Epistemology | Internet Encyclopedia of Philosophy

Epistemology is the study of knowledge. Epistemologists concern themselves with a number of tasks, which we might sort into two categories. Epistemology Epistemologyisthestudyofknowledge.Epistemologistsconcernthemselveswithanumberoftasks,whichwemightsortintotwocategories. First,wemustdeterminethenatureofknowledge;thatis,whatdoesitmeantosaythatsomeoneknows,orfailstoknow,something?Thisisamatterofunderstandingwhatknowledgeis,andhowtodistinguishbetweencasesinwhichsomeoneknowssomethingandcasesinwhichsomeonedoesnotknowsomething.Whilethereissomegeneralagreementaboutsomeaspectsofthisissue,weshallseethatthisquestionismuchmoredifficultthanonemightimagine. Second,wemustdeterminetheextentofhumanknowledge;thatis,howmuchdowe,orcanwe,know?Howcanweuseourreason,oursenses,thetestimonyofothers,andotherresourcestoacquireknowledge?Aretherelimitstowhatwecanknow?Forinstance,aresomethingsunknowable?Isitpossiblethatwedonotknownearlyasmuchaswethinkwedo?Shouldwehavealegitimateworryaboutskepticism,theviewthatwedonotorcannotknowanythingatall? Whilethisarticleprovidesonoverviewoftheimportantissues,itleavesthemostbasicquestionsunanswered;epistemologywillcontinuetobeanareaofphilosophicaldiscussionaslongasthesequestionsremain. TableofContents KindsofKnowledge TheNatureofPropositionalKnowledge Belief Truth Justification TheGettierProblem TheNo-False-BeliefCondition TheNo-DefeatersCondition CausalAccountsofKnowledge TheNatureofJustification Internalism Foundationalism Coherentism Externalism TheExtentofHumanKnowledge SourcesofKnowledge Skepticism CartesianSkepticism HumeanSkepticism Numericalvs.QualitativeIdentity Hume’sSkepticismaboutInduction Conclusion ReferencesandFurtherReading 1.KindsofKnowledge Theterm“epistemology”comesfromtheGreek“episteme,”meaning“knowledge,”and“logos,”meaning,roughly,“study,orscience,of.”“Logos”istherootofalltermsendingin“-ology”–suchaspsychology,anthropology–andof“logic,”andhasmanyotherrelatedmeanings. Theword“knowledge”anditscognatesareusedinavarietyofways.Onecommonuseoftheword“know”isasanexpressionofpsychologicalconviction.Forinstance,wemighthearsomeonesay,“Ijustknewitwouldn’train,butthenitdid.”Whilethismaybeanappropriateusage,philosopherstendtousetheword“know”inafactivesense,sothatonecannotknowsomethingthatisnotthecase.(Thispointisdiscussedatgreaterlengthinsection2bbelow.) Evenifwerestrictourselvestofactiveusages,therearestillmultiplesensesof“knowledge,”andsoweneedtodistinguishbetweenthem.Onekindofknowledgeisproceduralknowledge,sometimescalledcompetenceor“know-how;”forexample,onecanknowhowtorideabicycle,oronecanknowhowtodrivefromWashington,D.C.toNewYork.Anotherkindofknowledgeisacquaintanceknowledgeorfamiliarity;forinstance,onecanknowthedepartmentchairperson,oronecanknowPhiladelphia. Epistemologiststypicallydonotfocusonproceduraloracquaintanceknowledge,however,insteadpreferringtofocusonpropositionalknowledge.Apropositionissomethingwhichcanbeexpressedbyadeclarativesentence,andwhichpurportstodescribeafactorastateofaffairs,suchas“Dogsaremammals,”“2+2=7,”“Itiswrongtomurderinnocentpeopleforfun.”(Notethatapropositionmaybetrueorfalse;thatis,itneednotactuallyexpressafact.)Propositionalknowledge,then,canbecalledknowledge-that;statementsofpropositionalknowledge(orthelackthereof)areproperlyexpressedusing“that”-clauses,suchas“HeknowsthatHoustonisinTexas,”or“Shedoesnotknowthatthesquarerootof81is9.”Inwhatfollows,wewillbeconcernedonlywithpropositionalknowledge. Propositionalknowledge,obviously,encompassesknowledgeaboutawiderangeofmatters:scientificknowledge,geographicalknowledge,mathematicalknowledge,self-knowledge,andknowledgeaboutanyfieldofstudywhatever.Anytruthmight,inprinciple,beknowable,althoughtheremightbeunknowabletruths.Onegoalofepistemologyistodeterminethecriteriaforknowledgesothatwecanknowwhatcanorcannotbeknown,inotherwords,thestudyofepistemologyfundamentallyincludesthestudyofmeta-epistemology(whatwecanknowaboutknowledgeitself). Wecanalsodistinguishbetweendifferenttypesofpropositionalknowledge,basedonthesourceofthatknowledge.Non-empiricaloraprioriknowledgeispossibleindependentlyof,orpriorto,anyexperience,andrequiresonlytheuseofreason;examplesincludeknowledgeoflogicaltruthssuchasthelawofnon-contradiction,aswellasknowledgeofabstractclaims(suchasethicalclaimsorclaimsaboutvariousconceptualmatters).Empiricaloraposterioriknowledgeispossibleonlysubsequent,orposterior,tocertainsenseexperiences(inadditiontotheuseofreason);examplesincludeknowledgeofthecolororshapeofaphysicalobjectorknowledgeofgeographicallocations.(Somephilosophers,calledrationalists,believethatallknowledgeisultimatelygroundeduponreason;others,calledempiricists,believethatallknowledgeisultimatelygroundeduponexperience.)Athoroughepistemologyshould,ofcourse,addressallkindsofknowledge,althoughtheremightbedifferentstandardsforaprioriandaposterioriknowledge. Wecanalsodistinguishbetweenindividualknowledgeandcollectiveknowledge.Socialepistemologyisthesubfieldofepistemologythataddressesthewaythatgroups,institutions,orothercollectivebodiesmightcometoacquireknowledge. 