheRelevantAlternativesSolution,thepresentexplanationfortheplausibilityofAl'sfirstpremisecan'tbehappilyreversed:TryingtoaccountfortheplausibilityofthesubjunctiveconditionalIfHweretrue,Iwouldbelieveitwasfalse,byreferencetothe(presumed)factthatIdon'tknowthatHisfalsecertainlyseemstogetthingsbackwards.MuchbettertofollowtheproposedNozickeanrouteinexplainingtheplausibilityofdenyingknowledgebyreferencetotheconditional.

Further,aswasthecasewithnotknowing,theinvestigationsthatwouldreverseourotherpessimisticjudgmentsregardingyourstandingvis‐a‐visaskepticalhypothesiswouldalsoputyouinapositiontosaythatyouwouldn'tbelievethehypothesisisfalseifitweretrue.Thus,forinstance,tomakeitseemthatyoucantellthatthoseanimalsaren'tpaintedmules,youmustputyourselfinsuchapositionthatyouwouldn'tbelievetheyweren'tiftheywere.And,aswasthecasewithnotknowing,noneoftheseexplanationsbysubjunctiveconditionalsseemshappilyreversible.

Itseemsthatthisexplanation,SCA,fortheplausibilityofAl'sfirstpremisemustbe(atleastroughly)correctand,therefore,thatitpointstopartofthesolutiontoourpuzzle.

Indeed,somereaderswillwonderwhyIhaveclaimedonlythatourgeneraltendencynottocountinsensitivebeliefsasinstancesofknowledgeexplainsthatpremise'splausibilityandhavestoppedshortofacceptingsensitivityasanecessary

Page 30: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐28‐

conditionforknowledge26andthereforesimplyendorsingthatfirstpremiseastrue.Butwhilewe'vejustseenstronggroundsforsimplyacceptingAl'sfirstpremise,therearealsostronggroundsforacceptingAl'ssecondpremiseandforacceptingthedenialofitsconclusion.Wehavetostopshortsomewhere;wecan'tsimplyacceptallthreemembersofthistriadastrue.Tosolvethispuzzle,I'llclaimthatAl'sfirstpremise,whilenotsimplytrue,istrueaccordingtounusuallyhighstandardsforknowledge.But,I'llargue,mysolutionexplainswhythatpremiseseemstrueand,moregenerally,whysensitivityseemsnecessaryforknowledge.Ifmysolutionprovidesthebestexplanationforhowallthreemembersofourpuzzlingtriadseemtrue,thatwillbegoodreasonforstoppingshortwheremysolutiontellsusto,ratherthanwhereoneofitsinferiorrivals—boldskepticism,forexample—tellsusto.

9. NOZICK’SOWNSOLUTIONANDTHEABOMINABLECONJUNCTION

Nozick'sowntreatmentofAI,fromwhichSCAwasabstracted,fails.ThistreatmentisbasedonNozick'saccountofknowledgeastrue,sensitivebelief,where,veryroughly,one'struebeliefthatpissensitivetothetruthvalueofpifonewouldnothavebelievedthatpifphadbeenfalse.27Thus,Nozick'streatmentofAIinvolvesacceptingtheskeptic'sfirstpremise.But,atthesametime,andmuchmoreunfortunately,italsoinvolvesdenyingthesecond.Youdon'tknowthatyou'renotaBIV,Nozickclaims,becauseanybeliefyoumighthavetothiseffectisinsensitive:

26Or,giventheexceptionstothegeneraltendencythatwe'vediscussedinsections6and7,whyIhaven'tacceptedthatsomeproperlyChisholmedrefinementofthesensitivityrequirement(whichhasasinstancesofitconvincinginstancesofAl'sfirstpremise)isnecessaryforknowledge.

27ThoughthisstatementofNozick'saccountofknowledgeisrough,thatwillnotaffectmytreatment,whichwouldapplyequallywelltoNozick'sfullaccount.I'veskippedentirelyNozick'sfourthconditionforknowledge,butIbelievethisfourthconditiontoberedundant,anyway:Itautomaticallyholdswhenevertruebeliefispresent.Also,asI'vealreadynoted,Nozicktakesaccountofthemethodofbeliefformationinhisfinalversionofthethirdcondition.Thesamethinghappenswiththefourth.

Page 31: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐29‐

Youwouldhaveheldthisbeliefevenifitwerefalse(evenifyouwereaBIV).Bycontrast,Nozickclaims,yourbeliefthatyouhavehandsisasensitivebelief:Ifitwerefalse—ifyoudidn'thavehands—youwouldnotholdit.Soyoudoknowyouhavehandseventhoughyoudon'tknowthatyou'renotaBIV.Theskeptic'smistake—thesecondpremise—issupposingthatyoucanknowyouhavehandsonlyifyoualsoknowthatyou'renotaBIV.

OrsoNozickclaims.ThisisnottheplaceforageneralevaluationofNozick'sanalysisofpropositionalknowledge,soletusconfineourselvestotheresultsofthisanalysisasappliedtothebeliefsinquestioninAI.HereNozick'saccountdoesverywellinissuingtheintuitivelycorrectverdictfortherelevantparticularjudgmentsregardingwhatisknownandwhatisnot.Mostofuswouldjudgethatwedoknowsuchthingsasthatwehavehands,andthisisNozick'sverdict.And,whenaskepticalhypothesisiswellchosen,itdoesseemquiteplausibletomostofusthatwedon'tknowthatitdoesn'tobtain.Buttherearethreerelevantissuestoourpuzzle:IsthefirstpremiseofAItrue?Isthesecondpremisetrue?Istheconclusiontrue?Andit'seasytoendorsetheintuitivelycorrectanswertotwooutofthethreequestionsifyou'rewillingtotaketheimplausiblestandontheremainingone.

Nozicktakeshisimplausiblestandontheissueofthesecondpremise,denyingitinthefaceofitsevidentintuitiveappeal.28Acceptinghistreatment

28At2O5‐6Nozickadmitsthisappeal,andlaterhewrites,‘Thus,ifournotionofknowledgewasasstrongaswenaturallytendtothink(namely,closedunderknownlogicalimplication)thentheskepticwouldberight.(Butwhydowenaturallythinkthis?Furtherexplorationandexplanationisneededoftheintuitiverootsofthenaturalassumptionthatknowledgeisclosedunderknownlogicalimplication)’(242).

Nozickisquitehardonanti‐skepticswhochooserathertodenythefirstpremise;hewrites:‘Theskepticassertswedonotknowhispossibilitiesdon'tobtain,andheisright.Attemptstoavoidskepticismbyclaimingwedoknowthesethingsareboundtofail.Theskeptic'spossibilitiesmakeusuneasybecause,aswedeeplyrealize,wedonotknowtheydon'tobtain;itisnotsurprisingthatattemptstoshowwedoknowthesethingsleaveussuspicious,strikeusevenasbadfaith’(2O1).ButsimilarremarkscouldbemadeaboutNozick.AsNozickhimselfadmits,thesecondpremisehasitsownintuitiveappeal.Sowhynotsaythatwhatwe‘deeplyrealize’isthatifyoudon'tknowthatyou'renotaBIV,thenyoudon'tknowyouhavehands,andthattheskepticisrightaboutthis?Nozick'sdenialofthesecondpremiseleavesmeaboutas‘suspicious’asdoesadenialofthefirst,andthoughNozick'sdenialdoesn'tstrikemeasaninstanceofbadfaith,denialsofthefirstpremiseseemnobettercandidatesforthatcharge.

Page 32: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐30‐

involvesembracingtheabominableconjunctionthatwhileyoudon'tknowyou'renotabodiless(andhandless!)BIV,still,youknowyouhavehands.Thus,whilehisaccountdoesquitewellontherelevantparticularintuitionsregardingwhatisandisn'tknown,ityieldsanintuitivelybizarreresultonthecomparativejudgmentthesecondpremiseembodies.29

Aspromised,Iwon'thererehearsethepowerfulobjectionstoNozick'sanalysisofpropositionalknowledgethathavebeenputforward,30but,assumingthatthisanalysisisn'tindependentlyconvincingbeforeweturntotheproblemofskepticalhypotheses,31we'releftwithlittlereasontofollowNozickinchoosingtotakeanimplausiblestandpreciselywherehehasratherthansomeplaceelse.

Thisleavesusinabind.For,aswesawinsections5and8above,SCAisquitepowerful.Thatexplanationisthatwerealizethatanybeliefwemighthavetotheeffectthatan(effective)skepticalhypothesisdoesn'tobtainisinsensitive,andwe'reinclinedtothinkthatinsensitivebeliefsdon'tconstituteknowledge.HowcanweappropriatethatexplanationwithoutfollowingNozickinhavingtoimplausiblydenythesecondpremiseofAIandembracetheabominableconjunction?