2.TheNatureofPropositionalKnowledge Havingnarrowedourfocustopropositionalknowledge,wemustaskourselveswhat,exactly,constitutesknowledge.Whatdoesitmeanforsomeonetoknowsomething?Whatisthedifferencebetweensomeonewhoknowssomethingandsomeoneelsewhodoesnotknowit,orbetweensomethingoneknowsandsomethingonedoesnotknow?Sincethescopeofknowledgeissobroad,weneedageneralcharacterizationofknowledge,onewhichisapplicabletoanykindofpropositionwhatsoever.Epistemologistshaveusuallyundertakenthistaskbyseekingacorrectandcompleteanalysisoftheconceptofknowledge,inotherwordsasetofindividuallynecessaryandjointlysufficientconditionswhichdeterminewhethersomeoneknowssomething. a.Belief Letusbeginwiththeobservationthatknowledgeisamentalstate;thatis,knowledgeexistsinone’smind,andunthinkingthingscannotknowanything.Further,knowledgeisaspecifickindofmentalstate.While“that”-clausescanalsobeusedtodescribedesiresandintentions,thesecannotconstituteknowledge.Rather,knowledgeisakindofbelief.Ifonehasnobeliefsaboutaparticularmatter,onecannothaveknowledgeaboutit. Forinstance,supposethatIdesirethatIbegivenaraiseinsalary,andthatIintendtodowhateverIcantoearnone.SupposefurtherthatIamdoubtfulastowhetherIwillindeedbegivenaraise,duetotheintricaciesoftheuniversity’sbudgetandsuch.GiventhatIdonotbelievethatIwillbegivenaraise,IcannotbesaidtoknowthatIwill.OnlyifIaminclinedtobelievesomethingcanIcometoknowit.Similarly,thoughtsthatanindividualhasneverentertainedarenotamonghisbeliefs,andthuscannotbeincludedinhisbodyofknowledge. Somebeliefs,thosewhichtheindividualisactivelyentertaining,arecalledoccurrentbeliefs.Themajorityofanindividual’sbeliefsarenon-occurrent;thesearebeliefsthattheindividualhasinthebackgroundbutisnotentertainingataparticulartime.Correspondingly,mostofourknowledgeisnon-occurrent,orbackground,knowledge;onlyasmallamountofone’sknowledgeiseveractivelyonone’smind. b.Truth Knowledge,then,requiresbelief.Ofcourse,notallbeliefsconstituteknowledge.Beliefisnecessarybutnotsufficientforknowledge.Weareallsometimesmistakeninwhatwebelieve;inotherwords,whilesomeofourbeliefsaretrue,othersarefalse.Aswetrytoacquireknowledge,then,wearetryingtoincreaseourstockoftruebeliefs(whilesimultaneouslyminimizingourfalsebeliefs). Wemightsaythatthemosttypicalpurposeofbeliefsistodescribeorcapturethewaythingsactuallyare;thatis,whenoneformsabelief,oneisseekingamatchbetweenone’smindandtheworld.(Wesometimes,ofcourse,formbeliefsforotherreasons–tocreateapositiveattitude,todeceiveourselves,andsoforth–butwhenweseekknowledge,wearetryingtogetthingsright.)And,alas,wesometimesfailtoachievesuchamatch;someofourbeliefsdonotdescribethewaythingsactuallyare. Notethatweareassumingherethatthereissuchathingasobjectivetruth,sothatitispossibleforbeliefstomatchortofailtomatchwithreality.Thatis,inorderforsomeonetoknowsomething,theremustbesomethingoneknowsabout.Recallthatwearediscussingknowledgeinthefactivesense;iftherearenofactsofthematter,thenthere’snothingtoknow(ortofailtoknow).Thisassumptionisnotuniversallyaccepted–inparticular,itisnotsharedbysomeproponentsofrelativism–butitwillnotbedefendedhere.However,wecansaythattruthisaconditionofknowledge;thatis,ifabeliefisnottrue,itcannotconstituteknowledge.Accordingly,ifthereisnosuchthingastruth,thentherecanbenoknowledge.Evenifthereissuchathingastruth,ifthereisadomaininwhichtherearenotruths,thentherecanbenoknowledgewithinthatdomain.(Forexample,ifbeautyisintheeyeofthebeholder,thenabeliefthatsomethingisbeautifulcannotbetrueorfalse,andthuscannotconstituteknowledge.) c.Justification Knowledge,then,requiresfactualbelief.However,thisdoesnotsufficetocapturethenatureofknowledge.Justasknowledgerequiressuccessfullyachievingtheobjectiveoftruebelief,italsorequiressuccesswithregardtotheformationofthatbelief.Inotherwords,notalltruebeliefsconstituteknowledge;onlytruebeliefsarrivedatintherightwayconstituteknowledge. What,then,istherightwayofarrivingatbeliefs?Inadditiontotruth,whatotherpropertiesmustabeliefhaveinordertoconstituteknowledge?Wemightbeginbynotingthatsoundreasoningandsolidevidenceseemtobethewaytoacquireknowledge.Bycontrast,aluckyguesscannotconstituteknowledge.Similarly,misinformationandfaultyreasoningdonotseemlikearecipeforknowledge,eveniftheyhappentoleadtoatruebelief.Abeliefissaidtobejustifiedifitisobtainedintherightway.Whilejustificationseems,atfirstglance,tobeamatterofabelief’sbeingbasedonevidenceandreasoningratherthanonluckormisinformation,weshallseethatthereismuchdisagreementregardinghowtospelloutthedetails. Therequirementthatknowledgeinvolvejustificationdoesnotnecessarilymeanthatknowledgerequiresabsolutecertainty,however.Humansarefalliblebeings,andfallibilismistheviewthatitispossibletohaveknowledgeevenwhenone’struebeliefmighthaveturnedouttobefalse.Betweenbeliefswhichwerenecessarilytrueandthosewhicharetruesolelybyluckliesaspectrumofbeliefswithregardtowhichwehadsomedefeasiblereasontobelievethattheywouldbetrue.Forinstance,ifIheardtheweathermansaythatthereisa90%chanceofrain,andasaresultIformedthebeliefthatitwouldrain,thenmytruebeliefthatitwouldrainwasnottruepurelybyluck.Eventhoughtherewassomechancethatmybeliefmighthavebeenfalse,therewasasufficientbasisforthatbeliefforittoconstituteknowledge.Thisbasisisreferredtoasthejustificationforthatbelief.Wecanthensaythat,toconstituteknowledge,abeliefmustbebothtrueandjustified. Notethatbecauseofluck,abeliefcanbeunjustifiedyettrue;andbecauseofhumanfallibility,abeliefcanbejustifiedyetfalse.Inotherwords,truthandjustificationaretwoindependentconditionsofbeliefs.Thefactthatabeliefistruedoesnottelluswhetherornotitisjustified;thatdependsonhowthebeliefwasarrivedat.So,twopeoplemightholdthesametruebelief,butfordifferentreasons,sothatoneofthemisjustifiedandtheotherisunjustified.Similarly,thefactthatabeliefisjustifieddoesnottelluswhetherit’strueorfalse.Ofcourse,ajustifiedbeliefwillpresumablybemorelikelytobetruethantobefalse,andjustifiedbeliefswillpresumablybemorelikelyormoreprobabletobetruethanunjustifiedbeliefs.(Aswewillseeinsection3below,theexactnatureoftherelationshipbetweentruthandjustificationiscontentious.) d.TheGettierProblem Forsometime,thejustifiedtruebelief(JTB)accountwaswidelyagreedtocapturethenatureofknowledge.However,in1963,EdmundGettierpublishedashortbutwidelyinfluentialarticlewhichhasshapedmuchsubsequentworkinepistemology.Gettierprovidedtwoexamplesinwhichsomeonehadatrueandjustifiedbelief,butinwhichweseemtowanttodenythattheindividualhasknowledge,becauseluckstillseemstoplayaroleinhisbeliefhavingturnedouttobetrue. Consideranexample.Supposethattheclockoncampus(whichkeepsaccuratetimeandiswellmaintained)stoppedworkingat11:56pmlastnight,andhasyettoberepaired.Onmywaytomynoonclass,exactlytwelvehourslater,Iglanceattheclockandformthebeliefthatthetimeis11:56.Mybeliefistrue,ofcourse,sincethetimeisindeed11:56.Andmybeliefisjustified,asIhavenoreasontodoubtthattheclockisworking,andIcannotbeblamedforbasingbeliefsaboutthetimeonwhattheclocksays.Nonetheless,itseemsevidentthatIdonotknowthatthetimeis11:56.Afterall,ifIhadwalkedpasttheclockabitearlierorabitlater,Iwouldhaveendedupwithafalsebeliefratherthanatrueone. Thisexampleandotherslikeit,whileperhapssomewhatfar-fetched,seemtoshowthatitispossibleforjustifiedtruebelieftofailtoconstituteknowledge.Toputitanotherway,thejustificationconditionwasmeanttoensurethatknowledgewasbasedonsolidevidenceratherthanonluckormisinformation,butGettier-typeexamplesseemtoshowthatjustifiedtruebeliefcanstillinvolveluckandthusfallshortofknowledge.Thisproblemisreferredtoas“theGettierproblem.”Tosolvethisproblem,wemusteithershowthatallinstancesofjustifiedtruebeliefdoindeedconstituteknowledge,oralternativelyrefineouranalysisofknowledge. i.TheNo-False-BeliefCondition WemightthinkthatthereisasimpleandstraightforwardsolutiontotheGettierproblem.Notethatmyreasoningwastacitlybasedonmybeliefthattheclockisworkingproperly,andthatthisbeliefisfalse.Thisseemstoexplainwhathasgonewronginthisexample.Accordingly,wemightreviseouranalysisofknowledgebyinsistingthattoconstituteknowledge,abeliefmustbetrueandjustifiedandmustbeformedwithoutrelyingonanyfalsebeliefs.Inotherwords,wemightsay,justification,truth,andbeliefareallnecessaryforknowledge,buttheyarenotjointlysufficientforknowledge;thereisafourthcondition–namely,thatnofalsebeliefsbeessentiallyinvolvedinthereasoningthatledtothebelief–whichisalsonecessary. Unfortunately,thiswillnotsuffice;wecanmodifytheexamplesothatmybeliefisjustifiedandtrue,andisnotbasedonanyfalsebeliefs,butstillfallsshortofknowledge.Suppose,forinstance,thatIdonothaveanybeliefsabouttheclock’scurrentstate,butmerelythemoregeneralbeliefthattheclockusuallyisinworkingorder.Thisbelief,whichistrue,wouldsufficetojustifymybeliefthatthetimeisnow11:56;ofcourse,itstillseemsevidentthatIdonotknowthetime. ii.TheNo-DefeatersCondition However,theno-false-beliefconditiondoesnotseemtobecompletelymisguided;perhapswecanaddsomeotherconditiontojustificationandtruthtoyieldacorrectcharacterizationofknowledge.Notethat,evenifIdidn’tactivelyformthebeliefthattheclockiscurrentlyworkingproperly,itseemstobeimplicitinmyreasoning,andthefactthatitisfalseissurelyrelevanttotheproblem.Afterall,ifIwereasked,atthetimethatIlookedattheclock,whetheritisworkingproperly,Iwouldhavesaidthatitis.Conversely,ifIbelievedthattheclockwasn’tworkingproperly,Iwouldn’tbejustifiedinformingabeliefaboutthetimebasedonwhattheclocksays. Inotherwords,thepropositionthattheclockisworkingproperlyrightnowmeetsthefollowingconditions:itisafalseproposition,Idonotrealizethatitisafalseproposition,andifIhadrealizedthatitisafalseproposition,myjustificationformybeliefthatitis11:56wouldhavebeenundercutordefeated.Ifwecallpropositionssuchasthis“defeaters,”thenwecansaythattoconstituteknowledge,abeliefmustbetrueandjustified,andtheremustnotbeanydefeaterstothejustificationofthatbelief.Manyepistemologistsbelievethisanalysistobecorrect. iii.CausalAccountsofKnowledge