29WhatareNozick'sgroundsforrejectingthesecondpremise?Nozicknotesthatthepremiseisaninstanceofaverygeneralprincipletotheeffectthatknowledgeisclosedunderknownimplication(seenote33,below).Afteradmittingthattheclosureprincipleseemstrue(2O5‐6),Nozickclaimsthatit'swrong,andhisreasonsforthisclaimaremadeentirelyfromwithinhisanalysisofknowledge:Givenhisanalysis,knowledgewon'tbeclosed(seeespecially2O6‐8).SoNozickisrelyingonhisanalysistoshowusthatthesecondpremiseisfalsedespiteitsintuitiveappeal.Andindeed,Nozickhasdevelopedanddefendedhisanalysisofknowledge(inpart1ofchapter3)beforeheappliesittotheissueofskepticism(inpart2).

30Unfortunately,whatisperhapsthemostpowerfulattackonNozick'stheoryofknowledge,madebySaulKripkeinlectures,circa1985,hasnot,tothebestofmyknowledge,founditswayintoprint.ForthoseinterestedincriticalliteratureonNozick,agoodplacetostartiswithForbes1984andseveraloftheessaysinLuper‐Foy1987.Forstillfurtherreading,Luper‐Foy1987containsanexcellentbibliography.

31Asremarkedinnote29,NozickdependsheavilyontheindependentplausibilityofthisanalysistoprovidethemomentumforhistreatmentofAI.

Page 33: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐31‐

10. STRENGTHOFEPISTEMICPOSITIONANDAI’SSECONDPREMISE

Here'show:byincorporatingSCAintoacontextualistsolutiontoourpuzzlethatavoidssuchafumblingofAl'ssecondpremise.Indeed,Iproposeaverystrongendorsementofthatsecondpremise.

Recallthataccordingtocontextualisttheoriesofknowledgeattributions,howstrongasubject'sepistemicpositionmustbetomaketrueaspeaker'sattributionofknowledgetothatsubjectisaflexiblematterthatcanvaryaccordingtofeaturesofthespeaker'sconversationalcontext.Centraltocontextualism,then,isthenotionof(relative)strengthofepistemicposition.Inpresentinganddefendingcontextualism,I'vefoundthatmostlistenersfeelthattheyunderstandprettywellwhat'smeantwhenIclaim,forinstance,thatsometimesthestandardsforknowledgearehigherthanusual,orthatinsomeconversationalsituationsone'sepistemicpositionmustbestrongerthaninotherstocountasknowing.Butitwouldbegoodtoclarifythisimportantnotionofstrengthofepistemicpositionasbestwecanby,forinstance,supplyinganintuitivetestforwhenoneepistemicpositionisstrongerthananother.Thebestsuchdeviceisthatofcomparativeconditionals.OnecanhaveavarietyofgroundsforassentingtoconditionalslikeIfMugsyistall,thenWiltistall,andIfWiltisnottall,thenMugsyisnottall.ButoneverygoodbasisforassentingtotheseconditionalsisthecomparativeknowledgethatWiltisatleastastallasMugsy.Likewise,whereSisaputativesubjectofknowledge,PisatruepropositionthatSbelieves,andAandBaresituationsinwhichSisfound,wecanhavesimilarlycomparativegroundsforassentingtoconditionalsoftheformIfSknowsthatPinA,thenSknowsthatPinB.Insuchacase,thecomparativegroundsforourassentisourrealizationthatSisinatleastasstronganepistemicpositionwithrespecttoPinsituationBasheisinwithrespecttothatsamepropositioninsituationA,andthiscomparativeconditionalservesasagoodintuitivetestforthatcomparativefact:Itbringsthatfacttolight.

So,forinstance,toborrowsomeexamplesfromAlvinGoldman(1976),letHenrybeoursubject,andletWhatHenryisseeingisabarnbethethingHenryputativelyknows.BothinsituationF(for‘fakes’)andinsituationN(‘nofakes’),Henryisdrivingthroughthecountrysideand,havingnoreasontothinkthere's

Page 34: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐32‐

anythingunusualgoingon,veryfirmlybelieves,andtakeshimselftoknow,thattheobjecthe'sseeingisabarn.Andindeed,inbothcases,itisabarn.ButinF,unbeknownsttohim,Henryisinanareathatisfilledwithveryconvincingfakebarns—papier‐mâchèbarnfacades.Infact,wemaysupposethatHenryhasjustbeenfooledmorethantwentytimesbysuchfakes,althoughhe'snowlookingattheonlyactualbarnformilesaround,andsothistimetrulybelievesthatwhathe'sseeingisabarn.NisexactlylikeF,exceptthattherearenofakesinthearea—thethingsHenryhastakentobebarnshaveallactuallybeenbarns.Withregardtotheseexamples,theconditionalIfHenryknowsinF,thenheknowsinNseemstogetthecomparisonright,indicatingthatHenry'sinatleastasstronganepistemicpositioninsituationNasheisinsituationF.TheevidentfailureofIfHenryknowsinN,thenheknowsinFtogetthecomparisonrightshowsthatHenry'snotinasstrongapositiontoknowinFasinN.Together,theseresultsindicatethatHenry'sinastrongerepistemicpositioninNthaninF.

AsisimportanttoourdiscussionofAl'ssecondpremise,comparativeconditionalscansimilarlybeusedtotesttherelativestrengthofepistemicpositionofasinglesubjectwithrespecttodifferentpropositionsthatsubjectbelievesinthesamesituation:Thus,theintuitivecorrectnessofIfSknowsthatP,thenSknowsthatQandIfSdoesn'tknowthatQ,thenSdoesn'tknowthatPcanindicatethatSisinatleastasstronganepistemicpositionwithrespecttoQasshe'sinwithrespecttoP.32

Sometimesnoclearverdictresultswhenweattempttoevaluateaconditionalinthiscomparativeway,forthegoodreasonthatit'sunclearhowthetwoepistemicpositionswe'reevaluatingcomparewithoneanother.Thus,ifwecompareasituationinwhichHenryhasagoodlookatthebarnbutinwhichthereareacoupleoffakebarnsseveralmilesawaythatHenryhasn'tencounteredwithasituationinwhichtherearenofakesatallinHenry'svicinitybutinwhichhedoesn'thavequiteasgoodalookatthebarn,therelevantconditionalscanbedifficulttoevaluate.But,

32And,ofcourse,suchconditionalscanbeusedtomakeallmannerofothercomparisons:comparativestrengthoftheepistemicpositionsoftwodifferentsubjectswithrespecttothesamepropositionorwithrespecttodifferentpropositions,thestrengthoftheepistemicpositionofasubjectwithrespecttoonepropositioninonesituationascomparedwiththatsamesubject'sepistemicpositionwithrespecttoadifferentpropositioninadifferentsituation,etc.

Page 35: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐33‐

inmanyinstances,someoftherelevantconditionalsareclearlytrueoncomparativegrounds.

SuchisthecasewithinstancesofAl'ssecondpremise,wheretheskepticalhypothesisiswellchosen.Theyseemtrueandaretrue,Isuggest,forjustthiscomparativereason:Aswerealize,weareinatleastasgoodapositiontoknowthatthehypothesisisfalseaswe'reintoknowthetargetedpieceofpresumedordinaryknowledge.33Let'slookbrieflyatsomeinstances.Recallthefollowingepistemologicallyperplexingpairsofpropositions:

33Asiswellknown,instancesofAl'ssecondpremiseareofteninstancesoftheprinciplethatknowledgeisclosedunderknownlogicalimplication:Kp&K(pentailsq)→Kq.(InthenextparagraphIexplainwhythisisnotalwaysthecase,atleastwhentheclosureprincipleisn'tstrengthenedastheredescribed.)Asisalsowellknown,thereareexceptionstotheprinciplesoformulated,anditmighttakealotoftinkeringtogetitexactlyright.But,asNozick,thearchdenierofclosure,putsit,‘Wewouldbeill‐advised,however,toquibbleoverthedetailsofP[theprinciplethatknowledgeisclosedunderknownlogicalimplication].Althoughthesedetailsaredifficulttogetstraight,itwillcontinuetoappearthatsomethinglikePiscorrect’(1981:2O5).Nozickgoesontoclaimthatthisappearanceisdeceiving.IbelievethatsomethinglikePiscorrect,butthatdoesn'tcompetewithmypresentaccountofAl'ssecondpremise:Whenaconditionalisaninstanceoftheproperlyformulatedclosureprinciple,therelevantcomparativefactinvolvingstrengthofepistemicpositionholds.SeeBrueckner1985forargumentsthatthedenialofknowledgeclosureprinciples‘isnotafruitfulanti‐skepticalproject’(112).