RatherthanmodifyingtheJTBaccountofknowledgebyaddingafourthcondition,someepistemologistsseetheGettierproblemasreasontoseekasubstantiallydifferentalternative.Wehavenotedthatknowledgeshouldnotinvolveluck,andthatGettier-typeexamplesarethoseinwhichluckplayssomeroleintheformationofajustifiedtruebelief.Intypicalinstancesofknowledge,thefactorsresponsibleforthejustificationofabeliefarealsoresponsibleforitstruth.Forexample,whentheclockisworkingproperly,mybeliefisbothtrueandjustifiedbecauseit’sbasedontheclock,whichaccuratelydisplaysthetime.ButonefeaturethatallGettier-typeexampleshaveincommonisthelackofaclearconnectionbetweenthetruthandthejustificationofthebeliefinquestion.Forexample,mybeliefthatthetimeis11:56isjustifiedbecauseit’sbasedontheclock,butit’struebecauseIhappenedtowalkbyatjusttherightmoment.So,wemightinsistthattoconstituteknowledge,abeliefmustbebothtrueandjustified,anditstruthandjustificationmustbeconnectedsomehow. Thisnotionofaconnectionbetweenthetruthandthejustificationofabeliefturnsouttobedifficulttoformulateprecisely,butcausalaccountsofknowledgeseektocapturethespiritofthisproposalbymoresignificantlyalteringtheanalysisofknowledge.Suchaccountsmaintainthatinorderforsomeonetoknowaproposition,theremustbeacausalconnectionbetweenhisbeliefinthatpropositionandthefactthatthepropositionencapsulates.Thisretainsthetruthcondition,sinceapropositionmustbetrueinorderforittoencapsulateafact.However,itappearstobeincompatiblewithfallibilism,sinceitdoesnotallowforthepossibilitythatabeliefbejustifiedyetfalse.(Strictlyspeaking,causalaccountsofknowledgemakenoreferencetojustification,althoughwemightattempttoreformulatefallibilisminsomewhatmodifiedtermsinordertostatethisobservation.) Whilecausalaccountsofknowledgearenolongerthoughttobecorrect,theyhaveengenderedreliabilisttheoriesofknowledge,whichshallbediscussedinsection3bbelow. 3.TheNatureofJustification OnereasonthattheGettierproblemissoproblematicisthatneitherGettiernoranyonewhoprecededhimhasofferedasufficientlyclearandaccurateanalysisofjustification.Wehavesaidthatjustificationisamatterofabelief’shavingbeenformedintherightway,butwehaveyettosaywhatthatamountsto.Wemustnowconsiderthismattermoreclosely. Wehavenotedthatthegoalofourbelief-formingpracticesistoobtaintruthwhileavoidingerror,andthatjustificationisthefeatureofbeliefswhichareformedinsuchawayastobestpursuethisgoal.Ifwethink,then,ofthegoalofourbelief-formingpracticesasanattempttoestablishamatchbetweenone’smindandtheworld,andifwealsothinkoftheapplicationorwithholdingofthejustificationconditionasanevaluationofwhetherthismatchwasarrivedatintherightway,thenthereseemtobetwoobviousapproachestoconstruingjustification:namely,intermsofthebeliever’smind,orintermsoftheworld. a.Internalism Beliefisamentalstate,andbelief-formationisamentalprocess.Accordingly,onemightreason,whetherornotabeliefisjustified–whether,thatis,itisformedintherightway–canbedeterminedbyexaminingthethought-processesofthebelieverduringitsformation.Suchaview,whichmaintainsthatjustificationdependssolelyonfactorsinternaltothebeliever’smind,iscalledinternalism.(Theterm“internalism”hasdifferentmeaningsinothercontexts;here,itwillbeusedstrictlytorefertothistypeofviewaboutepistemicjustification.) Accordingtointernalism,theonlyfactorsthatarerelevanttothedeterminationofwhetherabeliefisjustifiedarethebeliever’sothermentalstates.Afterall,aninternalistwillargue,onlyanindividual’smentalstates–herbeliefsabouttheworld,hersensoryinputs(forexample,hersensedata)andherbeliefsabouttherelationsbetweenhervariousbeliefs–candeterminewhatnewbeliefsshewillform,soonlyanindividual’smentalstatescandeterminewhetheranyparticularbeliefisjustified.Inparticular,inordertobejustified,abeliefmustbeappropriatelybaseduponorsupportedbyothermentalstates. Thisraisesthequestionofwhatconstitutesthebasingorsupportrelationbetweenabeliefandone’sothermentalstates.Wemightwanttosaythat,inorderforbeliefAtobeappropriatelybasedonbeliefB(orbeliefsB1andB2,orB1,B2,and…Bn),thetruthofBmustsufficetoestablishthetruthofA,inotherwords,BmustentailA.(Weshallconsidertherelationshipbetweenbeliefsandsensoryinputsbelow.)However,ifwewanttoallowforourfallibility,wemustinsteadsaythatthetruthofBwouldgiveonegoodreasontobelievethatAisalsotrue(bymakingitlikelyorprobablethatAistrue).Anelaborationofwhatcountsasagoodreasonforbelief,accordingly,isanessentialpartofanyinternalistaccountofjustification. However,thereisanadditionalconditionthatwemustadd:beliefBmustitselfbejustified,sinceunjustifiedbeliefscannotconferjustificationonotherbeliefs.BecausebeliefBbemustalsobejustified,musttherebesomejustifiedbeliefCuponwhichBisbased?Ifso,Cmustitselfbejustified,anditmayderiveitsjustificationfromsomefurtherjustifiedbelief,D.Thischainofbeliefsderivingtheirjustificationfromotherbeliefsmaycontinueforever,leadingusinaninfiniteregress.Whiletheideaofaninfiniteregressmightseemtroubling,theprimarywaysofavoidingsucharegressmayhavetheirownproblemsaswell.Thisraisesthe“regressproblem,”whichbeginsfromobservingthatthereareonlyfourpossibilitiesastothestructureofone’sjustifiedbeliefs: Theseriesofjustifiedbeliefs,eachbasedupontheother,continuesinfinitely. Theseriesofjustifiedbeliefscirclesbacktoitsbeginning(AisbasedonB,BonC,ConD,andDonA). Theseriesofjustifiedbeliefsbeginswithanunjustifiedbelief. Theseriesofjustifiedbeliefsbeginswithabeliefwhichisjustified,butnotbyvirtueofbeingbasedonanotherjustifiedbelief. Thesealternativesseemtoexhaustthepossibilities.Thatis,ifonehasanyjustifiedbeliefs,oneofthesefourpossibilitiesmustdescribetherelationshipsbetweenthosebeliefs.Assuch,acompleteinternalistaccountofjustificationmustdecideamongthefour. i.Foundationalism Letus,then,considereachofthefourpossibilitiesmentionedabove.Alternative1seemsunacceptablebecausethehumanmindcancontainonlyfinitelymanybeliefs,andanythought-processthatleadstotheformationofanewbeliefmusthavesomestartingpoint.Alternative2seemsnobetter,sincecircularreasoningappearstobefallacious.Andalternative3hasalreadybeenruledout,sinceitrendersthesecondbeliefintheseries(and,thus,allsubsequentbeliefs)unjustified.Thatleavesalternative4,whichmust,byprocessofelimination,becorrect. Thislineofreasoning,whichistypicallyknownastheregressargument,leadstotheconclusionthattherearetwodifferentkindsofjustifiedbeliefs:thosewhichbeginaseriesofjustifiedbeliefs,andthosewhicharebasedonotherjustifiedbeliefs.Theformer,calledbasicbeliefs,areabletoconferjustificationonother,non-basicbeliefs,withoutthemselveshavingtheirjustificationconferreduponthembyotherbeliefs.Assuch,thereisanasymmetricalrelationshipbetweenbasicandnon-basicbeliefs.Suchaviewofthestructureofjustifiedbeliefisknownas“foundationalism.”Ingeneral,foundationalismentailsthatthereisanasymmetricalrelationshipbetweenanytwobeliefs:ifAisbasedonB,thenBcannotbebasedonA. Accordingly,itfollowsthatatleastsomebeliefs(namelybasicbeliefs)arejustifiedinsomewayotherthanbywayofarelationtootherbeliefs.Basicbeliefsmustbeself-justified,ormustderivetheirjustificationfromsomenon-doxasticsourcesuchassensoryinputs;theexactsourceofthejustificationofbasicbeliefsneedstobeexplainedbyanycompletefoundationalistaccountofjustification. ii.Coherentism Internalistsmightbedissatisfiedwithfoundationalism,sinceitallowsforthepossibilityofbeliefsthatarejustifiedwithoutbeingbaseduponotherbeliefs.Sinceitwasoursolutiontotheregressproblemthatledustofoundationalism,andsincenoneofthealternativesseempalatable,wemightlookforaflawintheproblemitself.Notethattheproblemisbasedonapivotalbuthithertounstatedassumption:namely,thatjustificationislinearinfashion.Thatis,thestatementoftheregressproblemassumesthatthebasingrelationparallelsalogicalargument,withonebeliefbeingbasedononeormoreotherbeliefsinanasymmetricalfashion. So,aninternalistwhofindsfoundationalismtobeproblematicmightdenythisassumption,maintaininginsteadthatjustificationistheresultofaholisticrelationshipamongbeliefs.Thatis,onemightmaintainthatbeliefsderivetheirjustificationbyinclusioninasetofbeliefswhichcoherewithoneanotherasawhole;aproponentofsuchaviewiscalledacoherentist. Acoherentist,then,seesjustificationasarelationofmutualsupportamongmanybeliefs,ratherthanaseriesofasymmetricalbeliefs.Abeliefderivesitsjustification,accordingtocoherentism,notbybeingbasedononeormoreotherbeliefs,butbyvirtueofitsmembershipinasetofbeliefsthatallfittogetherintherightway.(Thecoherentistneedstospecifywhatconstitutescoherence,ofcourse.Itmustbesomethingmorethanlogicalconsistency,sincetwounrelatedbeliefsmaybeconsistent.Rather,theremustbesomepositivesupportrelationship–forinstance,somesortofexplanatoryrelationship–betweenthemembersofacoherentsetinorderforthebeliefstobeindividuallyjustified.) Coherentismisvulnerabletothe“isolationobjection”.Itseemspossibleforasetofbeliefstobecoherent,butforallofthosebeliefstobeisolatedfromreality.Consider,forinstance,aworkoffiction.Allofthestatementsintheworkoffictionmightformacoherentset,butpresumablybelievingallandonlythestatementsinaworkoffictionwillnotrenderonejustified.Indeed,anyformofinternalismseemsvulnerabletothisobjection,andthusacompleteinternalistaccountofjustificationmustaddressit.Recallthatjustificationrequiresamatchbetweenone’smindandtheworld,andaninordinateemphasisontherelationsbetweenthebeliefsinone’smindseemstoignorethequestionofwhetherthosebeliefsmatchupwiththewaythingsactuallyare. b.Externalism Accordingly,onemightthinkthatfocusingsolelyonfactorsinternaltothebeliever’smindwillinevitablyleadtoamistakenaccountofjustification.Thealternative,then,isthatatleastsomefactorsexternaltothebeliever’sminddeterminewhetherornotsheisjustified.Aproponentofsuchaviewiscalledanexternalist. Accordingtoexternalism,theonlywaytoavoidtheisolationobjectionandensurethatknowledgedoesnotincludeluckistoconsidersomefactorsotherthantheindividual’sotherbeliefs.Whichfactors,then,shouldbeconsidered?Themostprominentversionofexternalism,calledreliabilism,suggeststhatweconsiderthesourceofabelief.Beliefscanbeformedasaresultofmanydifferentsources,suchassenseexperience,reason,testimony,memory.Moreprecisely,wemightspecifywhichsensewasused,whoprovidedthetestimony,whatsortofreasoningisused,orhowrecenttherelevantmemoryis.Foreverybelief,wecanindicatethecognitiveprocessthatledtoitsformation.Initssimplestandmoststraightforwardform,reliabilismmaintainsthatwhetherornotabeliefisjustifieddependsuponwhetherthatprocessisareliablesourceoftruebeliefs.Sinceweareseekingamatchbetweenourmindandtheworld,justifiedbeliefsarethosewhichresultfromprocesseswhichregularlyachievesuchamatch.So,forexample,usingvisiontodeterminethecolorofanobjectwhichiswell-litandrelativelynearisareliablebelief-formingprocessforapersonwithnormalvision,butnotforacolor-blindperson.Formingbeliefsonthebasisofthetestimonyofanexpertislikelytoyieldtruebeliefs,butformingbeliefsonthebasisofthetestimonyofcompulsiveliarsisnot.Ingeneral,ifabeliefistheresultofacognitiveprocesswhichreliably(mostofthetime–westillwanttoleaveroomforhumanfallibility)leadstotruebeliefs,thenthatbeliefisjustified. Theforegoingsuggestsoneimmediatechallengeforreliabilism.Theformationofabeliefisaone-timeevent,butthereliabilityoftheprocessdependsuponthelong-termperformanceofthatprocess.(Thiscanincludecounterfactualaswellasactualevents.Forinstance,acoinwhichisflippedonlyonceandlandsonheadsnonethelesshasa50%chanceoflandingontails,eventhoughitsactualperformancehasyieldedheads100%ofthetime.)Andthisrequiresthatwespecifywhichprocessisbeingused,sothatwecanevaluateitsperformanceinotherinstances.However,cognitiveprocessescanbedescribedinmoreorlessgeneralterms:forexample,thesamebelief-formingprocessmightbevariouslydescribedassenseexperience,vision,visionbyanormally-sightedperson,visionbyanormally-sightedpersonindaylight,visionbyanormally-sightedpersonindaylightwhilelookingatatree,visionbyanormally-sightedpersonindaylightwhilelookingatanelmtree,andsoforth.The“generalityproblem”notesthatsomeofthesedescriptionsmightspecifyareliableprocessbutothersmightspecifyanunreliableprocess,sothatwecannotknowwhetherabeliefisjustifiedorunjustifiedunlessweknowtheappropriatelevelofgeneralitytouseindescribingtheprocess. Evenifthegeneralityproblemcanbesolved,anotherproblemremainsforexternalism.KeithLehrerpresentsthisproblembywayofhisexampleofMr.Truetemp.Truetemphas,unbeknownsttohim,hadatempucomp–adevicewhichaccuratelyreadsthetemperatureandcausesaspontaneousbeliefaboutthattemperature–implantedinhisbrain.Asaresult,hehasmanytruebeliefsaboutthetemperature,buthedoesnotknowwhyhehasthemorwhattheirsourceis.Lehrerarguesthat,althoughTruetemp’sbelief-formingprocessisreliable,hisignoranceofthetempucomprendershistemperature-beliefsunjustified,andthusthatareliablecognitiveprocesscannotyieldjustificationunlessthebelieverisawareofthefactthattheprocessisreliable.Inotherwords,themerefactthattheprocessisreliabledoesnotsuffice,Lehrerconcludes,tojustifyanybeliefswhichareformedviathatprocess. 4.TheExtentofHumanKnowledge a.SourcesofKnowledge Giventheabovecharacterizationofknowledge,therearemanywaysthatonemightcometoknowsomething.Knowledgeofempiricalfactsaboutthephysicalworldwillnecessarilyinvolveperception,inotherwords,theuseofthesenses.Science,withitscollectionofdataandconductingofexperiments,istheparadigmofempiricalknowledge.However,muchofourmoremundaneknowledgecomesfromthesenses,aswelook,listen,smell,touch,andtastethevariousobjectsinourenvironments. Butallknowledgerequiressomeamountofreasoning.Datacollectedbyscientistsmustbeanalyzedbeforeknowledgeisyielded,andwedrawinferencesbasedonwhatoursensestellus.Andknowledgeofabstractornon-empiricalfactswillexclusivelyrelyuponreasoning.Inparticular,intuitionisoftenbelievedtobeasortofdirectaccesstoknowledgeoftheapriori. Onceknowledgeisobtained,itcanbesustainedandpassedontoothers.Memoryallowsustoknowsomethingthatweknewinthepast,even,perhaps,ifwenolongerremembertheoriginaljustification.Knowledgecanalsobetransmittedfromoneindividualtoanotherviatestimony;thatis,myjustificationforaparticularbeliefcouldamounttothefactthatsometrustedsourcehastoldmethatitistrue. b.Skepticism Inadditiontothenatureofknowledge,epistemologistsconcernthemselveswiththequestionoftheextentofhumanknowledge:howmuchdowe,orcanwe,know?Whateverturnsouttobethecorrectaccountofthenatureofknowledge,thereremainsthematterofwhetherweactuallyhaveanyknowledge.Ithasbeensuggestedthatwedonot,orcannot,knowanything,oratleastthatwedonotknowasmuchaswethinkwedo.Suchaviewiscalledskepticism. Wecandistinguishbetweenanumberofdifferentvarietiesofskepticism.First,onemightbeaskepticonlywithregardtocertaindomains,suchasmathematics,morality,ortheexternalworld(thisisthemostwell-knownvarietyofskepticism).Suchaskepticisalocalskeptic,ascontrastedwithaglobalskeptic,whomaintainsthatwecannotknowanythingatall.Also,sinceknowledgerequiresthatourbeliefsbebothtrueandjustified,askepticmightmaintainthatnoneofourbeliefsaretrueorthatnoneofthemarejustified(thelatterismuchmorecommonthantheformer). Whileitisquiteeasytochallengeanyclaimtoknowledgebygliblyasking,“Howdoyouknow?”,thisdoesnotsufficetoshowthatskepticismisanimportantposition.Likeanyphilosophicalstance,skepticismmustbesupportedbyanargument.Manyargumentshavebeenofferedindefenseofskepticism,andmanyresponsestothoseargumentshavebeenofferedinreturn.Here,weshallconsidertwoofthemostprominentargumentsinsupportofskepticismabouttheexternalworld. c.CartesianSkepticism InthefirstofhisMeditations,RenéDescartesoffersanargumentinsupportofskepticism,whichhethenattemptstorefuteinthelaterMeditations.Theargumentnotesthatsomeofourperceptionsareinaccurate.Oursensescantrickus;wesometimesmistakeadreamforawakingexperience,anditispossiblethatanevildemonissystematicallydeceivingus.