Whilerestrictionswillhavetobeputontheclosureprinciplethatwillweakenitincertainrespects,theremaybeotherrespectsinwhichitcanbestrengthened.SomeinstancesofAl'ssecondpremiseareconvincingeventhoughHiscompatiblewithO.Forinstance,theBIVhypothesisseemstounderminemyputativeknowledgeofI'minHoustonaswellasofIhavehands,but,ofcourse,thatI'mabodilessBIViscompatiblewithmybeinginHouston.PerhapsifSistoknowthatP,thenSmustknowthatnot‐QforanyQ(buthererestrictionsmustbeadded)suchthatifQweretrue,SwouldnotknowthatP.Thus,therangeofQsthatmustbeknownnottoobtainmaybebroadenedsoastoincludenotonlypropositionsthatareincompatiblewithP,butalsootherssuchthatiftheywerethecase,thenSwouldn'tknowthatP.ThoseQsthatareincompatiblewithPitselfwillthenbeseenasspecialcasesofthosethatareatoddswithS'sknowingthatP.BarryStrouddiscussesastrongerclosureprinciplesuchasthisinhis1984(25‐30).

Page 36: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐34‐

not‐H O

I'mnotaBIV. Ihavehands.

Thoseanimalsaren'tjust Thoseanimalsarezebras.cleverlypaintedmules.

Thepaperisn'tmistakenabout TheBullswonlastnight.whethertheBullswonlastnight.

Givennaturalbackgroundassumptions,wecansensethatthefollowingcomparativefactholdsforeachoftheabovepairs:IaminnobetterapositiontoknowthatOthanIamintoknowthatnot‐H.ThiscomparativefactisrevealedineachcasebythehighlyplausibleconditionalthatisAl'ssecondpremise:IfIdon'tknowthatnot‐H,thenIdon'tknowthatO.Closelytiedtothatcomparativefactineachcaseistherelatedandintuitivelycompellingrealizationthatitwouldbenowisertobetone'simmortalsoulonO'sbeingtruethantobetitonnot‐H'sbeingtrue.

Iproposethentoaccepttherelevantconditionalwithrespecttoeachoftheabovepairs,andtoacceptotherconvincinginstancesofAl'ssecondpremise.Indeed,theseconditionalsaretrueregardlessofhowhighorlowthestandardsforknowledgeareset.JustasthecomparativefactthatWiltisatleastastallasMugsyhastheresultthattheconditionalIfWiltisnottall,thenMugsyisnottallwillbetrueregardlessofhowhighorlowthestandardsfortallnessareset,sothecomparativefactthatI'minatleastasstronganepistemicpositionwithrespecttonot‐HasI'minwithrespecttoOwillresultinIfIdon'tknowthatnot‐H,thenIdon'tknowthatObeingtrueregardlessofhowhighorlowthestandardsforknowledgeareset.Thus,wewillneverhavetofollowNozickinacceptingtheabominableconjunction:thatconjunctionisfalseatanyepistemicstandard.

WiththatringingendorsementofAl'ssecondpremiseanchoredfirmlyinplace,wecanreturntothefirstpremise,hopingtoincorporateSCAintoacontextualistaccountofthatpremise'splausibility.

Page 37: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐35‐

11. STRENGTHANDSENSITIVITY

Ashasbecomeveryapparent,twonotionsthatarecentraltomyattempttosolveourpuzzleare,ontheonehand,theNozickeannotionofthesensitivityofbeliefsand,ontheother,thenotionofstrengthofepistemicposition.Whilebothnotionsstandinneedofagooddealofsharpeningandexplanation(onlysomeofwhichthey'llreceivehere),we'vealreadyobtainedinterestingresultsapplyingthemtotheepistemologicallyperplexingpairsofpropositionsdisplayedabove.Ineachcase,one'sbeliefinOissensitive,whileone'sbeliefinnot‐Hisinsensitive.Yet,atthesametime,oneisinatleastasstronganepistemicpositionwithrespecttonot‐HasoneisinwithrespecttoO.

Foreachofthesecondandthirdpairsofpropositions,onecouldgatherfurtherevidence,strengthenone'sepistemicpositionwithrespecttobothnot‐HandO,andmakeevenone'sbeliefthatnot‐Hsensitive.Butevenbeforethisfurtherevidenceisgathered,one'sbeliefthatOisalreadysensitive,despitethefactthatoneisinnostrongeranepistemicpositionwithrespecttothisOthanoneisinwithrespecttonot‐H.(Withrespecttothefirstpairofpropositions,itisdifficulttoimagineasituationinwhichoneisinsuchastrongpositionwithrespecttoone'snotbeingaBIVthatthisbeliefissensitive.)

Thisleadsustoanimportantinsightregardingskepticalhy‐potheses:One'sepistemicpositionwithrespecttopropositionstotheeffectthatskepticalhypothesesdon'tholdmustbestrongerthanitiswithrespecttoother,moreordinarypropositions(e.g.,ouraboveOs)ifbeliefinsuchpropositionsistobesensitive.

Anexplanationofourtwocentralnotionsintermsofpossibleworldswillprovideapartialandquiterough‐and‐ready,butstillsomewhatenlightening,pictureofhowthissituationcanarise.AnimportantcomponentofbeinginastrongepistemicpositionwithrespecttoPistohaveone'sbeliefastowhetherPistruematchthefactofthematterastowhetherPistrue,notonlyintheactualworld,butalsoattheworldssufficientlyclosetotheactualworld.Thatis,one'sbeliefshouldnotonlybetrue,butshouldbenon‐accidentallytrue,wherethisrequiresone's

Page 38: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐36‐

beliefastowhetherPistruetomatchthefactofthematteratnearbyworlds.Thefurtherawayonecangetfromtheactualworld,whilestillhavingitbethecasethatone'sbeliefmatchesthefactatworldsthatfarawayandcloser,thestrongerapositiononeisinwithrespecttoP.(Recallingtheresultsofsection6,weshouldremembereithertorestrictourattentionsolelytothoseworldsinwhichthesubjectusesthesamemethodofbelief‐formationsheusesintheactualworld,ortoweighsimilaritywithrespecttothesubject'smethodveryheavilyindeterminingtheclosenessofpossibleworldstotheactualworld.)Ifthetruth‐trackingofone'sbeliefastowhetherPextendsfarenoughfromactualitytoreachtheclosestnot‐Pworlds,thenonedoesn'tbelievethatPinthoseclosestnot‐Pworlds,andone'sbeliefthatPissensitive.Buthowfarfromactualitymusttruth‐trackingreach—howstronganepistemicpositionmustonebein—tomakeone'sbeliefthatPsensitive?That,ofcourse,dependsonhowdistantfromactualitytheclosestnot‐Pworldsare.

ConsidermybeliefthatIhavehands.Ibelievethisattheactualworld,andit'strue.What'smore,intheothernearbyworldsinwhichIhavehands,IbelievethatIdo.Therearealso,atleastinmyowncase,somealarminglycloseworldsinwhichIdon'thavehands.TheseincludeworldsinwhichIlostmyhandsyearsagowhileworkingonmyuncle'sgarbagetruck.Intheclosestofthesenot‐Pworlds,I'mnowfullyawareofthefactthatI'mhandless,andmybeliefastowhetherIhavehandsmatchesthefactofthematter.MybeliefastowhetherIhavehandsdoesn'tmatchthefactinvariousworldsinwhichI'maBIV,ofcourse,buttheseareverydistant.Whiletherearecloserworldsinwhichthematchfails,itseemsthatinafairlywiderangeofworldssurroundingtheactualworld,mybeliefastowhetherIhavehandsdoesagoodjobofmatchingthefactofthematter.Thus,I'minaprettystrongepi‐stemicpositionwithrespecttothatmatter.

NowletPbeI'mnotaBIV.Wherenot‐P(here,IamaBIV)isquiteremote,onecanbeinaquitestrongepistemicpositionwithrespecttoPmerelybybelievingthatPinallthenearbyworlds.AsIdobelievethisPinsuchnearbyworlds,I'minaprettystrongepistemicpositionwithrespecttothisP.Thiscanoccur,andinmycase,doesoccur,eventhoughone'sbeliefastowhetherPdoesn'tmatchthefactofthematterintheclosestnot‐Pworlds:Sinceeventheclosestofthenot‐Pworldsarequitedistant,one'sbeliefastowhetherPneedn'tmatchthefactofthematterthatfarfromtheactualworldforonetobeinaquitestrongpositionwithrespecttoP.