(Themodernversionoftheevildemonscenarioisthatyouareabrain-in-a-vat,becausescientistshaveremovedyourbrainfromyourskull,connectedittoasophisticatedcomputer,andimmerseditinavatofpreservativefluid.Thecomputerproduceswhatseemtobegenuinesenseexperiences,andalsorespondstoyourbrain’soutputtomakeitseemthatyouareabletomoveaboutinyourenvironmentasyoudidwhenyourbrainwasstillinyourbody.Whilethisscenariomayseemfar-fetched,wemustadmitthatitisatleastpossible.) Asaresult,someofourbeliefswillbefalse.Inordertobejustifiedinbelievingwhatwedo,wemusthavesomewaytodistinguishbetweenthosebeliefswhicharetrue(or,atleast,arelikelytobetrue)andthosewhicharenot.Butjustastherearenosignsthatwillallowustodistinguishbetweenwakinganddreaming,therearenosignsthatwillallowustodistinguishbetweenbeliefsthatareaccurateandbeliefswhicharetheresultofthemachinationsofanevildemon.Thisindistinguishabilitybetweentrustworthyanduntrustworthybelief,theargumentgoes,rendersallofourbeliefsunjustified,andthuswecannotknowanything.Asatisfactoryresponsetothisargument,then,mustshoweitherthatweareindeedabletodistinguishbetweentrueandfalsebeliefs,orthatweneednotbeabletomakesuchadistinction. d.HumeanSkepticism Accordingtotheindistinguishabilityskeptic,mysensescantellmehowthingsappear,butnothowtheyactuallyare.Weneedtousereasontoconstructanargumentthatleadsusfrombeliefsabouthowthingsappearto(justified)beliefsabouthowtheyare.Butevenifweareabletotrustourperceptions,sothatweknowthattheyareaccurate,DavidHumearguesthatthespecterofskepticismremains.Notethatweonlyperceiveaverysmallpartoftheuniverseatanygivenmoment,althoughwethinkthatwehaveknowledgeoftheworldbeyondthatwhichwearecurrentlyperceiving.Itfollows,then,thatthesensesalonecannotaccountforthisknowledge,andthatreasonmustsupplementthesensesinsomewayinordertoaccountforanysuchknowledge.However,Humeargues,reasonisincapableofprovidingjustificationforanybeliefabouttheexternalworldbeyondthescopeofourcurrentsenseperceptions.LetusconsidertwosuchpossibleargumentsandHume’scritiqueofthem. i.Numericalvs.QualitativeIdentity Wetypicallybelievethattheexternalworldis,forthemostpart,stable.Forinstance,IbelievethatmycarisparkedwhereIleftitthismorning,eventhoughIamnotcurrentlylookingatit.IfIweretogopeekoutthewindowrightnowandseemycar,Imightformthebeliefthatmycarhasbeeninthesamespaceallday.Whatisthebasisforthisbelief?Ifaskedtomakemyreasoningexplicit,Imightproceedasfollows: Ihavehadtwosense-experiencesofmycar:onethismorningandonejustnow. Thetwosense-experienceswere(moreorless)identical. Therefore,itislikelythattheobjectsthatcausedthemareidentical. Therefore,asingleobject–mycar–hasbeeninthatparkingspaceallday. Similarreasoningwouldundergirdallofourbeliefsaboutthepersistenceoftheexternalworldandalloftheobjectsweperceive.Butarethesebeliefsjustified?Humethinksnot,sincetheaboveargument(andallargumentslikeit)containsanequivocation.Inparticular,thefirstoccurrenceof“identical”referstoqualitativeidentity.Thetwosense-experiencesarenotoneandthesame,butaredistinct;whenwesaythattheyareidenticalwemeanthatoneissimilartotheotherinallofitsqualitiesorproperties.Butthesecondoccurrenceof“identical”referstonumericalidentity.Whenwesaythattheobjectsthatcausedthetwosense-experiencesareidentical,wemeanthatthereisoneobject,ratherthantwo,thatisresponsibleforbothofthem.Thisequivocation,Humeargues,renderstheargumentfallacious;accordingly,weneedanotherargumenttosupportourbeliefthatobjectspersistevenwhenwearenotobservingthem. ii.Hume’sSkepticismaboutInduction Supposethatasatisfactoryargumentcouldbefoundinsupportofourbeliefsinthepersistenceofphysicalobjects.Thiswouldprovideuswithknowledgethattheobjectsthatwehaveobservedhavepersistedevenwhenwewerenotobservingthem.Butinadditiontobelievingthattheseobjectshavepersistedupuntilnow,webelievethattheywillpersistinthefuture;wealsobelievethatobjectswehaveneverobservedsimilarlyhavepersistedandwillpersist.Inotherwords,weexpectthefuturetoberoughlylikethepast,andthepartsoftheuniversethatwehavenotobservedtoberoughlylikethepartsthatwehaveobserved.Forexample,Ibelievethatmycarwillpersistintothefuture.Whatisthebasisforthisbelief?Ifaskedtomakemyreasoningexplicit,Imightproceedasfollows: Mycarhasalwayspersistedinthepast. Natureisroughlyuniformacrosstimeandspace(andthusthefuturewillberoughlylikethepast). Therefore,mycarwillpersistinthefuture. Similarreasoningwouldundergirdallofourbeliefsaboutthefutureandabouttheunobserved.Aresuchbeliefsjustified?Again,Humethinksnot,sincetheaboveargument,andallargumentslikeit,containanunsupportedpremise,namelythesecondpremise,whichmightbecalledthePrincipleoftheUniformityofNature(PUN).Whyshouldwebelievethisprincipletobetrue?Humeinsiststhatweprovidesomereasoninsupportofthisbelief.Becausetheaboveargumentisaninductiveratherthanadeductiveargument,theproblemofshowingthatitisagoodargumentistypicallyreferredtoasthe“problemofinduction.”Wemightthinkthatthereisasimpleandstraightforwardsolutiontotheproblemofinduction,andthatwecanindeedprovidesupportforourbeliefthatPUNistrue.Suchanargumentwouldproceedasfollows: PUNhasalwaysbeentrueinthepast. Natureisroughlyuniformacrosstimeandspace(andthusthefuturewillberoughlylikethepast). Therefore,PUNwillbetrueinthefuture. Thisargument,however,iscircular;itssecondpremiseisPUNitself!Accordingly,weneedanotherargumenttosupportourbeliefthatPUNistrue,andthustojustifyourinductiveargumentsaboutthefutureandtheunobserved. 