Page 39: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐37‐

Butforone'sbeliefthatPtobesensitive,onemustnotbelievethatPintheclosestnot‐Pworlds.Sinceskepticalhypothesestendtofastenonsomewhatremote(andsometimesveryremote)possibilities,then,onecanbeinarelatively(andsometimesavery)strongpositionwithrespecttobeliefstotheeffectthattheydon'tobtain(sinceone'sbeliefastowhethertheyobtainmatchesthefactofthematteroverawiderangeofworldsclosesttotheactualworld),whilethesebeliefsremaininsensitive(sinceonewouldstillbelievethatthehypothesesdidn'tobtainintheclosestworldsinwhichtheydoobtain).Bycontrast,wherePissuchthattherearebothPandnot‐Pworldsveryclosetotheactualworld,one'sbeliefthatPmustbesensitive(onemustnotbelievethatPintheclosestnot‐Pworlds)inorderforonetobeinevenaminimallystrongepistemicpositionwithrespecttoP,and,conversely,oneneedn'tbeinaverystrongpositionforone'sbelieftobesensitive.

12. THERULEOFSENSITIVITYANDTHEBEGINNINGSOFANEW

CONTEXTUALISTSOLUTION

Theimportantinsightregardingskepticalhypotheses—thatone'sepistemicpositionwithrespecttopropositionstotheeffectthatskepticalhypothesesdon'tholdmustbestrongerthanitiswithrespecttootherpropositionsbeforebeliefsinsuchpropositionscanbesensitive—suggestsanewcontextualistaccountofhow,inpresentingAl,theskepticraisesthestandardsforknowledge.Let'scalltheconversationalrulethisnewaccountpositsasthemechanismbywhichtheskepticraisesthestandardsforknowledgethe‘RuleofSensitivity.’Althoughamoregeneralformulationofthisruleisdesirable,Iwillherestateitinsuchawaythatitappliesonlytoattributions(anddenials)ofknowledge,sincesuchapplicationsarewhat'sneededtoaddressthepresentpuzzle.34Solimited,ourruleissimplythis:Whenitis

34Introducingaskepticalhypothesisintoaconversationinanynumberofwaysotherthaninattributionsanddenialsofknowledgecanseemtoraisethestandardsforknowledge.Forinstance,insteadofarguing,‘Youdon'tknowthatthepaperisn'tmistakenabouttheresultoflastnight'sgame;therefore,youdon'tknowthattheBullswon,’askepticmayurge,‘Considerthisproposition:Thenewspaperismistakenaboutwho

Page 40: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐38‐

assertedthatsomesubjectSknows(ordoesnotknow)somepropositionP,thestandardsforknowledge(thestandardsforhowgoodanepistemicpositiononemustbeintocountasknowing)tendtoberaised,ifneedbetosuchalevelastorequireS'sbeliefinthatparticularPtobesensitiveforittocountasknowledge.WherethePinvolvedistotheeffectthataskepticalhypothesisdoesnotobtain,thenthisruledictatesthatthestandardswillberaisedtoaquitehighlevel,for,aswe'veseen,onemustbeinastrongerepistemicpositionwithrespecttoapropositionstatingthataskepticalhypothesisisfalse—relativetoother,moreordinary,propositions—beforeabeliefinsuchapropositioncanbesensitive.

Astoryintermsofpossibleworldsagainprovidesarough‐and‐ready,butstillperhapsenlightening,pictureofhowtheRuleofSensitivityoperates.Context,I'vesaid,determineshowstronganepistemicpositiononemustbeintocountasknowing.Picturethisrequirementasacontextuallydeterminedsphereofpossibleworlds,centeredontheactualworld,withinwhichasubject'sbeliefastowhetherPistruemustmatchthefactofthematterinorderforthesubjecttocountasknowing.(Giventheresultsofsection6,wemustagainremembereithertorestrictourattentionsolelytothoseworldsinwhichthesubjectusesthesamemethodofbeliefformationsheusesintheactualworld,ortoweighsimilaritywithrespecttothesubject'smethodveryheavilyindeterminingtheclosenessofpossibleworldstotheactualworld.)Callthisspherethesphereofepistemicallyrelevantworlds.Asthestandardsforknowledgegoup,thesphereofepistemicallyrelevantworldsbecomeslarger—thetruth‐trackingofone'sbeliefmustextendfurtherfromactualityforonetocountasknowing.Giventhispicture,theRuleofSensitivitycanbeformulatedasfollows:Whenit'sassertedthatSknows(ordoesn'tknow)thatP,then,ifnecessary,enlargethesphereofepistemicallyrelevantworldssothatitatleastincludestheclosestworldsinwhichPisfalse.

Apowerfulsolutiontoourpuzzleresultswhenwefollowthebasiccontextualiststrategy(seesection2)andutilizethisRuleofSensitivitytoexplain

wonthegame.Now,keepingthatpropositionclearlyinmind,answermethis:DoyoureallyknowthattheBullswon?’Ofcourse,aswiththeRuleofRelevance(seenote22),notjustanymentionofaskepticalhypothesisseemstotriggerthemechanismforraisingthestandardsofknowledgeI'mabouttoarticulate.

Page 41: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐39‐

howthestandardsforknowledgeareraisedbytheskeptic'spresentationofAI.Whilemanynoteworthyfeaturesandvirtuesofthissolutionarebestexplainedbycomparingitwiththeotherproposedsolutionstoourpuzzle,asI'lldoinfollowingsections,thebasicideaofthepresentsolutionisthis.InutilizingAItoattackourputativeknowledgeofO,theskepticinstinctivelychoosesherskepticalhypothesis,H,sothatitwillhavethesetwofeatures:(1)Wewillbeinatleastasstrongapositiontoknowthatnot‐Haswe'reintoknowthatO,but(2)Anybeliefwemighthavetotheeffectthatnot‐Hwillbeaninsensitivebelief(abeliefwewouldholdevenifnot‐Hwerefalse—thatis,evenifHweretrue).Givenfeature(2),theskeptic'sassertionthatwedon'tknowthatnot‐H,bytheRuleofSensitivity,drivesthestandardsforknowledgeuptosuchapointastomakethatassertiontrue.BytheRuleofSensitivity,recall,thestandardsforknowledgeareraisedtosuchalevelastorequireourbeliefthatnot‐Htobesensitivebeforeitcancountasknowledge.Sinceourbeliefthatnot‐Hisn'tsensitive(feature(2)),thestandardsaredrivenuptosuchalevelthatwedon'tcountasknowingthatnot‐H.Andsincewe'reinnostrongeranepistemicpositionwithrespecttoOthanwe'reinwithrespecttonot‐H(feature(1)),then,atthehighstandardsputinplacebytheskeptic'sassertionofAl'sfirstpremise,wealsofailtoknowthatO.Atthesehighstandards,theskeptictruthfullyassertshersecondpremise(which,recall,isalsotrueatlowerstandards),andthentruthfullyassertsAl'sconclusionthatwedon'tknowthatO.35ThisaccountsforthepersuasivenessofAI.Butsince,onthisaccount,theskepticgetstotruthfullystateherconclusiononlybyraisingthestandardsforknowledge,AIdoesn'tthreatenthetruthofourordinaryclaimstoknowtheveryOsourknowledgeofwhichtheskepticattacks.Forthefactthattheskepticcaninstallveryhighstandardsthatwedon'tliveuptohasnotendencytoshowthatwedon'tsatisfythemorerelaxedstandardsthatareinplaceinmoreordinaryconversationsanddebates.

35Again,I'mhereassumingaskeptic‐friendlyversionofcontextualism.Seethesecondimportantpointmadeattheendofsection2.

Page 42: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐40‐

13. THERULEOFSENSITIVITYANDSCA:ACOMPARISONOFOURNEW

SOLUTIONWITHTHEOTHERCONTEXTUALISTSOLUTIONSANDWITHNOZICK’SSOLUTION

Recallthattheproblemwiththeothercontextualistsolutionswe'veseenisthattheyfailtoadequatelyexplainwhyAl'sfirstpremisehastheintuitivepullithas(whentheskepticalhypothesisemployediswellchosen).OurnewcontextualistsolutiongainsanimportantadvantageoveritscontextualistrivalsbyincorporatingSCA.WeexplaintheplausibilityofAl'sfirstpremisebyreferencetothefollowingtwofacts.First,anybeliefwemighthavetotheeffectthataskepticalhypothesisdoesn'tobtain(wherethathypothesisiswellchosen)isinsensitive:aswerealize,wewouldholdthisbeliefevenifitwerefalse(evenifthehypothesisdidobtain).And,second,wehaveaverygeneralinclinationtothinkthatwedon'tknowthatPwhenwerealizethatourbeliefthatPisinsensitive—whenwerealizethatwewouldbelievethatPevenifPwerefalse.

WefollowNozickinemployingSCA.ButwedivergefromNozick'streatmentinouraccountofwhythesecondfactaboveholds.OnNozick'saccount,wehavethegeneralinclinationassertedtherebecauseourconceptofknowledgejustis,roughly,thatoftrue,sensitivebelief.Thiswouldaccountforourinclinationtodenythestatusofknowledgetoinsensitivebeliefsalright,butitwouldalsohaveushappilyassertingabominableconjunctions,which,infact,we'reloathetodo.

Ournewsolutionavoidsthisunhappinessbynotbuildingasensitivityrequirementintotheveryconceptofknowledge.Thenotionofsensitivity,rather,findsitshappierhomeinourcontextualistaccountofhowthestandardsforknowledgeareraised,andthesecondfactaboveisaccountedforasfollows.WhereS'sbeliefthatPisnotsensitive,SisnotinagoodenoughepistemicpositiontocountasknowingthatPbythestandardsthat,accordingtotheRuleofSensitivity,wouldbeputinplacebytheveryclaimthatSknows(ordoesn'tknow)thatP.Thus,anassertionthatSdoesn'tknowthatP,whereS'sbeliefthatPisinsensitive,willraisethestandardsforknowledgetoalevelhighenoughtomakethatdenialofknowledgetrue.ApositiveclaimthatSdoesknowsuchaP,ontheotherhand,is

Page 43: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐41‐

doomedtofailure:Themakingoftheclaimwillraisethestandardsforknowledgetoalevelhighenoughtomakethatclaimfalse.So,wheneverS'sbeliefthatPisinsensitive,wecantruthfullyassertthatSdoesn'tknowthatP,andcanonlyfalselysaythatSdoesknowthatP.Nowonder,then,thatthesecondfactholds!

Thus,wesuccessfullyincorporateSCA,explainingtheplausibilityofAl'sfirstpremise,withoutfollowingNozickinlicensingabominableconjunctions.

14. OURNEWCONTEXTUALISTSOLUTIONCLARIFIEDANDCOMPAREDWITHTHESTRAIGHTFORWARDSOLUTIONS

Thepuzzleofskepticalhypotheses,recall,concernsthepremisesofAltogetherwiththenegationofitsconclusion:

1. Idon'tknowthatnot‐H.

2. IfIdon'tthatnot‐H,thenIdon'tknowthatO.

not‐C.IdoknowthatO.

Asolutiontothepuzzlemust,ofcourse,issueaverdictastothetruthofeachofthesethree,butitmustalsoexplainwhywefindallofthemplausible.

Let'sbeclearaboutwhatourpresentcontextualistsolutionhastosayabouteachofthese.Ourverdictregarding(2)isthatit'strueregardlessofwhatepistemicstandardit'sevaluatedat,soitsplausibilityiseasilyaccountedfor.Butthis,combinedwithasimilarlyenthusiasticendorsementof(1),wouldlandusinboldskepticism.Weavoidthatfatebyendorsing(1)astrue,notatallstandards,butonlyattheunusuallyinflatedstandardsconducivetoskepticism.Thus,onoursolution,wedoknow,forinstance,thatwe'renotBIVs,accordingtoordinarylowstandardsforknowledge.But,though(1)isfalsewhenevaluatedaccordingtothoseordinarylowstandards,we'reabletoexplainitsplausibility,aswe'veseen,bymeansofthefactthatthehighstandardsatwhich(1)istruearepreciselythestandardsthatanassertionordenialofitputintoplay.Sinceattemptstoassert(1)areboundto

Page 44: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐42‐

resultintruth,andattemptstodenyitaredestinedtoproducefalsehood,36it'snosurprisethatwefinditsoplausible.

Butwhatof(not‐C)?Onthepresentsolution,claimstoknowordinarypropositionsaretrueaccordingtoordinarylowstandardsbutfalseaccordingtothehighlyinflatedstandardsthat,bytheRuleofSensitivity,areputinplacebytheassertionof(1).(Not‐C)seemsplausiblebecauseit'struewhenevaluatedatthestandardsmostnormallyappliedtoit.But,itwillbeasked,whydowefindtheseclaimstoknowplausibleevenwhenwe'reinacontextinwhichtheskeptichasraisedthestandardstosuchalevelthattheseclaimsarefalse?Alittlecautionisinorderhere.It'scontroversialjusthowintuitivelycorrect(not‐C)doesseemtousinsuchacontext.Mostofusfeelsomeambivalence.Suchambivalenceistobeexpectedwheneverwe'redealingwithapuzzleconsistingofmutuallyinconsistentpropositions,allofwhichareindividuallyplausible.Forwhenthepropositionsareconsideredtogether,onewillhavethisgoodreasonfordoubtingeachofthem:thattheothersseemtrue.Andit'sdifficulttodistinguishthedoubtof(not‐C)thatarisesfromthisverygeneralsource(thatitsfalsehoodfollowsfromotherthingsonefindsplausible)fromthatwhicharisesfromthefactthatthestandardsarehigh.Atanyrate,theverystrongpullthat(not‐C)continuestoexerton(atleastmostof)usevenwhenthestandardsarehighisexplainedinthemanneroutlinedinsection2:Evenwhilewe'reinacontextgovernedbyhighstandardsatwhichwedon'tcountasknowingthatO,weatthesametimerealizethatassoonaswefindourselvesinmoreordinaryconversationalcontexts,itwillnotonlybetrueforustoclaimtoknowtheseveryOsthattheskepticnowdeniesweknow,butitwillalsobewrong

36Butforcasesinwhichitseemsonecantruthfullysay‘Sknowsthatnot‐H,’despitethefactthatS'sbeliefthatnot‐Hisinsensitive,seechapter3,sectionJ(‘Low‐StrengthClaimstoKnowthatSkepticalHypothesesDoNotObtain’)ofmy199O.Insuchcases,givencertainfeaturesoftheconversationalsituation,theRuleofSensitivitydoesnotoperate.Theseconstituteexceptionstotherulethatonecannottruthfullycallaninsensitivebeliefknowledge.AsIexplainthere,Iwelcometheseexceptions,andwouldactuallybeabitworriedifthereweren'tsuchexceptions.Forit'safeatureofmytreatmentofAIthatwedoknowskepticalhypothesestobefalseaccordingtolowepistemicstandards.Iwouldfinditabitembarrassingifwecouldneverclaimtohavesuchknowledgebymeansofsimpleknowledgeattributions,andI'mreassuredbytheresultthatinspecialconversationalcircumstances,itseemswecantruthfullyclaimtoknowthatnot‐H,despitethefactthatourbeliefthatnot‐Hisinsensitive.

Page 45: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐43‐

forustodenythatweknowthesethings.It'seasy,then,tothinkthattheskeptic'spresentdenialmustbeequallyfalseandthatitwouldbeequallytrueforusnow,intheskeptic'spresence,toclaimtoknowthatO.

Theverdictsthepresentsolutionissuesregardingthetruthvaluesofthemembersofthetriadarecomplicatedbythefactthatoursisacontextualistsolution.Only(2)receivesthesameverdictregardlessofwhattheepistemicstandardsare;thetruthvaluesof(1)and(not‐C)varywithcontext.It'sjustthisvariancethatoursolutionsoessentiallyreliesoninexplaininghowwefallintoourpuzzlingconflictofintuitions.Noncontextualist(henceforth,‘straightforward’)solutions,ontheotherhand,mustchooseoneofthemembersofthistriadtodeny,claimingthislosertobefalseaccordingtotheinvariantepistemicstandardsthatgovernallattributionsanddenialsofknowledge:The‘Moorean’solutioninthiswaydenies(1),37the‘Nozickean’(2),andthe‘BoldSkeptial’solutionthusdenies(not‐C),acceptingthatwespeakfalselywhenever,eveninordinary,nonphilosophicaldiscussions,weclaimtoknowtheOinquestion.