5.Conclusion Thestudyofknowledgeisoneofthemostfundamentalaspectsofphilosophicalinquiry.Anyclaimtoknowledgemustbeevaluatedtodeterminewhetherornotitindeedconstitutesknowledge.Suchanevaluationessentiallyrequiresanunderstandingofwhatknowledgeisandhowmuchknowledgeispossible.Whilethisarticleprovidesonoverviewoftheimportantissues,itleavesthemostbasicquestionsunanswered;epistemologywillcontinuetobeanareaofphilosophicaldiscussionaslongasthesequestionsremain. 6.ReferencesandFurtherReading Alston,WilliamP.,1989.EpistemicJustification:EssaysintheTheoryofKnowledge.Ithaca,NY:CornellUniversityPress. Armstrong,David,1973.Belief,Truth,andKnowledge.Cambridge:CambridgeUniversityPress. Adefenseofreliabilism. BonJour,Laurence,1985.TheStructureofEmpiricalKnowledge.Cambridge,MA:HarvardUniversityPress. Adefenseofcoherentism. Chisholm,Roderick,1966.TheoryofKnowledge,EnglewoodCliffs,NJ:Prentice-Hall. Chisholm,Roderick,1977.TheoryofKnowledge,2ndedition.EnglewoodCliffs,NJ:Prentice-Hall. Chisholm,Roderick,1989.TheoryofKnowledge,3rdedition.EnglewoodCliffs,NJ:Prentice-Hall. Chisholmwasoneofthefirstauthorstoprovideasystematicanalysisofknowledge.Hisaccountofjustificationisfoundationalist. Descartes,Rene,1641.MeditationsonFirstPhilosophy.ReprintedinThePhilosophicalWritingsofDescartes(3volumes).Cottingham,StoothoffandMurdoch,trans.Cambridge:CambridgeUniversityPress. Descartespresentsaninfallibilistversionoffoundationalism,andattemptstorefuteskepticism. Dancy,JonathanandErnestSosa(eds.),1993.ACompaniontoEpistemology.Oxford:Blackwell. DeRose,Keith,1995.“SolvingtheSkepticalProblem”PhilosophicalReview,104,pp.1-52. DeRoseKeithandTedWarfield(eds.),1999.Skepticism:AContemporaryReader.Oxford:OxfordUniversityPress. Feldman,RichardandEarlConee,1985.“Evidentialism.”PhilosophicalStudies,48,pp.15-34. Theauthorspresentanddefendan(internalist)accountofjustificationaccordingtowhichabeliefisjustifiedorunjustifiedinvirtueofthebeliever’sevidence. Gettier,Edmund,1963.“IsJustifiedTrueBeliefKnowledge?”Analysis,23,pp.121-123. InwhichtheGettierproblemisintroduced. Goldman,Alvin,1976.“ACausalTheoryofKnowing.”JournalofPhilosophy,64,pp.357-372. Goldman,Alvin,1986.EpistemologyandCognition.Cambridge:HarvardUniversityPress. Perhapsthemostimportantdefenseofreliabilism. Haack,Susan,1991.“AFoundherentistTheoryofEmpiricalJustification,”InTheoryofKnowledge:ClassicalandContemporarySources(3rded.),Pojman,Louis(ed.),Belmont,CA:Wadsworth. Anattempttocombinecoherentismandfoundationalismintoaninternalistaccountofjustificationwhichissuperiortoeitherofthetwo. Hume,David,1739.ATreatiseonHumanNature.Oxford:OxfordUniversityPress. Hume,David,1751.AnEnquiryConcerningHumanUnderstanding.Indianapolis:Hackett. Lehrer,Keith,2000.TheoryofKnowledge(2nded.).Boulder,CO:Westview. Adefenseofcoherentism.ThisisalsowherewefindtheTruetempexample. Lehrer,KeithandStewartCohen,1983.“Justification,Truth,andCoherence.”Synthese,55,pp.191-207. Lewis,David,1996.“ElusiveKnowledge”AustralasianJournalofPhilosophy,74,pp.549-567. Locke,John,1689.AnEssayConcerningHumanUnderstanding.Oxford:Clarendon. Plato,MenoandTheaetetus.InCompleteWorks.J.Cooper,ed.Indianapolis:Hackett. PlatopresentsanddefendsaversionoftheJTBanalysisofknowledge. Pollock,JohnandJosephCruz,1999.ContemporaryTheoriesofKnowledge(2nded.).Lanham,MD:RowmanandLittlefield. Adefenseofnon-doxasticfoundationalism,inwhichthebasicstatesareperceptsratherthanbeliefs. Russell,Bertrand,1912.ProblemsofPhilosophy. RussellpresentsaGettier-typeexample,whichwaslargelyoverlookedformanyyears. AuthorInformation DavidA.Truncellito Email:[email protected] U.S.A. Anencyclopediaofphilosophyarticleswrittenbyprofessionalphilosophers. StayConnected &nbsp&nbsp &nbsp&nbsp &nbsp&nbsp BrowsebyTopicBrowsebyTopic SelectCategory HistoryofPhilosophy    17thCenturyEuropean    18thCenturyEuropean    19thCenturyEuropean    AncientPhilosophy    HistoryMisc.    HistoryofAnalytic    MedievalPhilosophy    Philosophers    RenaissancePhilosophy Metaphysics&Epistemology    Epistemology    Metaphysics    Mind&CognitiveScience    PhilosophyofLanguage    PhilosophyofReligion    Uncategorized PhilosophicalTraditions    AmericanPhilosophy    ChinesePhilosophy    ContinentalPhilosophy    FeministPhilosophy    IndianPhilosophy    IslamicPhilosophy    TraditionMisc. Science,Logic,&Mathematics    Logic    PhilosophyofMathematics    PhilosophyofScience ValueTheory    Aesthetics    Bioethics    Ethics    PhilosophyofLaw    PoliticalPhilosophy    ValueMisc.