Fromtheperspectiveofourpresentcontextualistsolution,eachofthesestraightforwardsolutionsresultsinpart,ofcourse,fromafailuretoseethetruthofcontextualism.ButwhichstraightforwardsolutionaninvariantistconfusedlyadoptswilldependonthestandardsthatdominateherevaluationofourbeliefsinOandinnot‐H.Ifherevaluationisdominatedbytherelativelylowstandardsthatgovernourordinary,out‐on‐the‐streettalkofknowledge,shewillendupaMoorean.Ifsheevaluatesthebeliefsinquestionaccordingtothehighstandardsthatareputintoplacebytheskeptic'spresentationofAl,boldskepticismistheresult.TheNozickeansolutionensuesfromevaluatingeachbeliefaccordingtothestandardsthatwouldmostoftenbeusedinevaluatingthatbelief.Forreasonswe'veseen,aclaimtoknow(oranadmissionthatonedoesn'tknow)thataskepticalhypothesisisfalsewill,bytheRuleofSensitivity,tendtoinviteaveryhighreading,atwhichtheadmissionistrueandtheclaimisfalse.ButaclaimtoknowthatOdoesn'tso

37Thisiscalledthe‘Moorean’solutionbecauseMoorerespondedinthiswaytothedreamargument.It'sfarfromcertainthatMoorewouldhavesorespondedtootherinstancesofAIthatutilizedifferentskepticalhypotheses.

Page 46: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐44‐

demandahighreading.Fromthepresentperspective,theNozickeanisreactingtothefactthatonecanusuallytruthfullyclaimthatonedoesknowthatOandcanusuallytruthfullyclaimnottoknowthatnot‐H.WhattheNozickeanmissesishowdifficultitistomakethesetwoclaimstogether:onceyouhaveadmittedthatyoudon'tknowthatnot‐H,itseemsthereverseofintuitivelycorrecttoclaimtoknowthatO,atleastuntiltheconversationalairiscleared.

Tosucceed,astraightforwardsolutionmustexplainwhatleadsourintuitionsastraywithrespecttotheunluckymemberofthetriadwhichthatsolutiondenies.Otherwise,we'llhavelittlereasonfordenyingjustthatmemberofthetriad.Nozickhimselfprovidesnosuchexplanationwithrespectto(2),parentheticallyleavingthisvitaltaskto‘furtherexploration,’38andotherNozickeans,ifanytherebe,havenot,tothebestofmyknowledge,progressedanyfartheralongthisfront.Mooreans,tothebestofmyknowledge,havefarednobetterinexplainingwhywe'resoreluctanttoclaimthestatusofknowledgeforourinsensitivebeliefs.It'sthedefendersofboldskepticismwho'vemadethemostprogresshere.Intheremainingsections,I'llexplainwhyourcontextualistsolutionissuperiortothatoftheboldskeptic.

15. BOLDSKEPTICISMANDTHEWARRANTEDASSERTABILITYMANEUVER

Almostallofthetime,itseemstoalmostallofusthatwedoknowtheOsthattheskepticclaimswedon'tknow.Accordingtotheboldskeptic,wheneverwesayorthinkthatweknowthesethings,wesayorthinksomethingfalse.Theboldskepticthusimplicatesus,speakersofEnglish,insystematicandwidespreadfalsehoodinouruse,inspeechandinthought,ofourverycommonword'know'.Equallyparadoxically,theboldskepticholdsthatwe'respeakingthetruthwheneverwesay

38Seethefirstparagraphofnote28,above.

Page 47: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐45‐

thatsomeonedoesn'tknowtheseOs,eventhoughitseemstomostofusthatwe'dthenbesayingsomethingquitefalse.Whatleadsusastray?PeterUngerandBarryStroudhavesuggestedonbehalfofboldskepticismthatalthoughwedon'tknowtheseO's,it'softenusefulforustoclaimthatwedoknowthem,andwearethereforeoftenwarrantedorjustifiedinmakingsuchclaims.Whatthenleadsusastrayisthis:Wemistakethisuseful/justified/warrantedassertabilityofknowledgeascriptionsfortruth.39Ontheothersideofthecoin,presumably,we'remistakingtheuseless/unwarranted/unjustifiedassertabilityofdenialsofknowledgeforfalsehood.

Twoseriousproblemsemergefortheboldskepticatthispoint.Thefirstisthatsuch‘warrantedassertabilitymaneuvers’couldbeattemptedbyadvocatesoftheothersolutionsaswell.Warrantedassertabilityindeedcanbemistakenfortruth,andunwarrantedassertabilityforfalsehood,butthisbyitselfdoesnotfavortheboldskeptic'ssolutionovertheotherstraightforwardapproaches.Eachofthestraightforwardapproachesdeniesamemberofthetriadconstitutingourpuzzle,andeachitseemscouldclaimthatthereasonthisloserthey'vechosenseemstrue,thoughit'sinfactfalse,isthatwe'reoftenwarrantedinassertingit,andwemistakethiswarrantedassertabilityfortruth.Thus,theMoorean,forinstance,couldclaimthatalthoughwedoindeedknowthatHisfalse,we'renotwarrantedinclaimingthatweknowthis(thoughthisclaimwouldbetrue),butareratherwarrantedinsayingthatwedon'tknow(thoughthislatterisfalse).Simplyattributingapparenttruthtowarrantedassertabilityisagamealmostanypartytothisdisputecanfairlyeasilyplay.40Thatthislineofthoughtwouldeventuallyworkoutanybetterfortheboldskepticthanforhisopponentswouldtakesomeshowing.41

39ThisisthebasiclineUngertakesinhisdefenseofboldskepticisminhis1975;seeespeciallypages50‐54.Stroud,thoughnothimselfadvocatingboldskepticism,doesseektodefendtheboldskepticalongtheselinesinchapter2ofhis1984;seeespeciallypages55‐82.

40Bycontrast,ournewcontextualistsolutionattributestheapparenttruthof(1)to(l)'struth(andnotjustitswarrantedassertability)attheverystandardsitsassertioninvokes.

41Formyownpart,forreasonsIcan'tgointohere,IthinktheresultingMooreanpositionwouldbeslightlymoredefensible;thus,ifIhadtorejectcontextualismandadoptastraightforwardsolution,I'dbeaMoorean.

Page 48: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐46‐

It'sat(1)thattheskeptichashisbesthopeofgaininganadvantageovermysolution,forthatpremiseindeeddoesseemtrue,and,unliketheskeptic,I'vestoppedshortoffullyendorsingit,makingdowithanexplanationofitsplausibility.Buttheskeptic'sotherproblemlurkshere.Usually,whilesolvingaphilosophicalpuzzleconsistingofasetofindividuallyplausiblebutmutuallyinconsistentclaims,oneonlyhastoexplain(away)theplausibilityofthosemembersofthesetonedenies,andoneisrelievedoftheburdenofexplainingtheplausibilityofthosemembersthatoneendorses,theirtruthandourabilitytorecognizethattruthbeingexplanationenoughoftheirapparenttruth.Buttruthdoesnotsufficetoexplainapparenttruthwhereonemakesusouttobeabsolutelyhorriblejudgesoftruthsofthekindinquestion.Thus,theskeptic'ssecondbigproblemisthat,becauseheholdsthatwe'resubjecttoconstantandradicalerrorastothescopeofourknowledge,consistentlythinkingweknowthingswhenwedon't,theskeptic,althoughhethinks(1)istrue,owesusanexplanationforitsplausibility.Giventhatourhabitofmistakingourignoranceforknowledgeissopervasive,whydoesn'titseemtousherethatweknowwhat,infact,wedon't—thattheseskepticalhypothesesarefalse?Whydoesourlackofknowledge,whichwe'resopervasivelyblindto,shinethroughsoclearlytousjustwheretheissueiswhetherweknowaskepticalhypothesistobefalse?

Theskeptic'sinitialanswerwillcertainlybethatwe'renotwarrantedinclaimingtoknowthatskepticalhypothesesdon'tobtain,andthuscan'tmistakewarrantedassertabilityfortruthhere.Butthen,toseewhyskepticalhypothesesareeffective,wemustbetoldwhywe'renotwarrantedinclaimingtoknowthatskepticalhypothesesarefalse,giventhat,accordingtotheskeptic,wearewarrantedinclaimingtoknowallmannerofotherthingsthatinfactwedon'tknow.Andhereskepticshavelittletooffer.Butiftheresultsofsections5and8abovearecorrect,theanswermustinvolvethelackofsensitivityenjoyedbyourbeliefsthatskepticalhypothesesdon'tobtain.Theskeptic'suseofSCAwilltakethisform:Althoughweknownothing(orverylittle),it'swhenourbeliefsareinsensitivethatwe'renotevenwarrantedinassertingthatweknowandwethereforerecognizeourlackofknowledge.ButtheskepticmustnowalsoaddressAl'ssecondpremise,makingsurehisendorsementofSCAismadeinsuchawayastoaccountforourintuitionshere.Indeed,whetherornothebuysintoSCA,theskepticfacesthis

Page 49: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐47‐

question:If,asheclaims,we'reusuallyunderthedelusionthatweknowthatO,butwecustomarilyrecognizethatwedon'tknowthatnot‐H,whyaren'twehappytoconjointhiserrorwiththatinsightandembracetheabominableconjunction?

Thismaylooklikeadifficultquestion,buttheskeptichasareadyanswer.Hisproblemisthatthewarrantedassertabilitymaneuverbyitselfdidn'treallysolveourpuzzle,butratherre‐introduceditinanewform.AndtheonlywayI'veseentoincorporateSCAintoatreatmentofAIthatalsohandlestheotherpiecesofourpuzzleistoemploytheideathatcontextuallysensitiveepistemicstandardsgovernouruseof'know',andtoposittheRuleofSensitivityasthemechanismbywhichtheAIskepticdrivesthosestandardsup,asI'veadvocatedhere.Butwiseinvariantiststypicallyacceptthatcontextuallyvaryingstandardsgovernouruseofascriptionsanddenialsofknowledge.Thesensibleinvariantistwilladmitthat,ofcourse,whatpassesforknowledgeinsomecontextswon'tsopassinothers.Beinganinvariantist,he'lldenythatthetruthconditionsofknowledgeattributionsvaryinthewaythecontextualistclaimstheydo.Butthecleverinvariantistwillmaintainthatthevaryingepistemicstandardsthatthecontextualistsupposesgovernthetruthconditionsofthesesentencesinfactgoverntheirconditionsofwarrantedassertability.42

Thisallowstheboldskeptictomimicanycontextualistsolution,andinparticularthesolutionI'madvocatinghere,bymeansofasimpletwist.Withrespecttomysolution,theboldskepticcanmaintainthattheRuleofSensitivityisarulefortheraisingoftheepistemicstandardsgoverningouruseofsentencesascribingknowledgetosubjects,alright,butinsistthatitgovernsthewarrantedassertabilityconditionsofthesesentences,ratherthantheirtruthconditions,which,he'll

42Stroudthusclaimsthatontheskeptic'sconceptionofourpractices,weoperateundercertain‘practicalconstraints’(1984,75)inoureverydayusesof'know',andassertsthatourstandardsforsayingweknowvaryfromcasetocase(65‐66).Thus,ontheskeptic'sconception,thestandardsforascribingknowledgethatweemployineverydayusedependuponour‘aimsandinterestsatthemoment’(65).Accordingtocontextualism,thesevaryingstandardsreflectacorrespondingvariationinthetruthconditionsforattributionsofknowledge.ButonStroud'sskeptic'sconception,whenweascribeknowledgeineverydaysituations,wearetypicallysayingsomethingliterallyfalse,although‘theexigenciesofaction’justifythesefalseattributions.ThebestexplorationofthistypeofideaisprovidedbyUngerinhis1984.

Page 50: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐48‐

maintain,remainconstantatalevelbeyondthereachofmeremortalstosatisfy.Thewarrantedassertabilitymaneuvercanthenbeemployed:Wemistakewarrantedassertabilityfortruth(andunwarrantedassertabilityforfalsehood).Thus,sincewe'reneverwarrantedinclaimingtoknowthatskepticalhypothesesdon'tobtain(duetotheoperationofthetwistedRuleofSensitivity),we'reledtojudge(correctly)thatsuchclaimstoknowledgewouldbefalse.AndsinceAl'ssecondpremiseisalwayswarranted,wejudge(againcorrectly)thatthispremiseistrue.ButsinceaclaimtoknowsomeOisusuallywarranted,duetothelowstandardsforwarrantedassertabilitythatwouldordinarilybeappliedtosuchaclaim,wejudge(incorrectly)thatweknowthisO.Thus,mysolution,likeothercontextualistsolutions,canbeeasilyadaptedtosuitthepurposesoftheboldskeptic.Theresultisatheoryparalleltomyowncontextualistsolution,whichdiffersinitssemanticsof'know':Accordingtothisparallelinvariantisttheory,thecontext‐sensitivevaryingepistemicstandardswe'vediscoveredgovernthewarrantedassertabilityconditionsofattributionsanddenialsofknowledge,ratherthantheirtruthconditions,whichareheldtobeinvariant.43Howshallwerationallydecidebetweenacontextualistsolution,andinparticulartheoneI'mheredefending,andtheboldskeptic'sanalogueofit?44

16. BOLDSKEPTICISMANDSYSTEMATICFALSEHOOD

Likeitscontextualistrelatives,ournewsolutionisdesignedlargelywiththegoalinmindofcreditingmostofourattributionsofknowledgewithtruth.Andnowonder.

43Goingbacktotheboldskeptic'sfirstproblem,notethatallthismaneuveringcanbemimickedbytheMoorean,whocanalsoholdthataRuleofSensitivitygovernsthewarrantedassertabilityconditionsofknowledgeascriptions.Liketheboldskeptic,theMooreancanholdthatthetruthconditionsofsuchattributionsofknowledgeremaininvariant,butintheMoorean'shands,theseconstantepistemicstandardswillbemeetablylow.

44ReadersofUnger's1984willseethestronginfluenceofthatexcellentbookonmyprocedurehere,thoughIcometoverydifferentconclusionsthanhedoesinthatwork.(Butseehismorerecent1986.)

Page 51: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐49‐

Weingeneraltakeitasastrikeagainstatheoryofacommontermofanaturallanguagethatitinvolvesthespeakersofthatlanguageinsystematicandwidespreadfalsehoodintheiruseofthatterm.Let'sborrowanexampleandsuppose,forinstance,thatacrazedphilosopherclaimedthattherearenophysicians,because,inadditiontoholdingamedicaldegree,anecessaryconditionforbeingaphysicianisthatonebeabletocureanyconceivableillness.45Onwhatgroundsshouldwerejectthisbizarreconjectureinfavorofamoretraditionalandlessdemandingaccountofwhatitistobeaphysician?OurlanguagecertainlycouldhavebeensuchthatS'shavingtheabilitytocureanyconceivableillnesswasatruthconditionof'Sisaphysician'(althoughtheword'physician'wouldnothavebeenveryusefulinthatcase).Invirtueofwhatisourlanguageinfactsuchthatthestrangetheoryisnottrueofit?I'mofcoursenotinapositiontogiveacompleteanswertothisquestion,butit'seminentlyreasonabletosupposethatsuchfactsasthese,regardingouruse,inthoughtandinspeech,oftheterm'physician'areinvolved:thatwetaketobephysiciansmanylicensedpractitionersofmedicinewhodon'tsatisfythedemandingrequirementalleged;thatweseriouslydescribethesepeopleasbeingphysicians;thatwedon'tdenythatthesepeoplearephysicians;etc.It'snodoubtlargelyinvirtueofsuchfactsasthesethatthetraditionalview,ratherthantheconjectureofourcrazedphilosopher,istrueofourlanguage.(Thecorrectnessofthetraditionalviewlargelyconsistsinsuchfacts.)Andthesefactsalsoprovideuswithourbestreasonsorevidenceforacceptingthetraditional,ratherthanthestrange,hypothesisregardingthesemanticsof'physician'.Inthiscase,thatthepeculiartheoryimplicatesusinsystematicandwidespreadfalsehoodinourspeechandthoughtinvolving'physicians'isa(constitutiveandevidential)strikeagainstthetheorythatprovesquitedecisive.

Ifourcrazedphilosophertriedtoaccountfortheabovefactsregardingouruseoftheterm'physician'viathequickandeasyconjecturethatthelessdemandingrequirementsthataremoretraditionallyassignedto'physician',whiletheydon'taccuratelyspecifythetruthconditionsofsentencesinvolvingthatterm,doarticulatethesesentences'warrantedassertabilityconditions,weshouldnot,onthe

45SeeStroud(1984,4O),whointurnborrowedtheexamplefromelsewhere.

Page 52: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐50‐

basisofthismaneuver,suspendourjudgmentagainsthiscontention.Thathistheoryinvolvesusinsystematicfalsehoodcontinuestoconstituteastrikeagainstit,andintheabsenceofquiteweightycounterbalancingconsiderationsthatfavorthestrangetheoryoverthetraditionalone,thisstrikeremainsdecisive.

Ofcourse,theproblemwiththishopelessnonstarterofatheoryisthattheredon'tseemtobeanysuchcounterbalancingconsiderationsinitsfavor.Bycontrast,boldskepticismcanappeartobesupportedbyskepticalargumentslikeAl.Thoughtheboldskeptic'sresolutionofourpuzzleinvolvesusinsystematicfalsehoodbecauseofitsunwaveringacceptanceofAl'sconclusion,itatthesametimecanseemtomakesenseofotherpiecesofthepuzzle(thatwe'reinclinedtosaythatwedon'tknowthatskepticalhypothesesarefalseandtosaythatwedon'tknowvariousordinarythingsifwedon'tknowthesehypothesestobefalse),makingthewarrantedassertabilitymaneuverseemmoremotivatedherethanitisinthehandsofourimaginedcrazedphilosopher.But,aswesawintheprevioussection,thisappearanceisdeceptive.Boldskepticism,byitself,doesnotexplaintheplausibilityofAl'spremises.Tohelptheskepticsolvethepuzzle,I'vehadtoascribetohimananalogueofournewsolution.46Butonceweseethattheskepticalpuzzlecanbesolvedjustaswellwithouttheboldskeptic'ssystematicfalsehood,we'releftwithnoreasonforpayingthathighpriceforasolution.47Indeed,sincetheboldskeptical

46Ofcourse,skepticsarefreetorefusethishelpandproposeothersolutions.Likepracticallyanyclaimtohaveprovidedthebestexplanationofsomething,myclaimhereishostagetothepossiblefuturedevelopmentofabetterexplanationcomingalong.

47Well,littlereason.Inhis1984,aspartofhiscaseforhisrelativistconclusionthatthere'snofactofthematterastowhethercontextualismorskepticalinvariantismiscorrect,Ungertriestobalancethisrelativedisadvantageofskepticalinvariantismagainstcontextualism'srelativedisadvantagethatitdoesnotmakethetruthconditionsofknowledgeattributionsappropriatelyindependentfromthecurrentintentsandinterestsofthosewhohappentobespeakingonagivenoccasion(37).Inpart3ofmy1992,Iarguethatcontextualismcanhandlethemostseriousconsequencesonemightsuspectwouldfollowfromthislackofindependence.Whateverindependenceconcernsmightremainwithcontextualismseemquiteswampedbythecostoftheboldskeptic'ssolution,which,asI'vehereargued,isquitehighindeed.

InhisreviewofUnger1984,Brueckner,relatingtheadvantagesofinvariantism,writes,‘Inparticular,speakers'intuitionsconcerningthecorrectuseof'know'seemtoconformtotheclosureprincipleforknowledgeassertedbytheinvariantistyetdeniedbythecontextualist’(1986a,512).Ifinvariantism,butnotcontextualism,upheldclosure,Iwouldtakethistobeaveryimportantadvantageforinvariantism—

Page 53: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐51‐

solutionandournewcontextualistsolutionunderconsiderationcloselyparalleleachother,there'snotmuchdifferenceinhowtheysolvethepuzzle.Thattheboldskepticalresolutioninvolvesusinsystematicfalsehoodisoneofthefewdifferencestobefoundhere,andit'saweightyconsiderationagainstthatresolution.And,withtherebeinglittleroomforweightycompensatingadvantagesforthisresolutionoverthecontextualist's(givenhowsimilartheyareinotherrespects),thisconsiderationprovesdecisive.So,aswiththecrazedphilosopher'stheoryof'physician',theboldskeptic'sresolutionofAIshouldberejectedbecauseitinvolvesusinsystematicandwidespreadfalsehoodinouruseofacommontermofourlanguage.

17. BEGGINGTHEQUESTIONAGAINSTTHESKEPTIC?

IfskepticsareallowedtoplayKingoftheMountain—theystartoffontop(nevermindhowtheygotthere)andit'stheanti‐skeptics'jobtoknockthemoff—displacingthemcanbeaverydifficulttask.Howdifficultdependsonseveralfactors,oneofwhichiswhatpremisestheanti‐skepticisallowedtoappealtoinanargumentdesignedtodethronetheskeptic.Iftheskepticwon'tallowanypremisestobeavailable,then,asThomasReidnoted,‘Itwouldbeimpossiblebyargumenttobeathimoutofthisstronghold;andhemustevenbelefttoenjoyhisscepticism’(1895,447).48If,tomakethegameabitmoreinteresting,aslimrangeofclaimsisallowedtopassinspectionandbeavailableforuseintheanti‐skepticalcampaign,then(asReidagainrecognized)it'softendifficulttosaywhat,ifanything,ofimportancewouldfollowfromthefactthattheskepticcanorcannotbeknockedfromhisperchbyargumentsfrompremisesofthatparticulartype.

perhapsevenweightyenoughtomakethecontestbetweenthetwotheoriesinteresting.But,asI'veargued,contextualismneednot,and,properlydeveloped,doesnot,takeanimplausiblestandontheissueofclosure.(Seesection1Oandespeciallynote33,above.)

48Idiscussthisinsection11.Bofmy1989.

Page 54: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐52‐

IhavelittleinterestinplayingKingoftheMountain.ButskepticalargumentslikeAIthreatentoshowthattheskepticneedn'tjustplaythisgame,butcangainthetopofthemountain—thatstartingfromourownbeliefsandintuitions,hecangiveusbetterreasonsforacceptinghisskepticismthanwehaveforrejectingit.I'veherearguedthattheboldskepticcannotwinthisbattle—thatofprovidingthebestresolutionofourpuzzlingconflictofintuitions.AlthoughAl'spremisesareinitiallyplausible,thebestresolutionfortheconflictofintuitionsgeneratedbyAIisnotthatoftheboldskeptic.

Alongtheway,I'vebeenassumingcertainthingsthatwebelievebutthattheskepticclaimswecan'tknow,therebyperhapsraisingtheconcernthatI'mbeggingthequestionagainsttheskeptic.Forinstance,inclaimingthatmybeliefthatIhavehandsissensitive,IbetraymyconvictionthatI'mnotaBIV,eitherintheactualworldorinanynearbyworlds.Indeed,I'mreadytoadmittotheskepticthatifIamaBIV,thenIdon'tknowIhavehands,accordingtoanystandardsforknowledge.But,ofcourse,asIfirmlybelieve,I'mnotaBIV.

Isitlegitimateformetousethisconvictioninadebateagainsttheskeptic?Notifwe'replayingKingoftheMountain.Butiftheskepticismarshallingdeeplyfeltintuitionsofoursinanattempttogiveusgoodreasonsforacceptinghisskepticism,it'slegitimatetopointoutthatotherofourbeliefsmilitateagainsthisposition,andaskwhyweshouldgivecredencetojustthosethatfavorhim.Andifwecanfurthershowthatthosebeliefsthatseemtofavorhissolutioncanbeaccommodatedinoursolutionbetterthanhecanaccommodatethoseofourbeliefsthatarehostiletohim,thebestconclusionwecandrawisthatwe'renotordinarilymistakenwhenweclaimorascribeknowledge,despitetheboldskeptic'sattempttoshowthatweare.Instead,themaininsightstobedrawnfromastudyofAlinvolvethecontext‐sensitivityofattributionsofknowledge,andtherolethattheRuleofSensitivityplaysinchangingtheepistemicstandardsthatgoverntheseattributions.49

49ThispaperisdedicatedtothememoryofKenKonyndyk,myfirstphilosophyteacher.ThankstoAnthonyBrueckner,JohnCarroll,GraemeForbes,RichardGrandy,MarkHeller,EdwardStein,HollyThomas,ananonymousreaderforthePhilosophicalReview,andtoaudiencesatRiceUniversityandattheUniversityofVirginia,Charlottesvilleforhelpfulcommentsonearlierdraftsofthispaper.SpecialthanksareduetoPeter

Page 55: SOLVING THE SKEPTICAL PROBLEM CONTENTS · 2017. 12. 22. · 2 Those who think that Hilary Putnam may have already disarmed BIV‐inspired skepticism should understand the BIV hypothesis

‐53‐

Ungerforhisimportantwritingsonthistopic,fromwhichI'velearnedmuch;forthreeyearsofalmostdailyphilosophicaldiscussions,manyofwhichwereonthetopicofknowledgeandskepticismandalmostallofwhichwereenjoyableandenlightening;andforhismanycommentsonvariousdraftsofthispaperanditsancestors.